diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,91221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.991242743284463,
+  "eval_steps": 500,
+  "global_step": 152000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 7.3482,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.5339,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 1.7782,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.6851,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.6615,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 1.8091,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.6687,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.7193,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 1.6301,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.717,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.77,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.7114,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.6862,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.6833,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.7883,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.7293,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.7219,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 1.6383,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 1.7184,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.7022,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 1.8617,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.7488,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.7833,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 1.7259,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 1.8184,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.6941,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.6485,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.7037,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 1.7738,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 1.7621,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 1.7382,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.7619,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 1.7008,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.7947,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 1.816,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 1.8832,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 1.9344,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 1.8483,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.7117,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.771,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 1.8244,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 1.8452,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-05,
+      "loss": 1.7776,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.9036,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 1.7655,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.9702,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-05,
+      "loss": 1.863,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 1.9294,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9e-05,
+      "loss": 1.7937,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 2.0343,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996709335614864e-05,
+      "loss": 1.8105,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993418671229725e-05,
+      "loss": 1.8692,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990128006844586e-05,
+      "loss": 1.9421,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998683734245945e-05,
+      "loss": 2.1437,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998354667807431e-05,
+      "loss": 1.8478,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998025601368916e-05,
+      "loss": 2.0921,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997696534930403e-05,
+      "loss": 1.8999,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973674684918886e-05,
+      "loss": 2.0129,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997038402053375e-05,
+      "loss": 1.8734,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996709335614861e-05,
+      "loss": 1.8718,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996380269176347e-05,
+      "loss": 1.8927,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996051202737833e-05,
+      "loss": 1.9535,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995722136299319e-05,
+      "loss": 1.9859,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995393069860805e-05,
+      "loss": 2.0209,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9950640034222915e-05,
+      "loss": 1.8816,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994734936983777e-05,
+      "loss": 1.9963,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994405870545264e-05,
+      "loss": 1.8657,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994076804106749e-05,
+      "loss": 1.8409,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993747737668236e-05,
+      "loss": 1.9182,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9934186712297214e-05,
+      "loss": 1.8224,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9930896047912075e-05,
+      "loss": 1.8197,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992760538352694e-05,
+      "loss": 2.0405,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99243147191418e-05,
+      "loss": 2.3088,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992102405475666e-05,
+      "loss": 1.992,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991773339037152e-05,
+      "loss": 2.0311,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9914442725986375e-05,
+      "loss": 1.9199,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991115206160124e-05,
+      "loss": 1.9382,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99078613972161e-05,
+      "loss": 1.9352,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9904570732830966e-05,
+      "loss": 2.0591,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990128006844582e-05,
+      "loss": 1.9647,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989798940406068e-05,
+      "loss": 2.0227,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989469873967554e-05,
+      "loss": 2.0953,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9891408075290404e-05,
+      "loss": 1.9944,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9888117410905265e-05,
+      "loss": 2.0137,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9884826746520126e-05,
+      "loss": 1.9343,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988153608213498e-05,
+      "loss": 1.8778,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987824541774985e-05,
+      "loss": 1.8393,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9874954753364703e-05,
+      "loss": 1.8069,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987166408897957e-05,
+      "loss": 1.9059,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9868373424594426e-05,
+      "loss": 2.0168,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986508276020929e-05,
+      "loss": 1.8167,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986179209582415e-05,
+      "loss": 2.0043,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985850143143901e-05,
+      "loss": 1.9917,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985521076705387e-05,
+      "loss": 2.0448,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985192010266873e-05,
+      "loss": 1.9476,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984862943828359e-05,
+      "loss": 2.0946,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9845338773898455e-05,
+      "loss": 2.0684,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984204810951331e-05,
+      "loss": 2.008,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983875744512818e-05,
+      "loss": 1.8645,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983546678074303e-05,
+      "loss": 2.0166,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983217611635789e-05,
+      "loss": 2.0869,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9828885451972754e-05,
+      "loss": 2.1347,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9825594787587616e-05,
+      "loss": 2.387,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982230412320248e-05,
+      "loss": 1.881,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.981901345881734e-05,
+      "loss": 1.946,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.981572279443219e-05,
+      "loss": 1.9341,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.981243213004706e-05,
+      "loss": 1.9135,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9809141465661915e-05,
+      "loss": 1.9465,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980585080127678e-05,
+      "loss": 1.9581,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980256013689164e-05,
+      "loss": 1.8768,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97992694725065e-05,
+      "loss": 1.8418,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.979597880812136e-05,
+      "loss": 1.9568,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.979268814373622e-05,
+      "loss": 1.8766,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.978939747935108e-05,
+      "loss": 1.8908,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9786106814965944e-05,
+      "loss": 1.8882,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9782816150580805e-05,
+      "loss": 1.9378,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977952548619567e-05,
+      "loss": 1.9714,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977623482181053e-05,
+      "loss": 1.8524,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977294415742539e-05,
+      "loss": 2.0527,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9769653493040244e-05,
+      "loss": 1.9816,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9766362828655105e-05,
+      "loss": 1.9366,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9763072164269966e-05,
+      "loss": 2.1826,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975978149988483e-05,
+      "loss": 1.8222,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975649083549969e-05,
+      "loss": 1.8031,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975320017111455e-05,
+      "loss": 1.9835,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974990950672941e-05,
+      "loss": 1.8512,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974661884234427e-05,
+      "loss": 2.0046,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9743328177959134e-05,
+      "loss": 1.9519,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9740037513573995e-05,
+      "loss": 2.0673,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9736746849188856e-05,
+      "loss": 1.8805,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.973345618480371e-05,
+      "loss": 1.9736,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.973016552041858e-05,
+      "loss": 1.8435,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.972687485603343e-05,
+      "loss": 1.8263,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.97235841916483e-05,
+      "loss": 2.0054,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9720293527263156e-05,
+      "loss": 2.0062,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.971700286287802e-05,
+      "loss": 1.8132,
+      "step": 1360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.971371219849288e-05,
+      "loss": 2.093,
+      "step": 1370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.971042153410774e-05,
+      "loss": 2.0691,
+      "step": 1380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.97071308697226e-05,
+      "loss": 1.8904,
+      "step": 1390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.970384020533746e-05,
+      "loss": 1.9591,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.970054954095232e-05,
+      "loss": 2.1616,
+      "step": 1410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9697258876567185e-05,
+      "loss": 1.9208,
+      "step": 1420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.969396821218204e-05,
+      "loss": 2.0019,
+      "step": 1430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.969067754779691e-05,
+      "loss": 1.8626,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.968738688341176e-05,
+      "loss": 1.8762,
+      "step": 1450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.968409621902662e-05,
+      "loss": 1.817,
+      "step": 1460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9680805554641484e-05,
+      "loss": 2.0039,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9677514890256346e-05,
+      "loss": 2.0954,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.967422422587121e-05,
+      "loss": 1.9142,
+      "step": 1490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.967093356148607e-05,
+      "loss": 1.8284,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.966764289710092e-05,
+      "loss": 1.9075,
+      "step": 1510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.966435223271579e-05,
+      "loss": 2.0033,
+      "step": 1520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9661061568330645e-05,
+      "loss": 1.9238,
+      "step": 1530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965777090394551e-05,
+      "loss": 1.7958,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965448023956037e-05,
+      "loss": 1.851,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965118957517523e-05,
+      "loss": 1.9057,
+      "step": 1560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.964789891079009e-05,
+      "loss": 1.8864,
+      "step": 1570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.964460824640495e-05,
+      "loss": 1.8419,
+      "step": 1580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.964131758201981e-05,
+      "loss": 2.0535,
+      "step": 1590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9638026917634674e-05,
+      "loss": 1.9968,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.963473625324953e-05,
+      "loss": 1.96,
+      "step": 1610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9631445588864397e-05,
+      "loss": 1.7994,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.962815492447925e-05,
+      "loss": 1.9103,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.962486426009412e-05,
+      "loss": 2.0749,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9621573595708974e-05,
+      "loss": 1.9974,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9618282931323835e-05,
+      "loss": 1.9591,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9614992266938696e-05,
+      "loss": 1.8008,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961170160255356e-05,
+      "loss": 2.027,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960841093816842e-05,
+      "loss": 1.9188,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960512027378328e-05,
+      "loss": 1.7814,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9601829609398134e-05,
+      "loss": 1.9746,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9598538945013e-05,
+      "loss": 2.3507,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959524828062786e-05,
+      "loss": 1.9207,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9591957616242725e-05,
+      "loss": 1.9319,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958866695185758e-05,
+      "loss": 1.8252,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958537628747244e-05,
+      "loss": 1.9385,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95820856230873e-05,
+      "loss": 1.9013,
+      "step": 1770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.957879495870216e-05,
+      "loss": 2.0051,
+      "step": 1780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9575504294317025e-05,
+      "loss": 1.9945,
+      "step": 1790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9572213629931886e-05,
+      "loss": 2.0283,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.956892296554674e-05,
+      "loss": 2.0614,
+      "step": 1810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.956563230116161e-05,
+      "loss": 1.899,
+      "step": 1820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.956234163677646e-05,
+      "loss": 1.7771,
+      "step": 1830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.955905097239133e-05,
+      "loss": 1.9604,
+      "step": 1840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9555760308006185e-05,
+      "loss": 1.9363,
+      "step": 1850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.955246964362105e-05,
+      "loss": 1.9437,
+      "step": 1860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.954917897923591e-05,
+      "loss": 1.9351,
+      "step": 1870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.954588831485077e-05,
+      "loss": 1.8908,
+      "step": 1880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.954259765046563e-05,
+      "loss": 1.9821,
+      "step": 1890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.953930698608049e-05,
+      "loss": 1.9842,
+      "step": 1900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.953601632169535e-05,
+      "loss": 1.8822,
+      "step": 1910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9532725657310214e-05,
+      "loss": 1.8444,
+      "step": 1920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9529434992925076e-05,
+      "loss": 1.9795,
+      "step": 1930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.952614432853994e-05,
+      "loss": 1.8084,
+      "step": 1940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.952285366415479e-05,
+      "loss": 1.7289,
+      "step": 1950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.951956299976966e-05,
+      "loss": 1.9072,
+      "step": 1960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9516272335384514e-05,
+      "loss": 1.9638,
+      "step": 1970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9512981670999375e-05,
+      "loss": 1.9122,
+      "step": 1980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9509691006614236e-05,
+      "loss": 2.1751,
+      "step": 1990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.95064003422291e-05,
+      "loss": 1.9946,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.950310967784396e-05,
+      "loss": 1.9864,
+      "step": 2010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949981901345882e-05,
+      "loss": 1.9887,
+      "step": 2020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949652834907368e-05,
+      "loss": 1.9318,
+      "step": 2030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949323768468854e-05,
+      "loss": 1.8801,
+      "step": 2040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9489947020303404e-05,
+      "loss": 1.8793,
+      "step": 2050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9486656355918265e-05,
+      "loss": 1.7896,
+      "step": 2060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9483365691533127e-05,
+      "loss": 1.9264,
+      "step": 2070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.948007502714798e-05,
+      "loss": 1.7282,
+      "step": 2080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.947678436276285e-05,
+      "loss": 1.8534,
+      "step": 2090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9473493698377704e-05,
+      "loss": 2.018,
+      "step": 2100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9470203033992565e-05,
+      "loss": 1.9512,
+      "step": 2110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9466912369607426e-05,
+      "loss": 1.8822,
+      "step": 2120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.946362170522229e-05,
+      "loss": 1.9326,
+      "step": 2130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.946033104083715e-05,
+      "loss": 1.7682,
+      "step": 2140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.945704037645201e-05,
+      "loss": 1.8136,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.945374971206687e-05,
+      "loss": 2.2371,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.945045904768173e-05,
+      "loss": 1.7795,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944716838329659e-05,
+      "loss": 1.9501,
+      "step": 2180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9443877718911455e-05,
+      "loss": 1.8275,
+      "step": 2190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944058705452631e-05,
+      "loss": 1.8818,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943729639014118e-05,
+      "loss": 1.9928,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943400572575603e-05,
+      "loss": 1.8979,
+      "step": 2220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943071506137089e-05,
+      "loss": 1.861,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9427424396985755e-05,
+      "loss": 2.011,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9424133732600616e-05,
+      "loss": 1.9168,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942084306821548e-05,
+      "loss": 2.1098,
+      "step": 2260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941755240383034e-05,
+      "loss": 1.8105,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941426173944519e-05,
+      "loss": 1.9218,
+      "step": 2280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.941097107506006e-05,
+      "loss": 2.0638,
+      "step": 2290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9407680410674915e-05,
+      "loss": 1.9848,
+      "step": 2300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.940438974628978e-05,
+      "loss": 1.718,
+      "step": 2310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.940109908190464e-05,
+      "loss": 1.9206,
+      "step": 2320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.93978084175195e-05,
+      "loss": 2.0166,
+      "step": 2330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.939451775313436e-05,
+      "loss": 2.1128,
+      "step": 2340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.939122708874922e-05,
+      "loss": 1.8732,
+      "step": 2350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.938793642436408e-05,
+      "loss": 1.8579,
+      "step": 2360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9384645759978944e-05,
+      "loss": 1.9509,
+      "step": 2370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.93813550955938e-05,
+      "loss": 1.8857,
+      "step": 2380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.937806443120867e-05,
+      "loss": 1.8754,
+      "step": 2390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.937477376682352e-05,
+      "loss": 1.8078,
+      "step": 2400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.937148310243839e-05,
+      "loss": 1.8664,
+      "step": 2410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9368192438053244e-05,
+      "loss": 1.83,
+      "step": 2420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9364901773668105e-05,
+      "loss": 1.9999,
+      "step": 2430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9361611109282966e-05,
+      "loss": 1.9869,
+      "step": 2440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.935832044489783e-05,
+      "loss": 1.9046,
+      "step": 2450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.935502978051269e-05,
+      "loss": 2.0755,
+      "step": 2460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.935173911612755e-05,
+      "loss": 1.9715,
+      "step": 2470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9348448451742405e-05,
+      "loss": 1.9085,
+      "step": 2480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.934515778735727e-05,
+      "loss": 1.9854,
+      "step": 2490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.934186712297213e-05,
+      "loss": 2.0501,
+      "step": 2500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9338576458586995e-05,
+      "loss": 1.9246,
+      "step": 2510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.933528579420185e-05,
+      "loss": 1.8044,
+      "step": 2520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.933199512981671e-05,
+      "loss": 1.9461,
+      "step": 2530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.932870446543157e-05,
+      "loss": 1.7756,
+      "step": 2540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9325413801046433e-05,
+      "loss": 1.9201,
+      "step": 2550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9322123136661295e-05,
+      "loss": 1.8932,
+      "step": 2560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9318832472276156e-05,
+      "loss": 1.8404,
+      "step": 2570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.931554180789101e-05,
+      "loss": 1.8578,
+      "step": 2580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.931225114350588e-05,
+      "loss": 1.842,
+      "step": 2590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.930896047912073e-05,
+      "loss": 1.7498,
+      "step": 2600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.93056698147356e-05,
+      "loss": 2.0665,
+      "step": 2610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9302379150350456e-05,
+      "loss": 1.8835,
+      "step": 2620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.929908848596532e-05,
+      "loss": 1.9448,
+      "step": 2630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.929579782158018e-05,
+      "loss": 2.0113,
+      "step": 2640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.929250715719504e-05,
+      "loss": 1.9871,
+      "step": 2650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92892164928099e-05,
+      "loss": 1.7794,
+      "step": 2660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.928592582842476e-05,
+      "loss": 1.8896,
+      "step": 2670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.928263516403962e-05,
+      "loss": 1.9832,
+      "step": 2680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9279344499654484e-05,
+      "loss": 1.8607,
+      "step": 2690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9276053835269346e-05,
+      "loss": 1.8563,
+      "step": 2700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927276317088421e-05,
+      "loss": 1.929,
+      "step": 2710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926947250649906e-05,
+      "loss": 1.8829,
+      "step": 2720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926618184211392e-05,
+      "loss": 2.0024,
+      "step": 2730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9262891177728784e-05,
+      "loss": 1.9082,
+      "step": 2740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9259600513343645e-05,
+      "loss": 1.9097,
+      "step": 2750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9256309848958507e-05,
+      "loss": 1.8617,
+      "step": 2760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925301918457337e-05,
+      "loss": 1.8923,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924972852018823e-05,
+      "loss": 1.9746,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924643785580309e-05,
+      "loss": 1.8894,
+      "step": 2790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.924314719141795e-05,
+      "loss": 1.9056,
+      "step": 2800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.923985652703281e-05,
+      "loss": 1.9707,
+      "step": 2810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9236565862647674e-05,
+      "loss": 2.0806,
+      "step": 2820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.923327519826253e-05,
+      "loss": 1.9778,
+      "step": 2830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.92299845338774e-05,
+      "loss": 1.8693,
+      "step": 2840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.922669386949225e-05,
+      "loss": 2.0113,
+      "step": 2850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.922340320510712e-05,
+      "loss": 1.965,
+      "step": 2860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9220112540721974e-05,
+      "loss": 1.8569,
+      "step": 2870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9216821876336835e-05,
+      "loss": 2.1064,
+      "step": 2880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9213531211951696e-05,
+      "loss": 1.8588,
+      "step": 2890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.921024054756656e-05,
+      "loss": 1.951,
+      "step": 2900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.920694988318142e-05,
+      "loss": 1.9863,
+      "step": 2910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.920365921879628e-05,
+      "loss": 2.0215,
+      "step": 2920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9200368554411135e-05,
+      "loss": 1.9515,
+      "step": 2930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9197077890026e-05,
+      "loss": 1.819,
+      "step": 2940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.919378722564086e-05,
+      "loss": 2.0559,
+      "step": 2950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9190496561255725e-05,
+      "loss": 1.9761,
+      "step": 2960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.918720589687058e-05,
+      "loss": 1.9183,
+      "step": 2970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.918391523248544e-05,
+      "loss": 2.0929,
+      "step": 2980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.91806245681003e-05,
+      "loss": 2.1125,
+      "step": 2990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9177333903715163e-05,
+      "loss": 1.9729,
+      "step": 3000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9174043239330025e-05,
+      "loss": 1.9254,
+      "step": 3010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9170752574944886e-05,
+      "loss": 1.9503,
+      "step": 3020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.916746191055974e-05,
+      "loss": 1.8245,
+      "step": 3030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.916417124617461e-05,
+      "loss": 1.8783,
+      "step": 3040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.916088058178946e-05,
+      "loss": 1.753,
+      "step": 3050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.915758991740433e-05,
+      "loss": 1.9049,
+      "step": 3060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9154299253019186e-05,
+      "loss": 2.0126,
+      "step": 3070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.915100858863405e-05,
+      "loss": 1.7852,
+      "step": 3080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.914771792424891e-05,
+      "loss": 1.7619,
+      "step": 3090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.914442725986377e-05,
+      "loss": 1.8758,
+      "step": 3100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.914113659547863e-05,
+      "loss": 1.7899,
+      "step": 3110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.913784593109349e-05,
+      "loss": 1.7791,
+      "step": 3120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9134555266708346e-05,
+      "loss": 1.9005,
+      "step": 3130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9131264602323214e-05,
+      "loss": 1.8747,
+      "step": 3140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.912797393793807e-05,
+      "loss": 2.0669,
+      "step": 3150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.912468327355294e-05,
+      "loss": 1.9952,
+      "step": 3160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.912139260916779e-05,
+      "loss": 1.9801,
+      "step": 3170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.911810194478265e-05,
+      "loss": 1.9435,
+      "step": 3180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9114811280397514e-05,
+      "loss": 1.8379,
+      "step": 3190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9111520616012375e-05,
+      "loss": 1.9017,
+      "step": 3200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9108229951627236e-05,
+      "loss": 2.1727,
+      "step": 3210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.91049392872421e-05,
+      "loss": 1.9167,
+      "step": 3220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910164862285695e-05,
+      "loss": 2.0586,
+      "step": 3230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909835795847182e-05,
+      "loss": 1.8575,
+      "step": 3240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9095067294086675e-05,
+      "loss": 1.9234,
+      "step": 3250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909177662970154e-05,
+      "loss": 1.8547,
+      "step": 3260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90884859653164e-05,
+      "loss": 1.9933,
+      "step": 3270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908519530093126e-05,
+      "loss": 1.8008,
+      "step": 3280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908190463654612e-05,
+      "loss": 2.0968,
+      "step": 3290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907861397216098e-05,
+      "loss": 2.2159,
+      "step": 3300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.907532330777584e-05,
+      "loss": 1.9033,
+      "step": 3310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9072032643390704e-05,
+      "loss": 1.7113,
+      "step": 3320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.906874197900556e-05,
+      "loss": 1.7739,
+      "step": 3330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9065451314620426e-05,
+      "loss": 1.7911,
+      "step": 3340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.906216065023528e-05,
+      "loss": 1.8077,
+      "step": 3350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.905886998585015e-05,
+      "loss": 1.903,
+      "step": 3360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9055579321465e-05,
+      "loss": 1.8583,
+      "step": 3370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9052288657079864e-05,
+      "loss": 1.8606,
+      "step": 3380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9048997992694726e-05,
+      "loss": 1.9606,
+      "step": 3390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.904570732830959e-05,
+      "loss": 2.1126,
+      "step": 3400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.904241666392445e-05,
+      "loss": 1.9519,
+      "step": 3410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.903912599953931e-05,
+      "loss": 1.7887,
+      "step": 3420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.903583533515417e-05,
+      "loss": 1.8341,
+      "step": 3430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.903254467076903e-05,
+      "loss": 1.9576,
+      "step": 3440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.902925400638389e-05,
+      "loss": 1.916,
+      "step": 3450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9025963341998755e-05,
+      "loss": 2.0817,
+      "step": 3460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9022672677613616e-05,
+      "loss": 1.981,
+      "step": 3470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.901938201322847e-05,
+      "loss": 1.8514,
+      "step": 3480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.901609134884333e-05,
+      "loss": 1.9675,
+      "step": 3490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.901280068445819e-05,
+      "loss": 1.8809,
+      "step": 3500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9009510020073054e-05,
+      "loss": 2.2347,
+      "step": 3510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9006219355687915e-05,
+      "loss": 1.8799,
+      "step": 3520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.900292869130278e-05,
+      "loss": 1.821,
+      "step": 3530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.899963802691764e-05,
+      "loss": 2.0434,
+      "step": 3540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.89963473625325e-05,
+      "loss": 1.8403,
+      "step": 3550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.899305669814736e-05,
+      "loss": 1.8211,
+      "step": 3560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.898976603376222e-05,
+      "loss": 2.0645,
+      "step": 3570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8986475369377076e-05,
+      "loss": 1.8497,
+      "step": 3580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8983184704991944e-05,
+      "loss": 1.7095,
+      "step": 3590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.89798940406068e-05,
+      "loss": 1.9234,
+      "step": 3600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.897660337622167e-05,
+      "loss": 1.9017,
+      "step": 3610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.897331271183652e-05,
+      "loss": 1.9615,
+      "step": 3620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.897002204745138e-05,
+      "loss": 1.8477,
+      "step": 3630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8966731383066244e-05,
+      "loss": 1.924,
+      "step": 3640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8963440718681105e-05,
+      "loss": 1.8955,
+      "step": 3650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8960150054295966e-05,
+      "loss": 2.0565,
+      "step": 3660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.895685938991083e-05,
+      "loss": 1.983,
+      "step": 3670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.895356872552568e-05,
+      "loss": 1.9916,
+      "step": 3680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.895027806114055e-05,
+      "loss": 1.9148,
+      "step": 3690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8946987396755405e-05,
+      "loss": 1.8098,
+      "step": 3700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894369673237027e-05,
+      "loss": 1.8759,
+      "step": 3710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894040606798513e-05,
+      "loss": 1.7087,
+      "step": 3720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893711540359999e-05,
+      "loss": 1.6715,
+      "step": 3730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893382473921485e-05,
+      "loss": 1.8722,
+      "step": 3740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893053407482971e-05,
+      "loss": 1.8387,
+      "step": 3750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892724341044457e-05,
+      "loss": 1.8821,
+      "step": 3760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8923952746059434e-05,
+      "loss": 1.8337,
+      "step": 3770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892066208167429e-05,
+      "loss": 1.8722,
+      "step": 3780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8917371417289156e-05,
+      "loss": 1.9513,
+      "step": 3790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891408075290401e-05,
+      "loss": 1.8798,
+      "step": 3800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891079008851888e-05,
+      "loss": 1.8129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.890749942413373e-05,
+      "loss": 2.0258,
+      "step": 3820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8904208759748594e-05,
+      "loss": 1.8908,
+      "step": 3830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8900918095363456e-05,
+      "loss": 1.715,
+      "step": 3840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.889762743097832e-05,
+      "loss": 1.8147,
+      "step": 3850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.889433676659318e-05,
+      "loss": 1.796,
+      "step": 3860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.889104610220804e-05,
+      "loss": 1.9086,
+      "step": 3870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8887755437822894e-05,
+      "loss": 1.9525,
+      "step": 3880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.888446477343776e-05,
+      "loss": 1.9139,
+      "step": 3890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8881174109052616e-05,
+      "loss": 1.8291,
+      "step": 3900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8877883444667485e-05,
+      "loss": 1.8403,
+      "step": 3910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.887459278028234e-05,
+      "loss": 1.9693,
+      "step": 3920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.88713021158972e-05,
+      "loss": 1.8304,
+      "step": 3930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.886801145151206e-05,
+      "loss": 1.97,
+      "step": 3940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.886472078712692e-05,
+      "loss": 1.9555,
+      "step": 3950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8861430122741784e-05,
+      "loss": 2.0348,
+      "step": 3960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8858139458356645e-05,
+      "loss": 1.7302,
+      "step": 3970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.88548487939715e-05,
+      "loss": 1.9505,
+      "step": 3980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.885155812958637e-05,
+      "loss": 1.8607,
+      "step": 3990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.884826746520122e-05,
+      "loss": 2.1188,
+      "step": 4000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.884497680081609e-05,
+      "loss": 1.8797,
+      "step": 4010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8841686136430945e-05,
+      "loss": 1.741,
+      "step": 4020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8838395472045806e-05,
+      "loss": 1.7861,
+      "step": 4030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.883510480766067e-05,
+      "loss": 2.158,
+      "step": 4040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.883181414327553e-05,
+      "loss": 1.9228,
+      "step": 4050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.882852347889039e-05,
+      "loss": 1.9805,
+      "step": 4060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.882523281450525e-05,
+      "loss": 1.9632,
+      "step": 4070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8821942150120106e-05,
+      "loss": 1.8809,
+      "step": 4080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8818651485734974e-05,
+      "loss": 1.8144,
+      "step": 4090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.881536082134983e-05,
+      "loss": 1.9866,
+      "step": 4100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8812070156964696e-05,
+      "loss": 1.7036,
+      "step": 4110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.880877949257955e-05,
+      "loss": 1.885,
+      "step": 4120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.880548882819441e-05,
+      "loss": 1.8645,
+      "step": 4130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.880219816380927e-05,
+      "loss": 1.9178,
+      "step": 4140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8798907499424135e-05,
+      "loss": 1.8295,
+      "step": 4150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8795616835038996e-05,
+      "loss": 1.7523,
+      "step": 4160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.879232617065386e-05,
+      "loss": 1.7377,
+      "step": 4170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.878903550626872e-05,
+      "loss": 1.834,
+      "step": 4180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.878574484188358e-05,
+      "loss": 1.6745,
+      "step": 4190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.878245417749844e-05,
+      "loss": 1.9536,
+      "step": 4200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87791635131133e-05,
+      "loss": 1.7971,
+      "step": 4210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8775872848728163e-05,
+      "loss": 1.9063,
+      "step": 4220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877258218434302e-05,
+      "loss": 1.7274,
+      "step": 4230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876929151995788e-05,
+      "loss": 1.8126,
+      "step": 4240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876600085557274e-05,
+      "loss": 1.8914,
+      "step": 4250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87627101911876e-05,
+      "loss": 1.7621,
+      "step": 4260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875941952680246e-05,
+      "loss": 1.8952,
+      "step": 4270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8756128862417324e-05,
+      "loss": 1.9618,
+      "step": 4280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8752838198032186e-05,
+      "loss": 1.975,
+      "step": 4290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874954753364705e-05,
+      "loss": 1.8928,
+      "step": 4300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874625686926191e-05,
+      "loss": 1.8302,
+      "step": 4310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.874296620487677e-05,
+      "loss": 1.8209,
+      "step": 4320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8739675540491624e-05,
+      "loss": 1.886,
+      "step": 4330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.873638487610649e-05,
+      "loss": 1.8889,
+      "step": 4340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8733094211721346e-05,
+      "loss": 1.9343,
+      "step": 4350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8729803547336214e-05,
+      "loss": 1.8728,
+      "step": 4360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.872651288295107e-05,
+      "loss": 1.817,
+      "step": 4370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.872322221856593e-05,
+      "loss": 1.791,
+      "step": 4380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.871993155418079e-05,
+      "loss": 2.0382,
+      "step": 4390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.871664088979565e-05,
+      "loss": 1.8611,
+      "step": 4400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8713350225410514e-05,
+      "loss": 1.9317,
+      "step": 4410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8710059561025375e-05,
+      "loss": 2.0094,
+      "step": 4420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.870676889664023e-05,
+      "loss": 2.0178,
+      "step": 4430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.87034782322551e-05,
+      "loss": 1.9583,
+      "step": 4440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.870018756786995e-05,
+      "loss": 1.9544,
+      "step": 4450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.869689690348482e-05,
+      "loss": 1.7916,
+      "step": 4460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8693606239099675e-05,
+      "loss": 1.881,
+      "step": 4470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8690315574714536e-05,
+      "loss": 1.8962,
+      "step": 4480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.86870249103294e-05,
+      "loss": 2.0597,
+      "step": 4490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.868373424594426e-05,
+      "loss": 1.9615,
+      "step": 4500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.868044358155912e-05,
+      "loss": 1.9067,
+      "step": 4510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.867715291717398e-05,
+      "loss": 1.8385,
+      "step": 4520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8673862252788836e-05,
+      "loss": 1.8345,
+      "step": 4530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8670571588403704e-05,
+      "loss": 1.861,
+      "step": 4540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.866728092401856e-05,
+      "loss": 1.7929,
+      "step": 4550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8663990259633426e-05,
+      "loss": 1.7962,
+      "step": 4560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.866069959524828e-05,
+      "loss": 1.9557,
+      "step": 4570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.865740893086314e-05,
+      "loss": 1.716,
+      "step": 4580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8654118266478e-05,
+      "loss": 1.7977,
+      "step": 4590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8650827602092865e-05,
+      "loss": 1.8899,
+      "step": 4600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8647536937707726e-05,
+      "loss": 1.8598,
+      "step": 4610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.864424627332259e-05,
+      "loss": 1.979,
+      "step": 4620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.864095560893744e-05,
+      "loss": 1.9174,
+      "step": 4630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.863766494455231e-05,
+      "loss": 1.8572,
+      "step": 4640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8634374280167164e-05,
+      "loss": 1.8991,
+      "step": 4650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.863108361578203e-05,
+      "loss": 2.0926,
+      "step": 4660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.862779295139689e-05,
+      "loss": 1.9904,
+      "step": 4670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.862450228701175e-05,
+      "loss": 1.6795,
+      "step": 4680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.862121162262661e-05,
+      "loss": 1.6608,
+      "step": 4690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861792095824147e-05,
+      "loss": 1.8704,
+      "step": 4700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861463029385633e-05,
+      "loss": 1.9444,
+      "step": 4710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861133962947119e-05,
+      "loss": 1.8511,
+      "step": 4720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860804896508605e-05,
+      "loss": 1.8062,
+      "step": 4730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8604758300700916e-05,
+      "loss": 1.9736,
+      "step": 4740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860146763631577e-05,
+      "loss": 1.8299,
+      "step": 4750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859817697193064e-05,
+      "loss": 1.9939,
+      "step": 4760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859488630754549e-05,
+      "loss": 1.8205,
+      "step": 4770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859159564316036e-05,
+      "loss": 1.9376,
+      "step": 4780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8588304978775215e-05,
+      "loss": 1.8424,
+      "step": 4790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8585014314390076e-05,
+      "loss": 2.1722,
+      "step": 4800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858172365000494e-05,
+      "loss": 1.9564,
+      "step": 4810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85784329856198e-05,
+      "loss": 2.0167,
+      "step": 4820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.857514232123466e-05,
+      "loss": 1.9115,
+      "step": 4830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.857185165684952e-05,
+      "loss": 1.8815,
+      "step": 4840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8568560992464376e-05,
+      "loss": 1.8246,
+      "step": 4850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8565270328079244e-05,
+      "loss": 1.8586,
+      "step": 4860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.85619796636941e-05,
+      "loss": 1.8437,
+      "step": 4870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8558688999308966e-05,
+      "loss": 1.906,
+      "step": 4880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.855539833492382e-05,
+      "loss": 1.8251,
+      "step": 4890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.855210767053868e-05,
+      "loss": 1.7247,
+      "step": 4900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8548817006153544e-05,
+      "loss": 1.7323,
+      "step": 4910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8545526341768405e-05,
+      "loss": 1.8871,
+      "step": 4920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8542235677383266e-05,
+      "loss": 1.9206,
+      "step": 4930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.853894501299813e-05,
+      "loss": 1.839,
+      "step": 4940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.853565434861299e-05,
+      "loss": 1.7973,
+      "step": 4950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.853236368422785e-05,
+      "loss": 1.7036,
+      "step": 4960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.852907301984271e-05,
+      "loss": 1.984,
+      "step": 4970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.852578235545757e-05,
+      "loss": 1.732,
+      "step": 4980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8522491691072434e-05,
+      "loss": 2.1892,
+      "step": 4990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.851920102668729e-05,
+      "loss": 1.8753,
+      "step": 5000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.851591036230215e-05,
+      "loss": 1.738,
+      "step": 5010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.851261969791701e-05,
+      "loss": 1.8545,
+      "step": 5020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.850932903353187e-05,
+      "loss": 1.8685,
+      "step": 5030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.850603836914673e-05,
+      "loss": 1.9065,
+      "step": 5040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8502747704761594e-05,
+      "loss": 1.8289,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8499457040376456e-05,
+      "loss": 1.7633,
+      "step": 5060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.849616637599132e-05,
+      "loss": 1.8847,
+      "step": 5070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.849287571160618e-05,
+      "loss": 1.8368,
+      "step": 5080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.848958504722104e-05,
+      "loss": 1.9366,
+      "step": 5090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8486294382835894e-05,
+      "loss": 1.9579,
+      "step": 5100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.848300371845076e-05,
+      "loss": 2.1089,
+      "step": 5110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8479713054065617e-05,
+      "loss": 1.7692,
+      "step": 5120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8476422389680485e-05,
+      "loss": 2.0565,
+      "step": 5130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.847313172529534e-05,
+      "loss": 2.125,
+      "step": 5140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84698410609102e-05,
+      "loss": 1.8919,
+      "step": 5150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.846655039652506e-05,
+      "loss": 1.8407,
+      "step": 5160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.846325973213992e-05,
+      "loss": 1.7856,
+      "step": 5170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8459969067754784e-05,
+      "loss": 1.9605,
+      "step": 5180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8456678403369645e-05,
+      "loss": 1.8081,
+      "step": 5190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84533877389845e-05,
+      "loss": 1.8959,
+      "step": 5200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.845009707459937e-05,
+      "loss": 1.8723,
+      "step": 5210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844680641021422e-05,
+      "loss": 1.9781,
+      "step": 5220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844351574582909e-05,
+      "loss": 1.7895,
+      "step": 5230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8440225081443945e-05,
+      "loss": 1.917,
+      "step": 5240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8436934417058806e-05,
+      "loss": 1.7596,
+      "step": 5250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843364375267367e-05,
+      "loss": 1.8061,
+      "step": 5260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843035308828853e-05,
+      "loss": 1.8981,
+      "step": 5270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842706242390339e-05,
+      "loss": 1.8623,
+      "step": 5280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842377175951825e-05,
+      "loss": 2.01,
+      "step": 5290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8420481095133106e-05,
+      "loss": 2.0324,
+      "step": 5300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8417190430747974e-05,
+      "loss": 1.7607,
+      "step": 5310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841389976636283e-05,
+      "loss": 1.7315,
+      "step": 5320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8410609101977696e-05,
+      "loss": 1.7867,
+      "step": 5330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.840731843759255e-05,
+      "loss": 2.0223,
+      "step": 5340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.840402777320741e-05,
+      "loss": 1.8845,
+      "step": 5350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8400737108822273e-05,
+      "loss": 1.7175,
+      "step": 5360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8397446444437135e-05,
+      "loss": 1.8147,
+      "step": 5370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8394155780051996e-05,
+      "loss": 2.0078,
+      "step": 5380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.839086511566686e-05,
+      "loss": 1.8579,
+      "step": 5390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.838757445128171e-05,
+      "loss": 1.9075,
+      "step": 5400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.838428378689658e-05,
+      "loss": 1.9317,
+      "step": 5410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8380993122511434e-05,
+      "loss": 2.0084,
+      "step": 5420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.83777024581263e-05,
+      "loss": 1.9265,
+      "step": 5430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.837441179374116e-05,
+      "loss": 1.753,
+      "step": 5440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.837112112935602e-05,
+      "loss": 1.8463,
+      "step": 5450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.836783046497088e-05,
+      "loss": 1.7196,
+      "step": 5460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.836453980058574e-05,
+      "loss": 1.777,
+      "step": 5470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.83612491362006e-05,
+      "loss": 1.8873,
+      "step": 5480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.835795847181546e-05,
+      "loss": 1.7196,
+      "step": 5490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.835466780743032e-05,
+      "loss": 1.7203,
+      "step": 5500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8351377143045186e-05,
+      "loss": 1.7229,
+      "step": 5510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.834808647866004e-05,
+      "loss": 1.7642,
+      "step": 5520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.834479581427491e-05,
+      "loss": 1.8184,
+      "step": 5530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.834150514988976e-05,
+      "loss": 1.8567,
+      "step": 5540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8338214485504624e-05,
+      "loss": 1.853,
+      "step": 5550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8334923821119485e-05,
+      "loss": 1.6153,
+      "step": 5560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8331633156734347e-05,
+      "loss": 1.9962,
+      "step": 5570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.832834249234921e-05,
+      "loss": 1.8045,
+      "step": 5580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.832505182796407e-05,
+      "loss": 1.5958,
+      "step": 5590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8321761163578924e-05,
+      "loss": 1.7563,
+      "step": 5600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.831847049919379e-05,
+      "loss": 1.7812,
+      "step": 5610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8315179834808646e-05,
+      "loss": 1.9425,
+      "step": 5620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8311889170423514e-05,
+      "loss": 1.8726,
+      "step": 5630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.830859850603837e-05,
+      "loss": 1.9261,
+      "step": 5640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.830530784165323e-05,
+      "loss": 1.7498,
+      "step": 5650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.830201717726809e-05,
+      "loss": 1.8798,
+      "step": 5660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.829872651288295e-05,
+      "loss": 2.2903,
+      "step": 5670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8295435848497814e-05,
+      "loss": 1.7391,
+      "step": 5680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8292145184112675e-05,
+      "loss": 1.8389,
+      "step": 5690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8288854519727536e-05,
+      "loss": 1.8517,
+      "step": 5700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82855638553424e-05,
+      "loss": 1.8308,
+      "step": 5710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.828227319095726e-05,
+      "loss": 1.9622,
+      "step": 5720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827898252657212e-05,
+      "loss": 1.9601,
+      "step": 5730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827569186218698e-05,
+      "loss": 1.7735,
+      "step": 5740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8272401197801836e-05,
+      "loss": 1.9597,
+      "step": 5750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82691105334167e-05,
+      "loss": 1.79,
+      "step": 5760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826581986903156e-05,
+      "loss": 1.832,
+      "step": 5770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826252920464642e-05,
+      "loss": 1.8457,
+      "step": 5780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825923854026128e-05,
+      "loss": 1.8679,
+      "step": 5790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825594787587614e-05,
+      "loss": 1.7806,
+      "step": 5800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8252657211491e-05,
+      "loss": 2.0841,
+      "step": 5810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8249366547105865e-05,
+      "loss": 1.97,
+      "step": 5820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8246075882720726e-05,
+      "loss": 1.8403,
+      "step": 5830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824278521833559e-05,
+      "loss": 1.8644,
+      "step": 5840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.823949455395044e-05,
+      "loss": 1.972,
+      "step": 5850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.823620388956531e-05,
+      "loss": 2.0377,
+      "step": 5860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8232913225180164e-05,
+      "loss": 1.6788,
+      "step": 5870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822962256079503e-05,
+      "loss": 1.7204,
+      "step": 5880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822633189640989e-05,
+      "loss": 1.9359,
+      "step": 5890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822304123202475e-05,
+      "loss": 1.7799,
+      "step": 5900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821975056763961e-05,
+      "loss": 1.9492,
+      "step": 5910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821645990325447e-05,
+      "loss": 1.9304,
+      "step": 5920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821316923886933e-05,
+      "loss": 1.914,
+      "step": 5930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.820987857448419e-05,
+      "loss": 2.0541,
+      "step": 5940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.820658791009905e-05,
+      "loss": 1.8732,
+      "step": 5950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8203297245713916e-05,
+      "loss": 1.8967,
+      "step": 5960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.820000658132877e-05,
+      "loss": 1.9064,
+      "step": 5970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819671591694364e-05,
+      "loss": 1.8677,
+      "step": 5980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819342525255849e-05,
+      "loss": 1.9425,
+      "step": 5990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8190134588173354e-05,
+      "loss": 1.7803,
+      "step": 6000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8186843923788215e-05,
+      "loss": 1.8053,
+      "step": 6010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8183553259403076e-05,
+      "loss": 1.8868,
+      "step": 6020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818026259501794e-05,
+      "loss": 1.806,
+      "step": 6030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.81769719306328e-05,
+      "loss": 1.8701,
+      "step": 6040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8173681266247653e-05,
+      "loss": 1.7132,
+      "step": 6050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817039060186252e-05,
+      "loss": 1.8105,
+      "step": 6060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8167099937477376e-05,
+      "loss": 1.9718,
+      "step": 6070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8163809273092244e-05,
+      "loss": 2.0107,
+      "step": 6080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.81605186087071e-05,
+      "loss": 1.8138,
+      "step": 6090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.815722794432196e-05,
+      "loss": 2.0459,
+      "step": 6100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.815393727993682e-05,
+      "loss": 1.7638,
+      "step": 6110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.815064661555168e-05,
+      "loss": 1.806,
+      "step": 6120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8147355951166544e-05,
+      "loss": 1.7442,
+      "step": 6130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8144065286781405e-05,
+      "loss": 1.8707,
+      "step": 6140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814077462239626e-05,
+      "loss": 2.0113,
+      "step": 6150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813748395801113e-05,
+      "loss": 1.8617,
+      "step": 6160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813419329362598e-05,
+      "loss": 1.8946,
+      "step": 6170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813090262924085e-05,
+      "loss": 1.7981,
+      "step": 6180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8127611964855704e-05,
+      "loss": 1.7861,
+      "step": 6190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8124321300470566e-05,
+      "loss": 1.9038,
+      "step": 6200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812103063608543e-05,
+      "loss": 1.7292,
+      "step": 6210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811773997170029e-05,
+      "loss": 1.7114,
+      "step": 6220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811444930731515e-05,
+      "loss": 1.934,
+      "step": 6230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811115864293001e-05,
+      "loss": 1.9502,
+      "step": 6240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8107867978544865e-05,
+      "loss": 2.0247,
+      "step": 6250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810457731415973e-05,
+      "loss": 1.9272,
+      "step": 6260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810128664977459e-05,
+      "loss": 1.9451,
+      "step": 6270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8097995985389456e-05,
+      "loss": 1.863,
+      "step": 6280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809470532100431e-05,
+      "loss": 1.9589,
+      "step": 6290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809141465661917e-05,
+      "loss": 1.9046,
+      "step": 6300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808812399223403e-05,
+      "loss": 1.8408,
+      "step": 6310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8084833327848894e-05,
+      "loss": 2.0051,
+      "step": 6320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8081542663463755e-05,
+      "loss": 1.8098,
+      "step": 6330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807825199907862e-05,
+      "loss": 1.8583,
+      "step": 6340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807496133469348e-05,
+      "loss": 1.8622,
+      "step": 6350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807167067030834e-05,
+      "loss": 1.7837,
+      "step": 6360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8068380005923194e-05,
+      "loss": 1.8612,
+      "step": 6370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806508934153806e-05,
+      "loss": 1.7846,
+      "step": 6380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8061798677152916e-05,
+      "loss": 1.7457,
+      "step": 6390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805850801276778e-05,
+      "loss": 1.7818,
+      "step": 6400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805521734838264e-05,
+      "loss": 2.0448,
+      "step": 6410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80519266839975e-05,
+      "loss": 1.9417,
+      "step": 6420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804863601961236e-05,
+      "loss": 1.9608,
+      "step": 6430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804534535522722e-05,
+      "loss": 1.6252,
+      "step": 6440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8042054690842084e-05,
+      "loss": 1.7988,
+      "step": 6450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8038764026456945e-05,
+      "loss": 1.8129,
+      "step": 6460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8035473362071806e-05,
+      "loss": 1.7885,
+      "step": 6470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803218269768667e-05,
+      "loss": 1.8218,
+      "step": 6480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802889203330153e-05,
+      "loss": 2.0066,
+      "step": 6490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8025601368916383e-05,
+      "loss": 1.741,
+      "step": 6500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802231070453125e-05,
+      "loss": 1.85,
+      "step": 6510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8019020040146106e-05,
+      "loss": 1.9927,
+      "step": 6520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801572937576097e-05,
+      "loss": 1.6563,
+      "step": 6530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801243871137583e-05,
+      "loss": 1.8027,
+      "step": 6540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800914804699069e-05,
+      "loss": 1.8765,
+      "step": 6550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800585738260555e-05,
+      "loss": 1.7888,
+      "step": 6560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800256671822041e-05,
+      "loss": 1.7956,
+      "step": 6570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7999276053835274e-05,
+      "loss": 1.7884,
+      "step": 6580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7995985389450135e-05,
+      "loss": 1.6752,
+      "step": 6590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799269472506499e-05,
+      "loss": 1.833,
+      "step": 6600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798940406067986e-05,
+      "loss": 1.8408,
+      "step": 6610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798611339629471e-05,
+      "loss": 1.9793,
+      "step": 6620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798282273190958e-05,
+      "loss": 1.7693,
+      "step": 6630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7979532067524434e-05,
+      "loss": 1.8522,
+      "step": 6640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7976241403139296e-05,
+      "loss": 1.9223,
+      "step": 6650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797295073875416e-05,
+      "loss": 1.8768,
+      "step": 6660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796966007436902e-05,
+      "loss": 1.8288,
+      "step": 6670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796636940998388e-05,
+      "loss": 1.8843,
+      "step": 6680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796307874559874e-05,
+      "loss": 1.9496,
+      "step": 6690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7959788081213595e-05,
+      "loss": 1.8028,
+      "step": 6700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795649741682846e-05,
+      "loss": 1.9103,
+      "step": 6710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795320675244332e-05,
+      "loss": 1.8263,
+      "step": 6720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7949916088058186e-05,
+      "loss": 1.8783,
+      "step": 6730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794662542367304e-05,
+      "loss": 1.5846,
+      "step": 6740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79433347592879e-05,
+      "loss": 1.7698,
+      "step": 6750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794004409490276e-05,
+      "loss": 1.8226,
+      "step": 6760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7936753430517624e-05,
+      "loss": 1.6777,
+      "step": 6770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7933462766132485e-05,
+      "loss": 1.8749,
+      "step": 6780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7930172101747347e-05,
+      "loss": 2.0425,
+      "step": 6790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79268814373622e-05,
+      "loss": 1.7175,
+      "step": 6800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792359077297707e-05,
+      "loss": 1.9123,
+      "step": 6810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7920300108591924e-05,
+      "loss": 1.8508,
+      "step": 6820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791700944420679e-05,
+      "loss": 1.8299,
+      "step": 6830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7913718779821646e-05,
+      "loss": 1.94,
+      "step": 6840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791042811543651e-05,
+      "loss": 1.6944,
+      "step": 6850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790713745105137e-05,
+      "loss": 1.8723,
+      "step": 6860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.790384678666623e-05,
+      "loss": 1.7967,
+      "step": 6870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.790055612228109e-05,
+      "loss": 1.8069,
+      "step": 6880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.789726545789595e-05,
+      "loss": 1.7871,
+      "step": 6890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.789397479351081e-05,
+      "loss": 1.8991,
+      "step": 6900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7890684129125675e-05,
+      "loss": 1.7123,
+      "step": 6910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.788739346474053e-05,
+      "loss": 1.7576,
+      "step": 6920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78841028003554e-05,
+      "loss": 1.7997,
+      "step": 6930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.788081213597025e-05,
+      "loss": 1.805,
+      "step": 6940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.787752147158511e-05,
+      "loss": 1.8525,
+      "step": 6950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7874230807199975e-05,
+      "loss": 1.8752,
+      "step": 6960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7870940142814836e-05,
+      "loss": 1.8621,
+      "step": 6970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78676494784297e-05,
+      "loss": 2.0351,
+      "step": 6980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.786435881404456e-05,
+      "loss": 1.8757,
+      "step": 6990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.786106814965941e-05,
+      "loss": 1.8023,
+      "step": 7000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785777748527428e-05,
+      "loss": 1.6963,
+      "step": 7010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7854486820889135e-05,
+      "loss": 1.8984,
+      "step": 7020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7851196156504003e-05,
+      "loss": 1.8613,
+      "step": 7030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784790549211886e-05,
+      "loss": 1.6638,
+      "step": 7040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784461482773372e-05,
+      "loss": 1.8564,
+      "step": 7050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784132416334858e-05,
+      "loss": 1.8301,
+      "step": 7060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783803349896344e-05,
+      "loss": 1.7105,
+      "step": 7070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78347428345783e-05,
+      "loss": 1.9194,
+      "step": 7080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7831452170193164e-05,
+      "loss": 1.7737,
+      "step": 7090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7828161505808026e-05,
+      "loss": 1.9588,
+      "step": 7100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782487084142289e-05,
+      "loss": 1.8482,
+      "step": 7110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782158017703775e-05,
+      "loss": 1.7108,
+      "step": 7120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781828951265261e-05,
+      "loss": 1.9564,
+      "step": 7130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7814998848267464e-05,
+      "loss": 1.9115,
+      "step": 7140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7811708183882325e-05,
+      "loss": 1.8706,
+      "step": 7150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7808417519497186e-05,
+      "loss": 1.8307,
+      "step": 7160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780512685511205e-05,
+      "loss": 1.834,
+      "step": 7170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780183619072691e-05,
+      "loss": 1.6916,
+      "step": 7180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779854552634177e-05,
+      "loss": 2.0029,
+      "step": 7190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779525486195663e-05,
+      "loss": 1.8758,
+      "step": 7200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779196419757149e-05,
+      "loss": 1.9611,
+      "step": 7210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7788673533186354e-05,
+      "loss": 1.986,
+      "step": 7220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7785382868801215e-05,
+      "loss": 1.9005,
+      "step": 7230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7782092204416077e-05,
+      "loss": 1.9581,
+      "step": 7240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777880154003093e-05,
+      "loss": 1.7896,
+      "step": 7250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77755108756458e-05,
+      "loss": 1.645,
+      "step": 7260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7772220211260654e-05,
+      "loss": 1.8068,
+      "step": 7270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776892954687552e-05,
+      "loss": 2.0632,
+      "step": 7280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7765638882490376e-05,
+      "loss": 1.8217,
+      "step": 7290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776234821810524e-05,
+      "loss": 1.9143,
+      "step": 7300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77590575537201e-05,
+      "loss": 1.7907,
+      "step": 7310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775576688933496e-05,
+      "loss": 1.8071,
+      "step": 7320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775247622494982e-05,
+      "loss": 1.859,
+      "step": 7330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774918556056468e-05,
+      "loss": 1.7917,
+      "step": 7340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774589489617954e-05,
+      "loss": 1.8097,
+      "step": 7350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7742604231794405e-05,
+      "loss": 2.0313,
+      "step": 7360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.773931356740926e-05,
+      "loss": 1.794,
+      "step": 7370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.773602290302413e-05,
+      "loss": 1.7582,
+      "step": 7380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.773273223863898e-05,
+      "loss": 1.7884,
+      "step": 7390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.772944157425384e-05,
+      "loss": 1.9092,
+      "step": 7400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7726150909868705e-05,
+      "loss": 1.9523,
+      "step": 7410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7722860245483566e-05,
+      "loss": 2.2473,
+      "step": 7420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.771956958109843e-05,
+      "loss": 1.8236,
+      "step": 7430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.771627891671329e-05,
+      "loss": 1.8567,
+      "step": 7440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.771298825232814e-05,
+      "loss": 1.8352,
+      "step": 7450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.770969758794301e-05,
+      "loss": 2.047,
+      "step": 7460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7706406923557865e-05,
+      "loss": 1.8747,
+      "step": 7470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7703116259172733e-05,
+      "loss": 1.6061,
+      "step": 7480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.769982559478759e-05,
+      "loss": 1.8715,
+      "step": 7490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.769653493040245e-05,
+      "loss": 1.8886,
+      "step": 7500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.769324426601731e-05,
+      "loss": 1.7883,
+      "step": 7510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.768995360163217e-05,
+      "loss": 1.8673,
+      "step": 7520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.768666293724703e-05,
+      "loss": 1.9719,
+      "step": 7530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7683372272861894e-05,
+      "loss": 1.7408,
+      "step": 7540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.768008160847675e-05,
+      "loss": 1.958,
+      "step": 7550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.767679094409162e-05,
+      "loss": 1.8567,
+      "step": 7560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.767350027970647e-05,
+      "loss": 1.8018,
+      "step": 7570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.767020961532134e-05,
+      "loss": 1.8316,
+      "step": 7580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7666918950936194e-05,
+      "loss": 1.7123,
+      "step": 7590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7663628286551055e-05,
+      "loss": 1.8418,
+      "step": 7600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7660337622165916e-05,
+      "loss": 1.804,
+      "step": 7610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.765704695778078e-05,
+      "loss": 1.8465,
+      "step": 7620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.765375629339564e-05,
+      "loss": 2.0641,
+      "step": 7630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.76504656290105e-05,
+      "loss": 1.6089,
+      "step": 7640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.764717496462536e-05,
+      "loss": 1.7567,
+      "step": 7650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.764388430024022e-05,
+      "loss": 1.7424,
+      "step": 7660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.764059363585508e-05,
+      "loss": 1.6721,
+      "step": 7670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7637302971469945e-05,
+      "loss": 1.7602,
+      "step": 7680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.76340123070848e-05,
+      "loss": 1.8824,
+      "step": 7690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.763072164269967e-05,
+      "loss": 1.8396,
+      "step": 7700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762743097831452e-05,
+      "loss": 1.7236,
+      "step": 7710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7624140313929383e-05,
+      "loss": 2.1056,
+      "step": 7720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7620849649544245e-05,
+      "loss": 1.8465,
+      "step": 7730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7617558985159106e-05,
+      "loss": 1.8217,
+      "step": 7740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761426832077397e-05,
+      "loss": 2.0706,
+      "step": 7750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761097765638883e-05,
+      "loss": 1.7653,
+      "step": 7760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760768699200368e-05,
+      "loss": 1.7907,
+      "step": 7770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760439632761855e-05,
+      "loss": 1.8046,
+      "step": 7780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7601105663233406e-05,
+      "loss": 1.8971,
+      "step": 7790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7597814998848274e-05,
+      "loss": 1.8664,
+      "step": 7800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759452433446313e-05,
+      "loss": 1.7311,
+      "step": 7810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759123367007799e-05,
+      "loss": 1.8847,
+      "step": 7820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758794300569285e-05,
+      "loss": 1.7249,
+      "step": 7830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758465234130771e-05,
+      "loss": 1.9519,
+      "step": 7840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758136167692257e-05,
+      "loss": 1.6159,
+      "step": 7850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7578071012537434e-05,
+      "loss": 1.7724,
+      "step": 7860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7574780348152296e-05,
+      "loss": 2.0237,
+      "step": 7870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.757148968376716e-05,
+      "loss": 1.8322,
+      "step": 7880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.756819901938201e-05,
+      "loss": 1.7482,
+      "step": 7890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.756490835499688e-05,
+      "loss": 1.7422,
+      "step": 7900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7561617690611734e-05,
+      "loss": 1.7547,
+      "step": 7910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7558327026226595e-05,
+      "loss": 1.9822,
+      "step": 7920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7555036361841457e-05,
+      "loss": 1.8327,
+      "step": 7930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.755174569745632e-05,
+      "loss": 1.8708,
+      "step": 7940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.754845503307118e-05,
+      "loss": 1.6975,
+      "step": 7950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.754516436868604e-05,
+      "loss": 1.6989,
+      "step": 7960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.75418737043009e-05,
+      "loss": 1.7392,
+      "step": 7970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.753858303991576e-05,
+      "loss": 2.0159,
+      "step": 7980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7535292375530624e-05,
+      "loss": 1.7091,
+      "step": 7990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7532001711145485e-05,
+      "loss": 1.9383,
+      "step": 8000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.752871104676035e-05,
+      "loss": 1.7918,
+      "step": 8010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.75254203823752e-05,
+      "loss": 1.8564,
+      "step": 8020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.752212971799007e-05,
+      "loss": 1.6672,
+      "step": 8030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7518839053604924e-05,
+      "loss": 1.8153,
+      "step": 8040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7515548389219785e-05,
+      "loss": 1.6993,
+      "step": 8050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7512257724834646e-05,
+      "loss": 2.0067,
+      "step": 8060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.750896706044951e-05,
+      "loss": 1.7911,
+      "step": 8070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.750567639606437e-05,
+      "loss": 1.7487,
+      "step": 8080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.750238573167923e-05,
+      "loss": 1.8886,
+      "step": 8090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.749909506729409e-05,
+      "loss": 1.8654,
+      "step": 8100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.749580440290895e-05,
+      "loss": 1.7661,
+      "step": 8110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.749251373852381e-05,
+      "loss": 1.7422,
+      "step": 8120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7489223074138675e-05,
+      "loss": 1.8239,
+      "step": 8130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.748593240975353e-05,
+      "loss": 1.9077,
+      "step": 8140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.74826417453684e-05,
+      "loss": 1.7574,
+      "step": 8150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.747935108098325e-05,
+      "loss": 1.7144,
+      "step": 8160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7476060416598113e-05,
+      "loss": 1.834,
+      "step": 8170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7472769752212975e-05,
+      "loss": 1.9514,
+      "step": 8180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7469479087827836e-05,
+      "loss": 1.7553,
+      "step": 8190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.74661884234427e-05,
+      "loss": 1.8665,
+      "step": 8200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.746289775905756e-05,
+      "loss": 1.8488,
+      "step": 8210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.745960709467241e-05,
+      "loss": 1.8358,
+      "step": 8220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.745631643028728e-05,
+      "loss": 1.7887,
+      "step": 8230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7453025765902136e-05,
+      "loss": 1.8582,
+      "step": 8240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7449735101517004e-05,
+      "loss": 1.9571,
+      "step": 8250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744644443713186e-05,
+      "loss": 1.8567,
+      "step": 8260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744315377274672e-05,
+      "loss": 1.7771,
+      "step": 8270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743986310836158e-05,
+      "loss": 1.8647,
+      "step": 8280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743657244397644e-05,
+      "loss": 1.812,
+      "step": 8290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.74332817795913e-05,
+      "loss": 1.6525,
+      "step": 8300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7429991115206164e-05,
+      "loss": 1.9195,
+      "step": 8310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742670045082102e-05,
+      "loss": 1.856,
+      "step": 8320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742340978643589e-05,
+      "loss": 1.7583,
+      "step": 8330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742011912205074e-05,
+      "loss": 2.0455,
+      "step": 8340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741682845766561e-05,
+      "loss": 1.8225,
+      "step": 8350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7413537793280464e-05,
+      "loss": 1.7568,
+      "step": 8360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7410247128895325e-05,
+      "loss": 1.9492,
+      "step": 8370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7406956464510186e-05,
+      "loss": 1.6763,
+      "step": 8380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.740366580012505e-05,
+      "loss": 1.8149,
+      "step": 8390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.740037513573991e-05,
+      "loss": 1.7876,
+      "step": 8400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.739708447135477e-05,
+      "loss": 1.7732,
+      "step": 8410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7393793806969625e-05,
+      "loss": 1.7815,
+      "step": 8420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.739050314258449e-05,
+      "loss": 2.0086,
+      "step": 8430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.738721247819935e-05,
+      "loss": 1.8783,
+      "step": 8440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7383921813814215e-05,
+      "loss": 1.768,
+      "step": 8450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.738063114942907e-05,
+      "loss": 1.6803,
+      "step": 8460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.737734048504393e-05,
+      "loss": 2.0804,
+      "step": 8470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.737404982065879e-05,
+      "loss": 1.686,
+      "step": 8480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7370759156273654e-05,
+      "loss": 2.0184,
+      "step": 8490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7367468491888515e-05,
+      "loss": 1.7979,
+      "step": 8500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7364177827503376e-05,
+      "loss": 1.6438,
+      "step": 8510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.736088716311823e-05,
+      "loss": 1.7079,
+      "step": 8520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.73575964987331e-05,
+      "loss": 2.4117,
+      "step": 8530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735430583434795e-05,
+      "loss": 1.7669,
+      "step": 8540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735101516996282e-05,
+      "loss": 1.7595,
+      "step": 8550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7347724505577676e-05,
+      "loss": 1.8028,
+      "step": 8560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.734443384119254e-05,
+      "loss": 1.7342,
+      "step": 8570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.73411431768074e-05,
+      "loss": 1.774,
+      "step": 8580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733785251242226e-05,
+      "loss": 1.6601,
+      "step": 8590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733456184803712e-05,
+      "loss": 1.8803,
+      "step": 8600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733127118365198e-05,
+      "loss": 1.9165,
+      "step": 8610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732798051926684e-05,
+      "loss": 1.7243,
+      "step": 8620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7324689854881705e-05,
+      "loss": 1.7007,
+      "step": 8630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7321399190496566e-05,
+      "loss": 1.8535,
+      "step": 8640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731810852611143e-05,
+      "loss": 2.0146,
+      "step": 8650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731481786172628e-05,
+      "loss": 1.8127,
+      "step": 8660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731152719734114e-05,
+      "loss": 1.836,
+      "step": 8670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7308236532956004e-05,
+      "loss": 1.804,
+      "step": 8680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7304945868570865e-05,
+      "loss": 1.8417,
+      "step": 8690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.730165520418573e-05,
+      "loss": 1.8828,
+      "step": 8700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729836453980059e-05,
+      "loss": 1.8972,
+      "step": 8710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729507387541545e-05,
+      "loss": 1.9066,
+      "step": 8720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729178321103031e-05,
+      "loss": 1.8735,
+      "step": 8730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.728849254664517e-05,
+      "loss": 1.8524,
+      "step": 8740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.728520188226003e-05,
+      "loss": 1.9365,
+      "step": 8750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7281911217874894e-05,
+      "loss": 1.889,
+      "step": 8760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727862055348975e-05,
+      "loss": 1.841,
+      "step": 8770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727532988910462e-05,
+      "loss": 1.9949,
+      "step": 8780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727203922471947e-05,
+      "loss": 1.7898,
+      "step": 8790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726874856033434e-05,
+      "loss": 1.6685,
+      "step": 8800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7265457895949194e-05,
+      "loss": 1.6734,
+      "step": 8810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7262167231564055e-05,
+      "loss": 1.8038,
+      "step": 8820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7258876567178916e-05,
+      "loss": 1.9342,
+      "step": 8830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725558590279378e-05,
+      "loss": 1.7444,
+      "step": 8840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725229523840864e-05,
+      "loss": 1.9162,
+      "step": 8850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72490045740235e-05,
+      "loss": 1.7915,
+      "step": 8860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7245713909638355e-05,
+      "loss": 1.9339,
+      "step": 8870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724242324525322e-05,
+      "loss": 1.8936,
+      "step": 8880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723913258086808e-05,
+      "loss": 1.6661,
+      "step": 8890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7235841916482945e-05,
+      "loss": 1.7564,
+      "step": 8900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.72325512520978e-05,
+      "loss": 1.7413,
+      "step": 8910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.722926058771266e-05,
+      "loss": 1.7147,
+      "step": 8920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.722596992332752e-05,
+      "loss": 1.793,
+      "step": 8930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7222679258942384e-05,
+      "loss": 1.887,
+      "step": 8940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7219388594557245e-05,
+      "loss": 1.9039,
+      "step": 8950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7216097930172106e-05,
+      "loss": 2.1685,
+      "step": 8960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.721280726578696e-05,
+      "loss": 2.1491,
+      "step": 8970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720951660140183e-05,
+      "loss": 1.8215,
+      "step": 8980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720622593701668e-05,
+      "loss": 1.8504,
+      "step": 8990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720293527263155e-05,
+      "loss": 1.8468,
+      "step": 9000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7199644608246406e-05,
+      "loss": 1.855,
+      "step": 9010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.719635394386127e-05,
+      "loss": 1.7153,
+      "step": 9020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.719306327947613e-05,
+      "loss": 1.6915,
+      "step": 9030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718977261509099e-05,
+      "loss": 1.7073,
+      "step": 9040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718648195070585e-05,
+      "loss": 1.7851,
+      "step": 9050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718319128632071e-05,
+      "loss": 1.7658,
+      "step": 9060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7179900621935567e-05,
+      "loss": 1.8059,
+      "step": 9070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7176609957550435e-05,
+      "loss": 2.0062,
+      "step": 9080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717331929316529e-05,
+      "loss": 1.7161,
+      "step": 9090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717002862878016e-05,
+      "loss": 1.7906,
+      "step": 9100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.716673796439501e-05,
+      "loss": 1.6509,
+      "step": 9110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.716344730000987e-05,
+      "loss": 1.7417,
+      "step": 9120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7160156635624734e-05,
+      "loss": 1.7295,
+      "step": 9130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7156865971239595e-05,
+      "loss": 1.9739,
+      "step": 9140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715357530685446e-05,
+      "loss": 1.8044,
+      "step": 9150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715028464246932e-05,
+      "loss": 1.8553,
+      "step": 9160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714699397808417e-05,
+      "loss": 1.7357,
+      "step": 9170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714370331369904e-05,
+      "loss": 1.7556,
+      "step": 9180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7140412649313895e-05,
+      "loss": 1.7741,
+      "step": 9190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713712198492876e-05,
+      "loss": 1.8549,
+      "step": 9200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713383132054362e-05,
+      "loss": 1.7562,
+      "step": 9210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713054065615848e-05,
+      "loss": 1.7104,
+      "step": 9220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712724999177334e-05,
+      "loss": 1.8899,
+      "step": 9230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71239593273882e-05,
+      "loss": 1.7647,
+      "step": 9240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712066866300306e-05,
+      "loss": 1.7662,
+      "step": 9250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7117377998617924e-05,
+      "loss": 1.614,
+      "step": 9260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711408733423278e-05,
+      "loss": 1.935,
+      "step": 9270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7110796669847646e-05,
+      "loss": 1.8535,
+      "step": 9280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71075060054625e-05,
+      "loss": 1.7098,
+      "step": 9290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710421534107737e-05,
+      "loss": 1.8076,
+      "step": 9300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710092467669222e-05,
+      "loss": 1.8302,
+      "step": 9310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7097634012307085e-05,
+      "loss": 1.8949,
+      "step": 9320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7094343347921946e-05,
+      "loss": 1.7122,
+      "step": 9330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709105268353681e-05,
+      "loss": 1.7469,
+      "step": 9340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708776201915167e-05,
+      "loss": 1.5696,
+      "step": 9350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708447135476653e-05,
+      "loss": 1.8221,
+      "step": 9360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708118069038139e-05,
+      "loss": 1.8399,
+      "step": 9370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707789002599625e-05,
+      "loss": 1.7465,
+      "step": 9380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7074599361611114e-05,
+      "loss": 1.7966,
+      "step": 9390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7071308697225975e-05,
+      "loss": 1.8078,
+      "step": 9400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7068018032840836e-05,
+      "loss": 1.7039,
+      "step": 9410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.706472736845569e-05,
+      "loss": 1.7467,
+      "step": 9420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.706143670407055e-05,
+      "loss": 1.7301,
+      "step": 9430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.705814603968541e-05,
+      "loss": 1.817,
+      "step": 9440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7054855375300274e-05,
+      "loss": 1.7036,
+      "step": 9450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7051564710915136e-05,
+      "loss": 1.6968,
+      "step": 9460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.704827404653e-05,
+      "loss": 1.5885,
+      "step": 9470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.704498338214486e-05,
+      "loss": 1.8784,
+      "step": 9480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.704169271775972e-05,
+      "loss": 1.9172,
+      "step": 9490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.703840205337458e-05,
+      "loss": 2.0382,
+      "step": 9500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.703511138898944e-05,
+      "loss": 1.9989,
+      "step": 9510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7031820724604296e-05,
+      "loss": 1.7131,
+      "step": 9520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7028530060219164e-05,
+      "loss": 1.8489,
+      "step": 9530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.702523939583402e-05,
+      "loss": 1.801,
+      "step": 9540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.702194873144889e-05,
+      "loss": 1.7497,
+      "step": 9550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.701865806706374e-05,
+      "loss": 1.935,
+      "step": 9560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.70153674026786e-05,
+      "loss": 1.6329,
+      "step": 9570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7012076738293464e-05,
+      "loss": 1.8627,
+      "step": 9580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7008786073908325e-05,
+      "loss": 1.7446,
+      "step": 9590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7005495409523187e-05,
+      "loss": 1.7631,
+      "step": 9600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.700220474513805e-05,
+      "loss": 1.8488,
+      "step": 9610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69989140807529e-05,
+      "loss": 1.9299,
+      "step": 9620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.699562341636777e-05,
+      "loss": 1.7187,
+      "step": 9630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6992332751982625e-05,
+      "loss": 1.9328,
+      "step": 9640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.698904208759749e-05,
+      "loss": 1.8464,
+      "step": 9650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.698575142321235e-05,
+      "loss": 1.7808,
+      "step": 9660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.698246075882721e-05,
+      "loss": 1.7972,
+      "step": 9670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.697917009444207e-05,
+      "loss": 1.8169,
+      "step": 9680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.697587943005693e-05,
+      "loss": 1.7271,
+      "step": 9690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.697258876567179e-05,
+      "loss": 1.7093,
+      "step": 9700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6969298101286654e-05,
+      "loss": 1.8551,
+      "step": 9710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.696600743690151e-05,
+      "loss": 1.8543,
+      "step": 9720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6962716772516376e-05,
+      "loss": 1.7714,
+      "step": 9730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.695942610813123e-05,
+      "loss": 1.7111,
+      "step": 9740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69561354437461e-05,
+      "loss": 1.8887,
+      "step": 9750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.695284477936095e-05,
+      "loss": 1.7162,
+      "step": 9760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6949554114975815e-05,
+      "loss": 1.8941,
+      "step": 9770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6946263450590676e-05,
+      "loss": 1.8642,
+      "step": 9780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694297278620554e-05,
+      "loss": 1.628,
+      "step": 9790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69396821218204e-05,
+      "loss": 1.6879,
+      "step": 9800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693639145743526e-05,
+      "loss": 1.6392,
+      "step": 9810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6933100793050114e-05,
+      "loss": 1.8194,
+      "step": 9820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692981012866498e-05,
+      "loss": 1.8217,
+      "step": 9830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692651946427984e-05,
+      "loss": 1.831,
+      "step": 9840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6923228799894705e-05,
+      "loss": 1.6803,
+      "step": 9850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691993813550956e-05,
+      "loss": 2.0363,
+      "step": 9860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691664747112442e-05,
+      "loss": 2.0016,
+      "step": 9870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691335680673928e-05,
+      "loss": 1.9232,
+      "step": 9880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691006614235414e-05,
+      "loss": 1.756,
+      "step": 9890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6906775477969004e-05,
+      "loss": 1.8758,
+      "step": 9900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6903484813583866e-05,
+      "loss": 1.9233,
+      "step": 9910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.690019414919872e-05,
+      "loss": 1.8498,
+      "step": 9920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.689690348481359e-05,
+      "loss": 1.7137,
+      "step": 9930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.689361282042844e-05,
+      "loss": 1.7067,
+      "step": 9940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.689032215604331e-05,
+      "loss": 1.748,
+      "step": 9950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6887031491658165e-05,
+      "loss": 1.7756,
+      "step": 9960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6883740827273026e-05,
+      "loss": 1.7903,
+      "step": 9970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.688045016288789e-05,
+      "loss": 1.7744,
+      "step": 9980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.687715949850275e-05,
+      "loss": 2.0327,
+      "step": 9990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.687386883411761e-05,
+      "loss": 1.717,
+      "step": 10000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.687057816973247e-05,
+      "loss": 1.9016,
+      "step": 10010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6867287505347326e-05,
+      "loss": 1.8011,
+      "step": 10020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6863996840962194e-05,
+      "loss": 1.9135,
+      "step": 10030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.686070617657705e-05,
+      "loss": 1.8,
+      "step": 10040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6857415512191917e-05,
+      "loss": 1.8567,
+      "step": 10050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.685412484780677e-05,
+      "loss": 1.8113,
+      "step": 10060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.685083418342163e-05,
+      "loss": 1.8133,
+      "step": 10070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6847543519036494e-05,
+      "loss": 1.7594,
+      "step": 10080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6844252854651355e-05,
+      "loss": 1.7667,
+      "step": 10090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6840962190266216e-05,
+      "loss": 1.8991,
+      "step": 10100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.683767152588108e-05,
+      "loss": 1.6456,
+      "step": 10110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.683438086149594e-05,
+      "loss": 1.6984,
+      "step": 10120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.68310901971108e-05,
+      "loss": 1.8991,
+      "step": 10130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.682779953272566e-05,
+      "loss": 1.8076,
+      "step": 10140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.682450886834052e-05,
+      "loss": 1.7209,
+      "step": 10150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6821218203955384e-05,
+      "loss": 2.2316,
+      "step": 10160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.681792753957024e-05,
+      "loss": 1.9247,
+      "step": 10170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.68146368751851e-05,
+      "loss": 1.8898,
+      "step": 10180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.681134621079996e-05,
+      "loss": 1.8407,
+      "step": 10190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.680805554641482e-05,
+      "loss": 1.8155,
+      "step": 10200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.680476488202968e-05,
+      "loss": 1.9241,
+      "step": 10210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6801474217644545e-05,
+      "loss": 1.8893,
+      "step": 10220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6798183553259406e-05,
+      "loss": 1.8554,
+      "step": 10230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.679489288887427e-05,
+      "loss": 1.935,
+      "step": 10240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.679160222448913e-05,
+      "loss": 1.7223,
+      "step": 10250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.678831156010399e-05,
+      "loss": 1.7825,
+      "step": 10260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6785020895718844e-05,
+      "loss": 1.6279,
+      "step": 10270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.678173023133371e-05,
+      "loss": 2.1041,
+      "step": 10280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6778439566948567e-05,
+      "loss": 1.7777,
+      "step": 10290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6775148902563435e-05,
+      "loss": 1.8113,
+      "step": 10300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677185823817829e-05,
+      "loss": 1.7071,
+      "step": 10310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676856757379315e-05,
+      "loss": 1.6782,
+      "step": 10320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676527690940801e-05,
+      "loss": 1.9358,
+      "step": 10330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676198624502287e-05,
+      "loss": 1.704,
+      "step": 10340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6758695580637734e-05,
+      "loss": 1.7695,
+      "step": 10350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6755404916252595e-05,
+      "loss": 1.8229,
+      "step": 10360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675211425186745e-05,
+      "loss": 1.8898,
+      "step": 10370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674882358748232e-05,
+      "loss": 1.8489,
+      "step": 10380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674553292309717e-05,
+      "loss": 1.7447,
+      "step": 10390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674224225871204e-05,
+      "loss": 1.8766,
+      "step": 10400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6738951594326895e-05,
+      "loss": 2.0936,
+      "step": 10410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6735660929941756e-05,
+      "loss": 1.6629,
+      "step": 10420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.673237026555662e-05,
+      "loss": 1.7132,
+      "step": 10430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.672907960117148e-05,
+      "loss": 2.2445,
+      "step": 10440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.672578893678634e-05,
+      "loss": 1.6919,
+      "step": 10450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.67224982724012e-05,
+      "loss": 2.3169,
+      "step": 10460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.671920760801606e-05,
+      "loss": 1.8412,
+      "step": 10470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6715916943630924e-05,
+      "loss": 1.8418,
+      "step": 10480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.671262627924578e-05,
+      "loss": 1.7044,
+      "step": 10490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6709335614860646e-05,
+      "loss": 1.7673,
+      "step": 10500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.67060449504755e-05,
+      "loss": 1.6812,
+      "step": 10510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.670275428609037e-05,
+      "loss": 1.7077,
+      "step": 10520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6699463621705223e-05,
+      "loss": 1.8261,
+      "step": 10530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6696172957320085e-05,
+      "loss": 2.1709,
+      "step": 10540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6692882292934946e-05,
+      "loss": 1.8813,
+      "step": 10550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.668959162854981e-05,
+      "loss": 1.7755,
+      "step": 10560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.668630096416467e-05,
+      "loss": 1.7585,
+      "step": 10570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.668301029977953e-05,
+      "loss": 1.6301,
+      "step": 10580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6679719635394384e-05,
+      "loss": 1.8408,
+      "step": 10590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.667642897100925e-05,
+      "loss": 1.8973,
+      "step": 10600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.667313830662411e-05,
+      "loss": 1.8126,
+      "step": 10610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6669847642238975e-05,
+      "loss": 2.2898,
+      "step": 10620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.666655697785383e-05,
+      "loss": 1.8336,
+      "step": 10630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.666326631346869e-05,
+      "loss": 1.7009,
+      "step": 10640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.665997564908355e-05,
+      "loss": 1.7655,
+      "step": 10650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.665668498469841e-05,
+      "loss": 1.7368,
+      "step": 10660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6653394320313274e-05,
+      "loss": 1.6739,
+      "step": 10670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6650103655928136e-05,
+      "loss": 1.9038,
+      "step": 10680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.664681299154299e-05,
+      "loss": 1.8275,
+      "step": 10690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.664352232715786e-05,
+      "loss": 1.8762,
+      "step": 10700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.664023166277271e-05,
+      "loss": 1.836,
+      "step": 10710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.663694099838758e-05,
+      "loss": 1.6825,
+      "step": 10720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6633650334002435e-05,
+      "loss": 1.8987,
+      "step": 10730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6630359669617297e-05,
+      "loss": 1.8138,
+      "step": 10740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662706900523216e-05,
+      "loss": 1.8345,
+      "step": 10750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662377834084702e-05,
+      "loss": 1.9643,
+      "step": 10760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662048767646188e-05,
+      "loss": 1.8312,
+      "step": 10770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661719701207674e-05,
+      "loss": 1.928,
+      "step": 10780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6613906347691596e-05,
+      "loss": 1.6919,
+      "step": 10790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6610615683306464e-05,
+      "loss": 1.7779,
+      "step": 10800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660732501892132e-05,
+      "loss": 1.6432,
+      "step": 10810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660403435453619e-05,
+      "loss": 1.9299,
+      "step": 10820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660074369015104e-05,
+      "loss": 1.8386,
+      "step": 10830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65974530257659e-05,
+      "loss": 1.6492,
+      "step": 10840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6594162361380764e-05,
+      "loss": 1.8553,
+      "step": 10850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6590871696995625e-05,
+      "loss": 1.688,
+      "step": 10860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6587581032610486e-05,
+      "loss": 1.8515,
+      "step": 10870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658429036822535e-05,
+      "loss": 1.7706,
+      "step": 10880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658099970384021e-05,
+      "loss": 2.0089,
+      "step": 10890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657770903945507e-05,
+      "loss": 1.8401,
+      "step": 10900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657441837506993e-05,
+      "loss": 1.8045,
+      "step": 10910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657112771068479e-05,
+      "loss": 1.6818,
+      "step": 10920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6567837046299654e-05,
+      "loss": 1.8094,
+      "step": 10930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.656454638191451e-05,
+      "loss": 1.798,
+      "step": 10940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.656125571752937e-05,
+      "loss": 1.7641,
+      "step": 10950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.655796505314423e-05,
+      "loss": 2.0401,
+      "step": 10960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.655467438875909e-05,
+      "loss": 1.6607,
+      "step": 10970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6551383724373953e-05,
+      "loss": 1.7634,
+      "step": 10980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6548093059988815e-05,
+      "loss": 1.7936,
+      "step": 10990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6544802395603676e-05,
+      "loss": 1.775,
+      "step": 11000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.654151173121854e-05,
+      "loss": 1.6354,
+      "step": 11010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.65382210668334e-05,
+      "loss": 1.7337,
+      "step": 11020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.653493040244826e-05,
+      "loss": 1.7611,
+      "step": 11030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6531639738063114e-05,
+      "loss": 1.736,
+      "step": 11040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.652834907367798e-05,
+      "loss": 1.9145,
+      "step": 11050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.652505840929284e-05,
+      "loss": 1.6825,
+      "step": 11060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6521767744907705e-05,
+      "loss": 1.6935,
+      "step": 11070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.651847708052256e-05,
+      "loss": 1.6431,
+      "step": 11080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.651518641613742e-05,
+      "loss": 1.7577,
+      "step": 11090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.651189575175228e-05,
+      "loss": 1.7833,
+      "step": 11100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.650860508736714e-05,
+      "loss": 1.7747,
+      "step": 11110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6505314422982004e-05,
+      "loss": 1.7178,
+      "step": 11120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6502023758596866e-05,
+      "loss": 1.9283,
+      "step": 11130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.649873309421172e-05,
+      "loss": 1.9038,
+      "step": 11140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.649544242982659e-05,
+      "loss": 1.6723,
+      "step": 11150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.649215176544144e-05,
+      "loss": 1.7699,
+      "step": 11160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.648886110105631e-05,
+      "loss": 1.8297,
+      "step": 11170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6485570436671165e-05,
+      "loss": 1.622,
+      "step": 11180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6482279772286026e-05,
+      "loss": 2.0829,
+      "step": 11190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.647898910790089e-05,
+      "loss": 1.7702,
+      "step": 11200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.647569844351575e-05,
+      "loss": 2.0824,
+      "step": 11210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.647240777913061e-05,
+      "loss": 1.9437,
+      "step": 11220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.646911711474547e-05,
+      "loss": 1.7226,
+      "step": 11230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6465826450360326e-05,
+      "loss": 1.735,
+      "step": 11240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6462535785975194e-05,
+      "loss": 1.6354,
+      "step": 11250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.645924512159005e-05,
+      "loss": 1.7965,
+      "step": 11260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6455954457204917e-05,
+      "loss": 1.8698,
+      "step": 11270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.645266379281977e-05,
+      "loss": 1.6727,
+      "step": 11280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644937312843463e-05,
+      "loss": 1.6657,
+      "step": 11290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6446082464049494e-05,
+      "loss": 1.6979,
+      "step": 11300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6442791799664355e-05,
+      "loss": 1.9158,
+      "step": 11310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6439501135279216e-05,
+      "loss": 1.8366,
+      "step": 11320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643621047089408e-05,
+      "loss": 1.8306,
+      "step": 11330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643291980650893e-05,
+      "loss": 1.7263,
+      "step": 11340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64296291421238e-05,
+      "loss": 1.9166,
+      "step": 11350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6426338477738654e-05,
+      "loss": 1.9081,
+      "step": 11360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642304781335352e-05,
+      "loss": 1.8741,
+      "step": 11370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641975714896838e-05,
+      "loss": 1.805,
+      "step": 11380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641646648458324e-05,
+      "loss": 1.9144,
+      "step": 11390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64131758201981e-05,
+      "loss": 1.6914,
+      "step": 11400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640988515581296e-05,
+      "loss": 1.7936,
+      "step": 11410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640659449142782e-05,
+      "loss": 1.8632,
+      "step": 11420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640330382704268e-05,
+      "loss": 1.6257,
+      "step": 11430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.640001316265754e-05,
+      "loss": 1.6155,
+      "step": 11440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6396722498272406e-05,
+      "loss": 1.822,
+      "step": 11450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.639343183388726e-05,
+      "loss": 2.084,
+      "step": 11460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.639014116950213e-05,
+      "loss": 1.6998,
+      "step": 11470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.638685050511698e-05,
+      "loss": 1.9133,
+      "step": 11480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6383559840731844e-05,
+      "loss": 1.6798,
+      "step": 11490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6380269176346705e-05,
+      "loss": 1.8388,
+      "step": 11500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.637697851196157e-05,
+      "loss": 1.7968,
+      "step": 11510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.637368784757643e-05,
+      "loss": 1.7986,
+      "step": 11520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.637039718319129e-05,
+      "loss": 1.813,
+      "step": 11530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6367106518806144e-05,
+      "loss": 1.7862,
+      "step": 11540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.636381585442101e-05,
+      "loss": 1.6579,
+      "step": 11550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6360525190035866e-05,
+      "loss": 1.7095,
+      "step": 11560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6357234525650734e-05,
+      "loss": 1.5637,
+      "step": 11570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.635394386126559e-05,
+      "loss": 1.7768,
+      "step": 11580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.635065319688045e-05,
+      "loss": 1.9121,
+      "step": 11590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.634736253249531e-05,
+      "loss": 1.8035,
+      "step": 11600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.634407186811017e-05,
+      "loss": 1.6596,
+      "step": 11610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6340781203725034e-05,
+      "loss": 1.9281,
+      "step": 11620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6337490539339895e-05,
+      "loss": 1.9708,
+      "step": 11630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6334199874954756e-05,
+      "loss": 1.8437,
+      "step": 11640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.633090921056962e-05,
+      "loss": 1.8061,
+      "step": 11650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.632761854618448e-05,
+      "loss": 1.7009,
+      "step": 11660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.632432788179934e-05,
+      "loss": 1.8172,
+      "step": 11670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.63210372174142e-05,
+      "loss": 2.0243,
+      "step": 11680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6317746553029056e-05,
+      "loss": 1.6624,
+      "step": 11690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.631445588864392e-05,
+      "loss": 1.8269,
+      "step": 11700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.631116522425878e-05,
+      "loss": 1.8258,
+      "step": 11710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.630787455987364e-05,
+      "loss": 1.8001,
+      "step": 11720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.63045838954885e-05,
+      "loss": 1.7584,
+      "step": 11730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.630129323110336e-05,
+      "loss": 1.6483,
+      "step": 11740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6298002566718224e-05,
+      "loss": 1.93,
+      "step": 11750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6294711902333085e-05,
+      "loss": 1.8346,
+      "step": 11760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6291421237947946e-05,
+      "loss": 1.8476,
+      "step": 11770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.628813057356281e-05,
+      "loss": 1.9114,
+      "step": 11780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.628483990917766e-05,
+      "loss": 1.8295,
+      "step": 11790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.628154924479253e-05,
+      "loss": 1.7608,
+      "step": 11800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6278258580407384e-05,
+      "loss": 1.7193,
+      "step": 11810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627496791602225e-05,
+      "loss": 1.6766,
+      "step": 11820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627167725163711e-05,
+      "loss": 1.9021,
+      "step": 11830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626838658725197e-05,
+      "loss": 2.006,
+      "step": 11840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626509592286683e-05,
+      "loss": 1.647,
+      "step": 11850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626180525848169e-05,
+      "loss": 1.6524,
+      "step": 11860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625851459409655e-05,
+      "loss": 1.828,
+      "step": 11870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625522392971141e-05,
+      "loss": 1.8803,
+      "step": 11880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625193326532627e-05,
+      "loss": 1.8966,
+      "step": 11890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6248642600941136e-05,
+      "loss": 1.6213,
+      "step": 11900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624535193655599e-05,
+      "loss": 1.6331,
+      "step": 11910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624206127217086e-05,
+      "loss": 1.8807,
+      "step": 11920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623877060778571e-05,
+      "loss": 1.6708,
+      "step": 11930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6235479943400574e-05,
+      "loss": 1.9688,
+      "step": 11940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6232189279015435e-05,
+      "loss": 1.6594,
+      "step": 11950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.62288986146303e-05,
+      "loss": 1.5874,
+      "step": 11960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.622560795024516e-05,
+      "loss": 1.9536,
+      "step": 11970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.622231728586002e-05,
+      "loss": 1.6596,
+      "step": 11980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6219026621474874e-05,
+      "loss": 1.7801,
+      "step": 11990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.621573595708974e-05,
+      "loss": 1.9632,
+      "step": 12000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6212445292704596e-05,
+      "loss": 1.6985,
+      "step": 12010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6209154628319464e-05,
+      "loss": 1.802,
+      "step": 12020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.620586396393432e-05,
+      "loss": 1.6967,
+      "step": 12030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.620257329954918e-05,
+      "loss": 1.7939,
+      "step": 12040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.619928263516404e-05,
+      "loss": 1.8073,
+      "step": 12050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61959919707789e-05,
+      "loss": 1.7454,
+      "step": 12060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6192701306393764e-05,
+      "loss": 1.8365,
+      "step": 12070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6189410642008625e-05,
+      "loss": 1.8226,
+      "step": 12080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.618611997762348e-05,
+      "loss": 1.7738,
+      "step": 12090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.618282931323835e-05,
+      "loss": 1.644,
+      "step": 12100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61795386488532e-05,
+      "loss": 1.7685,
+      "step": 12110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.617624798446807e-05,
+      "loss": 1.8296,
+      "step": 12120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6172957320082925e-05,
+      "loss": 1.7617,
+      "step": 12130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6169666655697786e-05,
+      "loss": 1.7629,
+      "step": 12140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.616637599131265e-05,
+      "loss": 1.6963,
+      "step": 12150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.616308532692751e-05,
+      "loss": 1.8874,
+      "step": 12160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.615979466254237e-05,
+      "loss": 1.6572,
+      "step": 12170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.615650399815723e-05,
+      "loss": 1.8243,
+      "step": 12180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6153213333772085e-05,
+      "loss": 2.0297,
+      "step": 12190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6149922669386953e-05,
+      "loss": 1.836,
+      "step": 12200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.614663200500181e-05,
+      "loss": 1.6797,
+      "step": 12210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6143341340616676e-05,
+      "loss": 1.8823,
+      "step": 12220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.614005067623153e-05,
+      "loss": 1.6692,
+      "step": 12230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.613676001184639e-05,
+      "loss": 1.7384,
+      "step": 12240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.613346934746125e-05,
+      "loss": 1.7681,
+      "step": 12250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6130178683076114e-05,
+      "loss": 1.7589,
+      "step": 12260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6126888018690976e-05,
+      "loss": 2.0858,
+      "step": 12270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.612359735430584e-05,
+      "loss": 1.7864,
+      "step": 12280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61203066899207e-05,
+      "loss": 1.7951,
+      "step": 12290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611701602553556e-05,
+      "loss": 1.5966,
+      "step": 12300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6113725361150414e-05,
+      "loss": 1.8252,
+      "step": 12310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611043469676528e-05,
+      "loss": 1.8482,
+      "step": 12320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6107144032380136e-05,
+      "loss": 1.71,
+      "step": 12330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6103853367995e-05,
+      "loss": 1.833,
+      "step": 12340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610056270360986e-05,
+      "loss": 1.9119,
+      "step": 12350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609727203922472e-05,
+      "loss": 1.6478,
+      "step": 12360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609398137483958e-05,
+      "loss": 1.7087,
+      "step": 12370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609069071045444e-05,
+      "loss": 1.8644,
+      "step": 12380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6087400046069304e-05,
+      "loss": 1.6668,
+      "step": 12390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6084109381684165e-05,
+      "loss": 1.6633,
+      "step": 12400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6080818717299027e-05,
+      "loss": 1.7857,
+      "step": 12410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607752805291389e-05,
+      "loss": 1.8384,
+      "step": 12420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607423738852875e-05,
+      "loss": 1.8356,
+      "step": 12430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6070946724143604e-05,
+      "loss": 1.6629,
+      "step": 12440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.606765605975847e-05,
+      "loss": 1.7057,
+      "step": 12450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6064365395373326e-05,
+      "loss": 1.7922,
+      "step": 12460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.606107473098819e-05,
+      "loss": 1.8729,
+      "step": 12470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.605778406660305e-05,
+      "loss": 1.7882,
+      "step": 12480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.605449340221791e-05,
+      "loss": 1.6194,
+      "step": 12490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.605120273783277e-05,
+      "loss": 1.795,
+      "step": 12500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.604791207344763e-05,
+      "loss": 1.7097,
+      "step": 12510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6044621409062494e-05,
+      "loss": 1.5799,
+      "step": 12520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6041330744677355e-05,
+      "loss": 1.7727,
+      "step": 12530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.603804008029221e-05,
+      "loss": 1.8539,
+      "step": 12540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.603474941590708e-05,
+      "loss": 1.7059,
+      "step": 12550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.603145875152193e-05,
+      "loss": 1.6014,
+      "step": 12560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.60281680871368e-05,
+      "loss": 1.7489,
+      "step": 12570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6024877422751655e-05,
+      "loss": 1.8405,
+      "step": 12580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6021586758366516e-05,
+      "loss": 1.8133,
+      "step": 12590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.601829609398138e-05,
+      "loss": 1.9722,
+      "step": 12600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.601500542959624e-05,
+      "loss": 1.64,
+      "step": 12610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.60117147652111e-05,
+      "loss": 1.7315,
+      "step": 12620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.600842410082596e-05,
+      "loss": 1.8457,
+      "step": 12630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6005133436440815e-05,
+      "loss": 1.6749,
+      "step": 12640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6001842772055683e-05,
+      "loss": 1.6783,
+      "step": 12650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.599855210767054e-05,
+      "loss": 1.7349,
+      "step": 12660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5995261443285406e-05,
+      "loss": 1.7466,
+      "step": 12670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.599197077890026e-05,
+      "loss": 1.8239,
+      "step": 12680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.598868011451512e-05,
+      "loss": 1.8931,
+      "step": 12690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.598538945012998e-05,
+      "loss": 1.7012,
+      "step": 12700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5982098785744844e-05,
+      "loss": 1.894,
+      "step": 12710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5978808121359706e-05,
+      "loss": 1.7148,
+      "step": 12720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.597551745697457e-05,
+      "loss": 1.6645,
+      "step": 12730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.597222679258942e-05,
+      "loss": 1.7888,
+      "step": 12740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.596893612820429e-05,
+      "loss": 1.7411,
+      "step": 12750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5965645463819144e-05,
+      "loss": 1.8284,
+      "step": 12760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.596235479943401e-05,
+      "loss": 1.7694,
+      "step": 12770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5959064135048866e-05,
+      "loss": 1.7975,
+      "step": 12780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.595577347066373e-05,
+      "loss": 2.0382,
+      "step": 12790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.595248280627859e-05,
+      "loss": 1.8277,
+      "step": 12800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594919214189345e-05,
+      "loss": 1.8154,
+      "step": 12810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594590147750831e-05,
+      "loss": 1.8646,
+      "step": 12820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594261081312317e-05,
+      "loss": 1.8062,
+      "step": 12830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593932014873803e-05,
+      "loss": 1.797,
+      "step": 12840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5936029484352895e-05,
+      "loss": 1.6719,
+      "step": 12850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593273881996775e-05,
+      "loss": 1.8128,
+      "step": 12860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592944815558262e-05,
+      "loss": 1.7134,
+      "step": 12870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592615749119747e-05,
+      "loss": 1.846,
+      "step": 12880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5922866826812334e-05,
+      "loss": 1.8224,
+      "step": 12890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5919576162427195e-05,
+      "loss": 1.7991,
+      "step": 12900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5916285498042056e-05,
+      "loss": 1.7454,
+      "step": 12910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591299483365692e-05,
+      "loss": 1.8909,
+      "step": 12920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590970416927178e-05,
+      "loss": 1.918,
+      "step": 12930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590641350488663e-05,
+      "loss": 1.6349,
+      "step": 12940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59031228405015e-05,
+      "loss": 1.6841,
+      "step": 12950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5899832176116356e-05,
+      "loss": 1.811,
+      "step": 12960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5896541511731224e-05,
+      "loss": 1.7271,
+      "step": 12970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.589325084734608e-05,
+      "loss": 1.7954,
+      "step": 12980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.588996018296094e-05,
+      "loss": 1.6201,
+      "step": 12990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.58866695185758e-05,
+      "loss": 1.7566,
+      "step": 13000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.588337885419066e-05,
+      "loss": 1.8555,
+      "step": 13010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.588008818980552e-05,
+      "loss": 1.8546,
+      "step": 13020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5876797525420384e-05,
+      "loss": 1.7245,
+      "step": 13030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5873506861035246e-05,
+      "loss": 1.8273,
+      "step": 13040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.587021619665011e-05,
+      "loss": 1.7834,
+      "step": 13050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.586692553226497e-05,
+      "loss": 1.9061,
+      "step": 13060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.586363486787983e-05,
+      "loss": 1.7857,
+      "step": 13070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5860344203494684e-05,
+      "loss": 1.9791,
+      "step": 13080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5857053539109545e-05,
+      "loss": 1.8825,
+      "step": 13090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5853762874724407e-05,
+      "loss": 1.8695,
+      "step": 13100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.585047221033927e-05,
+      "loss": 1.6539,
+      "step": 13110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.584718154595413e-05,
+      "loss": 1.7528,
+      "step": 13120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.584389088156899e-05,
+      "loss": 1.8447,
+      "step": 13130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.584060021718385e-05,
+      "loss": 1.6579,
+      "step": 13140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.583730955279871e-05,
+      "loss": 1.8471,
+      "step": 13150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5834018888413574e-05,
+      "loss": 1.7104,
+      "step": 13160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5830728224028435e-05,
+      "loss": 1.9727,
+      "step": 13170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.58274375596433e-05,
+      "loss": 1.8506,
+      "step": 13180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.582414689525815e-05,
+      "loss": 1.8494,
+      "step": 13190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.582085623087302e-05,
+      "loss": 1.6243,
+      "step": 13200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5817565566487874e-05,
+      "loss": 1.7485,
+      "step": 13210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.581427490210274e-05,
+      "loss": 1.8351,
+      "step": 13220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5810984237717596e-05,
+      "loss": 1.7828,
+      "step": 13230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.580769357333246e-05,
+      "loss": 1.8011,
+      "step": 13240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.580440290894732e-05,
+      "loss": 1.7638,
+      "step": 13250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.580111224456218e-05,
+      "loss": 1.6964,
+      "step": 13260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.579782158017704e-05,
+      "loss": 1.9208,
+      "step": 13270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57945309157919e-05,
+      "loss": 2.1007,
+      "step": 13280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5791240251406764e-05,
+      "loss": 1.7464,
+      "step": 13290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5787949587021625e-05,
+      "loss": 1.8231,
+      "step": 13300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.578465892263648e-05,
+      "loss": 1.6391,
+      "step": 13310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.578136825825135e-05,
+      "loss": 1.6823,
+      "step": 13320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57780775938662e-05,
+      "loss": 1.6863,
+      "step": 13330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577478692948107e-05,
+      "loss": 1.6483,
+      "step": 13340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5771496265095925e-05,
+      "loss": 1.7297,
+      "step": 13350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5768205600710786e-05,
+      "loss": 1.6575,
+      "step": 13360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576491493632565e-05,
+      "loss": 1.6455,
+      "step": 13370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576162427194051e-05,
+      "loss": 1.8309,
+      "step": 13380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575833360755537e-05,
+      "loss": 1.9611,
+      "step": 13390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575504294317023e-05,
+      "loss": 1.7031,
+      "step": 13400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5751752278785086e-05,
+      "loss": 1.8506,
+      "step": 13410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5748461614399954e-05,
+      "loss": 1.8056,
+      "step": 13420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574517095001481e-05,
+      "loss": 1.7854,
+      "step": 13430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5741880285629676e-05,
+      "loss": 1.869,
+      "step": 13440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573858962124453e-05,
+      "loss": 1.6265,
+      "step": 13450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573529895685939e-05,
+      "loss": 1.8134,
+      "step": 13460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.573200829247425e-05,
+      "loss": 1.9387,
+      "step": 13470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5728717628089114e-05,
+      "loss": 1.8971,
+      "step": 13480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5725426963703976e-05,
+      "loss": 2.0235,
+      "step": 13490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.572213629931884e-05,
+      "loss": 1.7557,
+      "step": 13500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.571884563493369e-05,
+      "loss": 1.799,
+      "step": 13510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.571555497054856e-05,
+      "loss": 1.9915,
+      "step": 13520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5712264306163414e-05,
+      "loss": 1.54,
+      "step": 13530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.570897364177828e-05,
+      "loss": 1.639,
+      "step": 13540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5705682977393137e-05,
+      "loss": 1.8209,
+      "step": 13550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5702392313008e-05,
+      "loss": 1.846,
+      "step": 13560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.569910164862286e-05,
+      "loss": 1.7919,
+      "step": 13570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.569581098423772e-05,
+      "loss": 1.743,
+      "step": 13580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.569252031985258e-05,
+      "loss": 1.7612,
+      "step": 13590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.568922965546744e-05,
+      "loss": 1.6973,
+      "step": 13600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.56859389910823e-05,
+      "loss": 1.8316,
+      "step": 13610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5682648326697165e-05,
+      "loss": 1.8752,
+      "step": 13620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567935766231202e-05,
+      "loss": 1.685,
+      "step": 13630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567606699792689e-05,
+      "loss": 1.7353,
+      "step": 13640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567277633354174e-05,
+      "loss": 1.7257,
+      "step": 13650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5669485669156604e-05,
+      "loss": 1.7889,
+      "step": 13660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5666195004771465e-05,
+      "loss": 1.6775,
+      "step": 13670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5662904340386326e-05,
+      "loss": 1.6404,
+      "step": 13680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.565961367600119e-05,
+      "loss": 1.6225,
+      "step": 13690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.565632301161605e-05,
+      "loss": 1.7393,
+      "step": 13700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.56530323472309e-05,
+      "loss": 1.6955,
+      "step": 13710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564974168284577e-05,
+      "loss": 1.649,
+      "step": 13720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5646451018460626e-05,
+      "loss": 1.7774,
+      "step": 13730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5643160354075494e-05,
+      "loss": 1.8681,
+      "step": 13740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563986968969035e-05,
+      "loss": 2.0015,
+      "step": 13750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563657902530521e-05,
+      "loss": 1.7103,
+      "step": 13760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563328836092007e-05,
+      "loss": 1.6273,
+      "step": 13770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.562999769653493e-05,
+      "loss": 1.8855,
+      "step": 13780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.562670703214979e-05,
+      "loss": 1.6675,
+      "step": 13790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5623416367764655e-05,
+      "loss": 1.6481,
+      "step": 13800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5620125703379516e-05,
+      "loss": 1.7597,
+      "step": 13810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561683503899438e-05,
+      "loss": 1.6284,
+      "step": 13820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561354437460923e-05,
+      "loss": 1.7711,
+      "step": 13830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.56102537102241e-05,
+      "loss": 1.714,
+      "step": 13840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5606963045838954e-05,
+      "loss": 1.667,
+      "step": 13850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5603672381453815e-05,
+      "loss": 1.7574,
+      "step": 13860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560038171706868e-05,
+      "loss": 1.7835,
+      "step": 13870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559709105268354e-05,
+      "loss": 1.5447,
+      "step": 13880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55938003882984e-05,
+      "loss": 1.7256,
+      "step": 13890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559050972391326e-05,
+      "loss": 1.5874,
+      "step": 13900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558721905952812e-05,
+      "loss": 1.861,
+      "step": 13910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558392839514298e-05,
+      "loss": 1.6972,
+      "step": 13920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5580637730757844e-05,
+      "loss": 1.6327,
+      "step": 13930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5577347066372706e-05,
+      "loss": 1.8682,
+      "step": 13940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557405640198757e-05,
+      "loss": 1.6898,
+      "step": 13950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557076573760242e-05,
+      "loss": 1.729,
+      "step": 13960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556747507321729e-05,
+      "loss": 1.7134,
+      "step": 13970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5564184408832144e-05,
+      "loss": 1.8337,
+      "step": 13980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5560893744447005e-05,
+      "loss": 1.8269,
+      "step": 13990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5557603080061866e-05,
+      "loss": 1.6143,
+      "step": 14000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.555431241567673e-05,
+      "loss": 1.9963,
+      "step": 14010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.555102175129159e-05,
+      "loss": 1.7877,
+      "step": 14020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.554773108690645e-05,
+      "loss": 1.8179,
+      "step": 14030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.554444042252131e-05,
+      "loss": 1.6379,
+      "step": 14040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.554114975813617e-05,
+      "loss": 1.7613,
+      "step": 14050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.553785909375103e-05,
+      "loss": 1.6941,
+      "step": 14060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5534568429365895e-05,
+      "loss": 1.7286,
+      "step": 14070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.553127776498075e-05,
+      "loss": 1.7812,
+      "step": 14080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.552798710059562e-05,
+      "loss": 1.8656,
+      "step": 14090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.552469643621047e-05,
+      "loss": 1.6995,
+      "step": 14100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5521405771825334e-05,
+      "loss": 1.9602,
+      "step": 14110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5518115107440195e-05,
+      "loss": 1.7074,
+      "step": 14120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5514824443055056e-05,
+      "loss": 1.6707,
+      "step": 14130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.551153377866992e-05,
+      "loss": 1.8209,
+      "step": 14140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.550824311428478e-05,
+      "loss": 1.9049,
+      "step": 14150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.550495244989963e-05,
+      "loss": 1.7574,
+      "step": 14160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.55016617855145e-05,
+      "loss": 1.9156,
+      "step": 14170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5498371121129356e-05,
+      "loss": 1.7892,
+      "step": 14180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5495080456744224e-05,
+      "loss": 1.7163,
+      "step": 14190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.549178979235908e-05,
+      "loss": 1.9269,
+      "step": 14200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.548849912797394e-05,
+      "loss": 1.6487,
+      "step": 14210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54852084635888e-05,
+      "loss": 1.6622,
+      "step": 14220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.548191779920366e-05,
+      "loss": 1.6554,
+      "step": 14230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.547862713481852e-05,
+      "loss": 1.6528,
+      "step": 14240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5475336470433385e-05,
+      "loss": 1.717,
+      "step": 14250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.547204580604824e-05,
+      "loss": 1.8423,
+      "step": 14260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546875514166311e-05,
+      "loss": 1.6687,
+      "step": 14270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546546447727796e-05,
+      "loss": 1.7145,
+      "step": 14280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546217381289283e-05,
+      "loss": 1.7268,
+      "step": 14290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5458883148507684e-05,
+      "loss": 1.6722,
+      "step": 14300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5455592484122545e-05,
+      "loss": 1.795,
+      "step": 14310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.545230181973741e-05,
+      "loss": 1.8584,
+      "step": 14320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544901115535227e-05,
+      "loss": 1.843,
+      "step": 14330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544572049096713e-05,
+      "loss": 1.9205,
+      "step": 14340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544242982658199e-05,
+      "loss": 1.6512,
+      "step": 14350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5439139162196845e-05,
+      "loss": 1.7437,
+      "step": 14360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543584849781171e-05,
+      "loss": 1.8326,
+      "step": 14370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543255783342657e-05,
+      "loss": 1.806,
+      "step": 14380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5429267169041436e-05,
+      "loss": 1.6589,
+      "step": 14390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542597650465629e-05,
+      "loss": 1.7425,
+      "step": 14400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542268584027115e-05,
+      "loss": 1.8652,
+      "step": 14410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541939517588601e-05,
+      "loss": 1.6376,
+      "step": 14420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5416104511500874e-05,
+      "loss": 1.7695,
+      "step": 14430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5412813847115735e-05,
+      "loss": 1.7618,
+      "step": 14440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5409523182730596e-05,
+      "loss": 1.8064,
+      "step": 14450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540623251834545e-05,
+      "loss": 1.8029,
+      "step": 14460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540294185396032e-05,
+      "loss": 1.8172,
+      "step": 14470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5399651189575173e-05,
+      "loss": 1.6558,
+      "step": 14480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.539636052519004e-05,
+      "loss": 1.8773,
+      "step": 14490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5393069860804896e-05,
+      "loss": 1.6361,
+      "step": 14500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.538977919641976e-05,
+      "loss": 1.8212,
+      "step": 14510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.538648853203462e-05,
+      "loss": 1.899,
+      "step": 14520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.538319786764948e-05,
+      "loss": 1.842,
+      "step": 14530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.537990720326434e-05,
+      "loss": 1.6222,
+      "step": 14540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.53766165388792e-05,
+      "loss": 1.6495,
+      "step": 14550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5373325874494064e-05,
+      "loss": 1.7511,
+      "step": 14560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5370035210108925e-05,
+      "loss": 1.7721,
+      "step": 14570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5366744545723786e-05,
+      "loss": 1.8482,
+      "step": 14580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.536345388133865e-05,
+      "loss": 1.7513,
+      "step": 14590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.53601632169535e-05,
+      "loss": 1.7456,
+      "step": 14600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.535687255256836e-05,
+      "loss": 1.8556,
+      "step": 14610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5353581888183224e-05,
+      "loss": 1.8188,
+      "step": 14620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5350291223798086e-05,
+      "loss": 1.7406,
+      "step": 14630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.534700055941295e-05,
+      "loss": 1.9728,
+      "step": 14640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.534370989502781e-05,
+      "loss": 1.7172,
+      "step": 14650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.534041923064267e-05,
+      "loss": 1.6168,
+      "step": 14660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.533712856625753e-05,
+      "loss": 1.8106,
+      "step": 14670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.533383790187239e-05,
+      "loss": 1.6342,
+      "step": 14680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.533054723748725e-05,
+      "loss": 1.6596,
+      "step": 14690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5327256573102114e-05,
+      "loss": 1.8231,
+      "step": 14700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.532396590871697e-05,
+      "loss": 1.7476,
+      "step": 14710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.532067524433184e-05,
+      "loss": 1.7863,
+      "step": 14720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.531738457994669e-05,
+      "loss": 1.8224,
+      "step": 14730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.531409391556156e-05,
+      "loss": 1.6283,
+      "step": 14740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5310803251176414e-05,
+      "loss": 1.8076,
+      "step": 14750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5307512586791275e-05,
+      "loss": 1.9085,
+      "step": 14760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5304221922406137e-05,
+      "loss": 1.7555,
+      "step": 14770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5300931258021e-05,
+      "loss": 1.7563,
+      "step": 14780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.529764059363586e-05,
+      "loss": 1.6,
+      "step": 14790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.529434992925072e-05,
+      "loss": 1.6532,
+      "step": 14800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5291059264865575e-05,
+      "loss": 1.5316,
+      "step": 14810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.528776860048044e-05,
+      "loss": 1.6724,
+      "step": 14820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52844779360953e-05,
+      "loss": 1.6197,
+      "step": 14830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5281187271710165e-05,
+      "loss": 1.6786,
+      "step": 14840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527789660732502e-05,
+      "loss": 1.8908,
+      "step": 14850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527460594293988e-05,
+      "loss": 1.602,
+      "step": 14860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527131527855474e-05,
+      "loss": 1.7485,
+      "step": 14870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5268024614169604e-05,
+      "loss": 1.8126,
+      "step": 14880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5264733949784465e-05,
+      "loss": 1.7327,
+      "step": 14890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5261443285399326e-05,
+      "loss": 1.6107,
+      "step": 14900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525815262101418e-05,
+      "loss": 1.7771,
+      "step": 14910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525486195662905e-05,
+      "loss": 1.6692,
+      "step": 14920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52515712922439e-05,
+      "loss": 1.8088,
+      "step": 14930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524828062785877e-05,
+      "loss": 1.8713,
+      "step": 14940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5244989963473626e-05,
+      "loss": 1.7367,
+      "step": 14950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524169929908849e-05,
+      "loss": 1.745,
+      "step": 14960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523840863470335e-05,
+      "loss": 1.7165,
+      "step": 14970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523511797031821e-05,
+      "loss": 1.7782,
+      "step": 14980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523182730593307e-05,
+      "loss": 1.7894,
+      "step": 14990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.522853664154793e-05,
+      "loss": 1.9633,
+      "step": 15000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.522524597716279e-05,
+      "loss": 1.7343,
+      "step": 15010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5221955312777655e-05,
+      "loss": 1.7762,
+      "step": 15020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.521866464839251e-05,
+      "loss": 1.7015,
+      "step": 15030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.521537398400738e-05,
+      "loss": 1.8084,
+      "step": 15040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.521208331962223e-05,
+      "loss": 1.5922,
+      "step": 15050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.520879265523709e-05,
+      "loss": 1.83,
+      "step": 15060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5205501990851954e-05,
+      "loss": 1.8147,
+      "step": 15070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5202211326466816e-05,
+      "loss": 2.142,
+      "step": 15080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.519892066208168e-05,
+      "loss": 1.7075,
+      "step": 15090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.519562999769654e-05,
+      "loss": 1.7024,
+      "step": 15100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.519233933331139e-05,
+      "loss": 1.7561,
+      "step": 15110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.518904866892626e-05,
+      "loss": 1.6627,
+      "step": 15120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5185758004541115e-05,
+      "loss": 1.8012,
+      "step": 15130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.518246734015598e-05,
+      "loss": 1.6552,
+      "step": 15140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.517917667577084e-05,
+      "loss": 1.6786,
+      "step": 15150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.51758860113857e-05,
+      "loss": 1.6627,
+      "step": 15160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.517259534700056e-05,
+      "loss": 1.671,
+      "step": 15170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.516930468261542e-05,
+      "loss": 2.1417,
+      "step": 15180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.516601401823028e-05,
+      "loss": 1.7347,
+      "step": 15190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5162723353845144e-05,
+      "loss": 1.731,
+      "step": 15200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.515943268946e-05,
+      "loss": 1.6905,
+      "step": 15210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5156142025074867e-05,
+      "loss": 1.7662,
+      "step": 15220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.515285136068972e-05,
+      "loss": 1.6711,
+      "step": 15230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.514956069630459e-05,
+      "loss": 1.6061,
+      "step": 15240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5146270031919444e-05,
+      "loss": 1.9225,
+      "step": 15250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5142979367534305e-05,
+      "loss": 1.7764,
+      "step": 15260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5139688703149166e-05,
+      "loss": 1.7423,
+      "step": 15270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.513639803876403e-05,
+      "loss": 1.7384,
+      "step": 15280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.513310737437889e-05,
+      "loss": 1.8602,
+      "step": 15290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.512981670999375e-05,
+      "loss": 1.677,
+      "step": 15300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.512652604560861e-05,
+      "loss": 1.8548,
+      "step": 15310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.512323538122347e-05,
+      "loss": 1.7861,
+      "step": 15320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5119944716838334e-05,
+      "loss": 1.8387,
+      "step": 15330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5116654052453195e-05,
+      "loss": 1.8259,
+      "step": 15340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5113363388068056e-05,
+      "loss": 1.6555,
+      "step": 15350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.511007272368291e-05,
+      "loss": 1.7778,
+      "step": 15360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510678205929777e-05,
+      "loss": 1.7551,
+      "step": 15370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510349139491263e-05,
+      "loss": 1.6273,
+      "step": 15380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5100200730527495e-05,
+      "loss": 1.8265,
+      "step": 15390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5096910066142356e-05,
+      "loss": 1.6562,
+      "step": 15400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509361940175722e-05,
+      "loss": 1.6652,
+      "step": 15410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509032873737208e-05,
+      "loss": 1.6971,
+      "step": 15420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508703807298694e-05,
+      "loss": 1.8638,
+      "step": 15430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50837474086018e-05,
+      "loss": 1.7108,
+      "step": 15440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508045674421666e-05,
+      "loss": 1.7463,
+      "step": 15450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507716607983152e-05,
+      "loss": 1.6474,
+      "step": 15460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5073875415446385e-05,
+      "loss": 1.6965,
+      "step": 15470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507058475106124e-05,
+      "loss": 1.627,
+      "step": 15480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506729408667611e-05,
+      "loss": 1.6354,
+      "step": 15490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.506400342229096e-05,
+      "loss": 1.5983,
+      "step": 15500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.506071275790582e-05,
+      "loss": 1.6864,
+      "step": 15510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.5057422093520684e-05,
+      "loss": 1.7135,
+      "step": 15520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.5054131429135545e-05,
+      "loss": 1.7154,
+      "step": 15530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.505084076475041e-05,
+      "loss": 1.6891,
+      "step": 15540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.504755010036527e-05,
+      "loss": 1.8748,
+      "step": 15550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.504425943598012e-05,
+      "loss": 1.7354,
+      "step": 15560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.504096877159499e-05,
+      "loss": 1.8074,
+      "step": 15570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.5037678107209845e-05,
+      "loss": 1.7699,
+      "step": 15580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.503438744282471e-05,
+      "loss": 1.771,
+      "step": 15590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.503109677843957e-05,
+      "loss": 1.766,
+      "step": 15600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.502780611405443e-05,
+      "loss": 1.7482,
+      "step": 15610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.502451544966929e-05,
+      "loss": 1.7092,
+      "step": 15620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.502122478528415e-05,
+      "loss": 1.7295,
+      "step": 15630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.501793412089901e-05,
+      "loss": 1.8398,
+      "step": 15640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.5014643456513874e-05,
+      "loss": 1.6134,
+      "step": 15650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.501135279212873e-05,
+      "loss": 1.8853,
+      "step": 15660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.5008062127743596e-05,
+      "loss": 1.5941,
+      "step": 15670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.500477146335845e-05,
+      "loss": 1.8255,
+      "step": 15680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.500148079897332e-05,
+      "loss": 1.7531,
+      "step": 15690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4998190134588173e-05,
+      "loss": 2.002,
+      "step": 15700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4994899470203035e-05,
+      "loss": 1.6416,
+      "step": 15710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4991608805817896e-05,
+      "loss": 1.7299,
+      "step": 15720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.498831814143276e-05,
+      "loss": 1.862,
+      "step": 15730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.498502747704762e-05,
+      "loss": 1.7748,
+      "step": 15740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.498173681266248e-05,
+      "loss": 1.8877,
+      "step": 15750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4978446148277334e-05,
+      "loss": 1.9642,
+      "step": 15760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49751554838922e-05,
+      "loss": 1.8953,
+      "step": 15770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.497186481950706e-05,
+      "loss": 1.7721,
+      "step": 15780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4968574155121925e-05,
+      "loss": 1.7471,
+      "step": 15790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.496528349073678e-05,
+      "loss": 1.6508,
+      "step": 15800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.496199282635164e-05,
+      "loss": 1.6644,
+      "step": 15810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49587021619665e-05,
+      "loss": 1.7677,
+      "step": 15820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.495541149758136e-05,
+      "loss": 1.8035,
+      "step": 15830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4952120833196224e-05,
+      "loss": 1.8589,
+      "step": 15840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4948830168811086e-05,
+      "loss": 1.7698,
+      "step": 15850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494553950442594e-05,
+      "loss": 1.7872,
+      "step": 15860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494224884004081e-05,
+      "loss": 1.6149,
+      "step": 15870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493895817565566e-05,
+      "loss": 1.6339,
+      "step": 15880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493566751127053e-05,
+      "loss": 1.6829,
+      "step": 15890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4932376846885385e-05,
+      "loss": 1.725,
+      "step": 15900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4929086182500247e-05,
+      "loss": 1.6228,
+      "step": 15910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492579551811511e-05,
+      "loss": 1.6924,
+      "step": 15920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492250485372997e-05,
+      "loss": 1.768,
+      "step": 15930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491921418934483e-05,
+      "loss": 1.8408,
+      "step": 15940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491592352495969e-05,
+      "loss": 1.858,
+      "step": 15950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4912632860574546e-05,
+      "loss": 1.6632,
+      "step": 15960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4909342196189414e-05,
+      "loss": 1.7909,
+      "step": 15970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490605153180427e-05,
+      "loss": 1.8963,
+      "step": 15980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490276086741914e-05,
+      "loss": 1.8167,
+      "step": 15990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489947020303399e-05,
+      "loss": 1.7725,
+      "step": 16000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.489617953864885e-05,
+      "loss": 1.6255,
+      "step": 16010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4892888874263714e-05,
+      "loss": 1.8724,
+      "step": 16020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4889598209878575e-05,
+      "loss": 1.5703,
+      "step": 16030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4886307545493436e-05,
+      "loss": 1.6142,
+      "step": 16040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.48830168811083e-05,
+      "loss": 1.7345,
+      "step": 16050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.487972621672316e-05,
+      "loss": 1.8193,
+      "step": 16060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.487643555233802e-05,
+      "loss": 1.7469,
+      "step": 16070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.487314488795288e-05,
+      "loss": 1.6476,
+      "step": 16080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.486985422356774e-05,
+      "loss": 1.6031,
+      "step": 16090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4866563559182604e-05,
+      "loss": 1.6252,
+      "step": 16100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.486327289479746e-05,
+      "loss": 1.7669,
+      "step": 16110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.485998223041232e-05,
+      "loss": 1.7331,
+      "step": 16120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.485669156602718e-05,
+      "loss": 1.7204,
+      "step": 16130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.485340090164204e-05,
+      "loss": 1.8207,
+      "step": 16140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4850110237256903e-05,
+      "loss": 1.7497,
+      "step": 16150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4846819572871765e-05,
+      "loss": 2.0923,
+      "step": 16160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4843528908486626e-05,
+      "loss": 1.6573,
+      "step": 16170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.484023824410149e-05,
+      "loss": 1.7793,
+      "step": 16180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.483694757971635e-05,
+      "loss": 1.5975,
+      "step": 16190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.483365691533121e-05,
+      "loss": 1.695,
+      "step": 16200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.483036625094607e-05,
+      "loss": 1.7061,
+      "step": 16210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.482707558656093e-05,
+      "loss": 1.8158,
+      "step": 16220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.482378492217579e-05,
+      "loss": 1.7682,
+      "step": 16230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4820494257790655e-05,
+      "loss": 1.8461,
+      "step": 16240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.481720359340551e-05,
+      "loss": 1.8102,
+      "step": 16250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.481391292902038e-05,
+      "loss": 1.7889,
+      "step": 16260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.481062226463523e-05,
+      "loss": 1.692,
+      "step": 16270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.480733160025009e-05,
+      "loss": 1.8733,
+      "step": 16280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4804040935864954e-05,
+      "loss": 1.788,
+      "step": 16290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4800750271479816e-05,
+      "loss": 1.9928,
+      "step": 16300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.479745960709468e-05,
+      "loss": 1.7083,
+      "step": 16310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.479416894270954e-05,
+      "loss": 1.8449,
+      "step": 16320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.479087827832439e-05,
+      "loss": 1.6971,
+      "step": 16330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.478758761393926e-05,
+      "loss": 1.622,
+      "step": 16340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4784296949554115e-05,
+      "loss": 1.7669,
+      "step": 16350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.478100628516898e-05,
+      "loss": 1.5767,
+      "step": 16360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477771562078384e-05,
+      "loss": 1.8583,
+      "step": 16370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47744249563987e-05,
+      "loss": 1.8136,
+      "step": 16380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477113429201356e-05,
+      "loss": 1.784,
+      "step": 16390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476784362762842e-05,
+      "loss": 1.748,
+      "step": 16400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476455296324328e-05,
+      "loss": 1.7717,
+      "step": 16410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4761262298858144e-05,
+      "loss": 2.1155,
+      "step": 16420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4757971634473e-05,
+      "loss": 2.1173,
+      "step": 16430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475468097008787e-05,
+      "loss": 1.7139,
+      "step": 16440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475139030570272e-05,
+      "loss": 1.5743,
+      "step": 16450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474809964131759e-05,
+      "loss": 1.7315,
+      "step": 16460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4744808976932444e-05,
+      "loss": 1.7641,
+      "step": 16470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4741518312547305e-05,
+      "loss": 1.7666,
+      "step": 16480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4738227648162166e-05,
+      "loss": 1.689,
+      "step": 16490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473493698377703e-05,
+      "loss": 1.8177,
+      "step": 16500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473164631939189e-05,
+      "loss": 1.5781,
+      "step": 16510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.472835565500675e-05,
+      "loss": 1.8356,
+      "step": 16520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4725064990621604e-05,
+      "loss": 1.7715,
+      "step": 16530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.472177432623647e-05,
+      "loss": 1.6873,
+      "step": 16540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.471848366185133e-05,
+      "loss": 1.6929,
+      "step": 16550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4715192997466195e-05,
+      "loss": 1.8099,
+      "step": 16560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.471190233308105e-05,
+      "loss": 1.8956,
+      "step": 16570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.470861166869591e-05,
+      "loss": 1.9114,
+      "step": 16580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.470532100431077e-05,
+      "loss": 1.6322,
+      "step": 16590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.470203033992563e-05,
+      "loss": 1.656,
+      "step": 16600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4698739675540495e-05,
+      "loss": 1.671,
+      "step": 16610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4695449011155356e-05,
+      "loss": 1.6374,
+      "step": 16620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.469215834677021e-05,
+      "loss": 1.5932,
+      "step": 16630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.468886768238508e-05,
+      "loss": 2.0275,
+      "step": 16640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.468557701799993e-05,
+      "loss": 1.8642,
+      "step": 16650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46822863536148e-05,
+      "loss": 1.7615,
+      "step": 16660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4678995689229655e-05,
+      "loss": 1.7804,
+      "step": 16670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.467570502484452e-05,
+      "loss": 1.7933,
+      "step": 16680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.467241436045938e-05,
+      "loss": 1.8037,
+      "step": 16690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.466912369607424e-05,
+      "loss": 1.6485,
+      "step": 16700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46658330316891e-05,
+      "loss": 1.8309,
+      "step": 16710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.466254236730396e-05,
+      "loss": 1.8846,
+      "step": 16720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4659251702918816e-05,
+      "loss": 1.8072,
+      "step": 16730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4655961038533684e-05,
+      "loss": 1.8943,
+      "step": 16740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.465267037414854e-05,
+      "loss": 1.9123,
+      "step": 16750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.464937970976341e-05,
+      "loss": 1.6159,
+      "step": 16760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.464608904537826e-05,
+      "loss": 1.7756,
+      "step": 16770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.464279838099312e-05,
+      "loss": 1.7066,
+      "step": 16780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4639507716607984e-05,
+      "loss": 1.6832,
+      "step": 16790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4636217052222845e-05,
+      "loss": 1.6142,
+      "step": 16800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4632926387837706e-05,
+      "loss": 1.7234,
+      "step": 16810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.462963572345257e-05,
+      "loss": 1.8007,
+      "step": 16820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.462634505906743e-05,
+      "loss": 1.6168,
+      "step": 16830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.462305439468229e-05,
+      "loss": 1.9632,
+      "step": 16840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461976373029715e-05,
+      "loss": 1.7497,
+      "step": 16850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461647306591201e-05,
+      "loss": 1.8478,
+      "step": 16860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4613182401526874e-05,
+      "loss": 1.6334,
+      "step": 16870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460989173714173e-05,
+      "loss": 1.7739,
+      "step": 16880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460660107275659e-05,
+      "loss": 1.637,
+      "step": 16890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460331040837145e-05,
+      "loss": 1.7238,
+      "step": 16900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460001974398631e-05,
+      "loss": 1.6839,
+      "step": 16910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4596729079601174e-05,
+      "loss": 1.8136,
+      "step": 16920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4593438415216035e-05,
+      "loss": 1.7397,
+      "step": 16930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4590147750830896e-05,
+      "loss": 1.8198,
+      "step": 16940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458685708644576e-05,
+      "loss": 1.5818,
+      "step": 16950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458356642206062e-05,
+      "loss": 1.7258,
+      "step": 16960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458027575767548e-05,
+      "loss": 1.725,
+      "step": 16970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4576985093290334e-05,
+      "loss": 1.6812,
+      "step": 16980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45736944289052e-05,
+      "loss": 1.7419,
+      "step": 16990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457040376452006e-05,
+      "loss": 1.7473,
+      "step": 17000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4567113100134925e-05,
+      "loss": 1.8262,
+      "step": 17010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456382243574978e-05,
+      "loss": 1.8815,
+      "step": 17020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.456053177136464e-05,
+      "loss": 1.8024,
+      "step": 17030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.45572411069795e-05,
+      "loss": 1.6283,
+      "step": 17040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.455395044259436e-05,
+      "loss": 1.9301,
+      "step": 17050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4550659778209225e-05,
+      "loss": 1.6818,
+      "step": 17060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4547369113824086e-05,
+      "loss": 1.62,
+      "step": 17070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.454407844943894e-05,
+      "loss": 1.7772,
+      "step": 17080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.454078778505381e-05,
+      "loss": 2.0472,
+      "step": 17090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.453749712066866e-05,
+      "loss": 1.5708,
+      "step": 17100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.453420645628353e-05,
+      "loss": 1.8939,
+      "step": 17110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4530915791898385e-05,
+      "loss": 1.8064,
+      "step": 17120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.452762512751325e-05,
+      "loss": 1.7829,
+      "step": 17130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.452433446312811e-05,
+      "loss": 1.7315,
+      "step": 17140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.452104379874297e-05,
+      "loss": 1.5912,
+      "step": 17150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.451775313435783e-05,
+      "loss": 1.6841,
+      "step": 17160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.451446246997269e-05,
+      "loss": 1.7634,
+      "step": 17170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4511171805587546e-05,
+      "loss": 1.6914,
+      "step": 17180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4507881141202414e-05,
+      "loss": 1.896,
+      "step": 17190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.450459047681727e-05,
+      "loss": 1.601,
+      "step": 17200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.450129981243214e-05,
+      "loss": 1.5853,
+      "step": 17210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.449800914804699e-05,
+      "loss": 1.5299,
+      "step": 17220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.449471848366185e-05,
+      "loss": 1.6882,
+      "step": 17230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4491427819276714e-05,
+      "loss": 1.8578,
+      "step": 17240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4488137154891575e-05,
+      "loss": 1.779,
+      "step": 17250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4484846490506436e-05,
+      "loss": 1.6859,
+      "step": 17260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44815558261213e-05,
+      "loss": 1.7339,
+      "step": 17270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.447826516173615e-05,
+      "loss": 1.6125,
+      "step": 17280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.447497449735102e-05,
+      "loss": 1.8085,
+      "step": 17290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4471683832965875e-05,
+      "loss": 1.7675,
+      "step": 17300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.446839316858074e-05,
+      "loss": 1.7161,
+      "step": 17310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44651025041956e-05,
+      "loss": 1.7077,
+      "step": 17320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.446181183981046e-05,
+      "loss": 1.7375,
+      "step": 17330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.445852117542532e-05,
+      "loss": 1.7161,
+      "step": 17340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.445523051104018e-05,
+      "loss": 1.757,
+      "step": 17350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.445193984665504e-05,
+      "loss": 1.787,
+      "step": 17360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4448649182269904e-05,
+      "loss": 1.6612,
+      "step": 17370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444535851788476e-05,
+      "loss": 1.8955,
+      "step": 17380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4442067853499626e-05,
+      "loss": 1.9529,
+      "step": 17390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443877718911448e-05,
+      "loss": 1.883,
+      "step": 17400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443548652472935e-05,
+      "loss": 1.8729,
+      "step": 17410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44321958603442e-05,
+      "loss": 1.6312,
+      "step": 17420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4428905195959064e-05,
+      "loss": 1.5909,
+      "step": 17430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4425614531573926e-05,
+      "loss": 1.7709,
+      "step": 17440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442232386718879e-05,
+      "loss": 1.805,
+      "step": 17450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441903320280365e-05,
+      "loss": 1.6484,
+      "step": 17460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441574253841851e-05,
+      "loss": 1.7871,
+      "step": 17470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4412451874033364e-05,
+      "loss": 1.6281,
+      "step": 17480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440916120964823e-05,
+      "loss": 1.7602,
+      "step": 17490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4405870545263086e-05,
+      "loss": 1.7234,
+      "step": 17500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4402579880877954e-05,
+      "loss": 1.5365,
+      "step": 17510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439928921649281e-05,
+      "loss": 1.8081,
+      "step": 17520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439599855210767e-05,
+      "loss": 1.5951,
+      "step": 17530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.439270788772253e-05,
+      "loss": 1.6471,
+      "step": 17540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.438941722333739e-05,
+      "loss": 1.8399,
+      "step": 17550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4386126558952254e-05,
+      "loss": 1.8193,
+      "step": 17560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4382835894567115e-05,
+      "loss": 1.5794,
+      "step": 17570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4379545230181977e-05,
+      "loss": 1.821,
+      "step": 17580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.437625456579684e-05,
+      "loss": 1.9478,
+      "step": 17590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.43729639014117e-05,
+      "loss": 1.6797,
+      "step": 17600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.436967323702656e-05,
+      "loss": 1.8305,
+      "step": 17610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.436638257264142e-05,
+      "loss": 1.6252,
+      "step": 17620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4363091908256276e-05,
+      "loss": 1.676,
+      "step": 17630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.435980124387114e-05,
+      "loss": 1.7125,
+      "step": 17640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4356510579486e-05,
+      "loss": 1.649,
+      "step": 17650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.435321991510086e-05,
+      "loss": 1.7296,
+      "step": 17660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.434992925071572e-05,
+      "loss": 1.5723,
+      "step": 17670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.434663858633058e-05,
+      "loss": 1.6152,
+      "step": 17680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4343347921945444e-05,
+      "loss": 1.8308,
+      "step": 17690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4340057257560305e-05,
+      "loss": 1.8541,
+      "step": 17700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4336766593175166e-05,
+      "loss": 1.5502,
+      "step": 17710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.433347592879003e-05,
+      "loss": 1.6743,
+      "step": 17720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.433018526440488e-05,
+      "loss": 1.8529,
+      "step": 17730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.432689460001975e-05,
+      "loss": 1.8196,
+      "step": 17740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4323603935634605e-05,
+      "loss": 1.6337,
+      "step": 17750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.432031327124947e-05,
+      "loss": 1.7758,
+      "step": 17760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.431702260686433e-05,
+      "loss": 1.7941,
+      "step": 17770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.431373194247919e-05,
+      "loss": 1.7585,
+      "step": 17780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.431044127809405e-05,
+      "loss": 1.7017,
+      "step": 17790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.430715061370891e-05,
+      "loss": 1.6197,
+      "step": 17800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.430385994932377e-05,
+      "loss": 1.7785,
+      "step": 17810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4300569284938633e-05,
+      "loss": 1.7419,
+      "step": 17820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.429727862055349e-05,
+      "loss": 1.6608,
+      "step": 17830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4293987956168356e-05,
+      "loss": 1.8221,
+      "step": 17840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.429069729178321e-05,
+      "loss": 1.7339,
+      "step": 17850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.428740662739808e-05,
+      "loss": 1.8389,
+      "step": 17860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.428411596301293e-05,
+      "loss": 1.5791,
+      "step": 17870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4280825298627794e-05,
+      "loss": 1.8333,
+      "step": 17880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4277534634242656e-05,
+      "loss": 1.9584,
+      "step": 17890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427424396985752e-05,
+      "loss": 1.7294,
+      "step": 17900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427095330547238e-05,
+      "loss": 1.7596,
+      "step": 17910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426766264108724e-05,
+      "loss": 1.8259,
+      "step": 17920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4264371976702094e-05,
+      "loss": 1.8076,
+      "step": 17930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426108131231696e-05,
+      "loss": 1.6435,
+      "step": 17940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4257790647931816e-05,
+      "loss": 1.7664,
+      "step": 17950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4254499983546684e-05,
+      "loss": 1.6735,
+      "step": 17960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425120931916154e-05,
+      "loss": 1.7375,
+      "step": 17970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42479186547764e-05,
+      "loss": 1.6977,
+      "step": 17980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424462799039126e-05,
+      "loss": 1.7976,
+      "step": 17990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424133732600612e-05,
+      "loss": 1.8278,
+      "step": 18000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4238046661620984e-05,
+      "loss": 1.7742,
+      "step": 18010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4234755997235845e-05,
+      "loss": 1.8974,
+      "step": 18020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42314653328507e-05,
+      "loss": 1.7889,
+      "step": 18030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.422817466846557e-05,
+      "loss": 2.0267,
+      "step": 18040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.422488400408042e-05,
+      "loss": 1.7823,
+      "step": 18050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.422159333969529e-05,
+      "loss": 1.8009,
+      "step": 18060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4218302675310145e-05,
+      "loss": 1.6525,
+      "step": 18070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4215012010925006e-05,
+      "loss": 1.8184,
+      "step": 18080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.421172134653987e-05,
+      "loss": 1.6891,
+      "step": 18090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.420843068215473e-05,
+      "loss": 1.7261,
+      "step": 18100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.420514001776959e-05,
+      "loss": 1.7294,
+      "step": 18110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.420184935338445e-05,
+      "loss": 1.6177,
+      "step": 18120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4198558688999306e-05,
+      "loss": 1.6914,
+      "step": 18130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4195268024614174e-05,
+      "loss": 1.8694,
+      "step": 18140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.419197736022903e-05,
+      "loss": 1.6652,
+      "step": 18150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4188686695843896e-05,
+      "loss": 1.6828,
+      "step": 18160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.418539603145875e-05,
+      "loss": 1.5974,
+      "step": 18170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.418210536707361e-05,
+      "loss": 1.6936,
+      "step": 18180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.417881470268847e-05,
+      "loss": 1.7619,
+      "step": 18190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4175524038303334e-05,
+      "loss": 1.6461,
+      "step": 18200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4172233373918196e-05,
+      "loss": 1.8905,
+      "step": 18210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.416894270953306e-05,
+      "loss": 1.697,
+      "step": 18220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.416565204514792e-05,
+      "loss": 1.8097,
+      "step": 18230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.416236138076278e-05,
+      "loss": 1.7674,
+      "step": 18240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4159070716377634e-05,
+      "loss": 1.7619,
+      "step": 18250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.41557800519925e-05,
+      "loss": 2.0526,
+      "step": 18260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4152489387607357e-05,
+      "loss": 2.0831,
+      "step": 18270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.414919872322222e-05,
+      "loss": 1.6645,
+      "step": 18280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.414590805883708e-05,
+      "loss": 1.6821,
+      "step": 18290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.414261739445194e-05,
+      "loss": 1.949,
+      "step": 18300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.41393267300668e-05,
+      "loss": 1.6895,
+      "step": 18310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.413603606568166e-05,
+      "loss": 1.7015,
+      "step": 18320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4132745401296524e-05,
+      "loss": 1.7156,
+      "step": 18330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4129454736911385e-05,
+      "loss": 1.752,
+      "step": 18340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.412616407252625e-05,
+      "loss": 1.7345,
+      "step": 18350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.412287340814111e-05,
+      "loss": 1.8759,
+      "step": 18360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411958274375597e-05,
+      "loss": 1.6923,
+      "step": 18370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4116292079370824e-05,
+      "loss": 1.7463,
+      "step": 18380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411300141498569e-05,
+      "loss": 1.8249,
+      "step": 18390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4109710750600546e-05,
+      "loss": 1.6317,
+      "step": 18400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410642008621541e-05,
+      "loss": 1.8215,
+      "step": 18410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410312942183027e-05,
+      "loss": 1.7774,
+      "step": 18420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409983875744513e-05,
+      "loss": 1.6637,
+      "step": 18430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409654809305999e-05,
+      "loss": 1.6055,
+      "step": 18440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409325742867485e-05,
+      "loss": 1.7794,
+      "step": 18450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4089966764289714e-05,
+      "loss": 1.6888,
+      "step": 18460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4086676099904575e-05,
+      "loss": 1.6317,
+      "step": 18470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408338543551943e-05,
+      "loss": 1.8785,
+      "step": 18480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40800947711343e-05,
+      "loss": 1.6176,
+      "step": 18490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407680410674915e-05,
+      "loss": 2.1114,
+      "step": 18500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407351344236402e-05,
+      "loss": 1.7217,
+      "step": 18510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4070222777978875e-05,
+      "loss": 1.7191,
+      "step": 18520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4066932113593736e-05,
+      "loss": 1.7246,
+      "step": 18530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40636414492086e-05,
+      "loss": 1.8694,
+      "step": 18540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.406035078482346e-05,
+      "loss": 1.7406,
+      "step": 18550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.405706012043832e-05,
+      "loss": 1.7713,
+      "step": 18560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.405376945605318e-05,
+      "loss": 1.6461,
+      "step": 18570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4050478791668036e-05,
+      "loss": 1.7514,
+      "step": 18580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4047188127282904e-05,
+      "loss": 1.8778,
+      "step": 18590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.404389746289776e-05,
+      "loss": 1.7439,
+      "step": 18600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4040606798512626e-05,
+      "loss": 1.7856,
+      "step": 18610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.403731613412748e-05,
+      "loss": 1.8709,
+      "step": 18620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.403402546974234e-05,
+      "loss": 1.7435,
+      "step": 18630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.40307348053572e-05,
+      "loss": 1.6636,
+      "step": 18640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4027444140972064e-05,
+      "loss": 1.762,
+      "step": 18650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4024153476586926e-05,
+      "loss": 1.4998,
+      "step": 18660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.402086281220179e-05,
+      "loss": 1.7837,
+      "step": 18670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.401757214781664e-05,
+      "loss": 1.6219,
+      "step": 18680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.401428148343151e-05,
+      "loss": 1.8014,
+      "step": 18690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4010990819046364e-05,
+      "loss": 1.5901,
+      "step": 18700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.400770015466123e-05,
+      "loss": 1.7323,
+      "step": 18710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4004409490276087e-05,
+      "loss": 1.6907,
+      "step": 18720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.400111882589095e-05,
+      "loss": 1.99,
+      "step": 18730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.399782816150581e-05,
+      "loss": 1.6704,
+      "step": 18740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.399453749712067e-05,
+      "loss": 1.5246,
+      "step": 18750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.399124683273553e-05,
+      "loss": 1.7102,
+      "step": 18760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.398795616835039e-05,
+      "loss": 1.7779,
+      "step": 18770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.398466550396525e-05,
+      "loss": 1.7274,
+      "step": 18780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3981374839580115e-05,
+      "loss": 1.8552,
+      "step": 18790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.397808417519497e-05,
+      "loss": 1.7686,
+      "step": 18800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.397479351080984e-05,
+      "loss": 1.7604,
+      "step": 18810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.397150284642469e-05,
+      "loss": 1.8193,
+      "step": 18820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3968212182039554e-05,
+      "loss": 1.7789,
+      "step": 18830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3964921517654415e-05,
+      "loss": 1.8073,
+      "step": 18840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3961630853269276e-05,
+      "loss": 1.7658,
+      "step": 18850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.395834018888414e-05,
+      "loss": 1.9115,
+      "step": 18860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3955049524499e-05,
+      "loss": 1.7628,
+      "step": 18870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.395175886011385e-05,
+      "loss": 1.8104,
+      "step": 18880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394846819572872e-05,
+      "loss": 1.7,
+      "step": 18890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3945177531343576e-05,
+      "loss": 1.7358,
+      "step": 18900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3941886866958444e-05,
+      "loss": 1.7238,
+      "step": 18910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.39385962025733e-05,
+      "loss": 1.6289,
+      "step": 18920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393530553818816e-05,
+      "loss": 1.6932,
+      "step": 18930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393201487380302e-05,
+      "loss": 1.7841,
+      "step": 18940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392872420941788e-05,
+      "loss": 1.6456,
+      "step": 18950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392543354503274e-05,
+      "loss": 1.8017,
+      "step": 18960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3922142880647605e-05,
+      "loss": 1.8974,
+      "step": 18970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3918852216262466e-05,
+      "loss": 1.5489,
+      "step": 18980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391556155187733e-05,
+      "loss": 1.735,
+      "step": 18990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391227088749219e-05,
+      "loss": 1.9482,
+      "step": 19000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390898022310705e-05,
+      "loss": 1.6939,
+      "step": 19010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3905689558721904e-05,
+      "loss": 1.8198,
+      "step": 19020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390239889433677e-05,
+      "loss": 1.7553,
+      "step": 19030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389910822995163e-05,
+      "loss": 1.7207,
+      "step": 19040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389581756556649e-05,
+      "loss": 1.575,
+      "step": 19050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.389252690118135e-05,
+      "loss": 1.6882,
+      "step": 19060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.388923623679621e-05,
+      "loss": 1.6145,
+      "step": 19070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.388594557241107e-05,
+      "loss": 1.7402,
+      "step": 19080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.388265490802593e-05,
+      "loss": 1.7474,
+      "step": 19090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3879364243640794e-05,
+      "loss": 1.8259,
+      "step": 19100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3876073579255656e-05,
+      "loss": 1.7142,
+      "step": 19110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.387278291487052e-05,
+      "loss": 1.6522,
+      "step": 19120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.386949225048538e-05,
+      "loss": 1.7748,
+      "step": 19130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.386620158610024e-05,
+      "loss": 2.1873,
+      "step": 19140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3862910921715094e-05,
+      "loss": 1.6914,
+      "step": 19150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.385962025732996e-05,
+      "loss": 1.6481,
+      "step": 19160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3856329592944816e-05,
+      "loss": 1.6912,
+      "step": 19170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.385303892855968e-05,
+      "loss": 1.8204,
+      "step": 19180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.384974826417454e-05,
+      "loss": 1.5818,
+      "step": 19190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.38464575997894e-05,
+      "loss": 1.7766,
+      "step": 19200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.384316693540426e-05,
+      "loss": 1.9009,
+      "step": 19210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.383987627101912e-05,
+      "loss": 1.7608,
+      "step": 19220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3836585606633984e-05,
+      "loss": 1.731,
+      "step": 19230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3833294942248845e-05,
+      "loss": 1.9734,
+      "step": 19240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.38300042778637e-05,
+      "loss": 1.7881,
+      "step": 19250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.382671361347857e-05,
+      "loss": 1.755,
+      "step": 19260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.382342294909342e-05,
+      "loss": 1.851,
+      "step": 19270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.382013228470829e-05,
+      "loss": 1.7106,
+      "step": 19280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3816841620323145e-05,
+      "loss": 1.6269,
+      "step": 19290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3813550955938006e-05,
+      "loss": 1.8018,
+      "step": 19300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.381026029155287e-05,
+      "loss": 1.7926,
+      "step": 19310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.380696962716773e-05,
+      "loss": 1.8014,
+      "step": 19320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.380367896278259e-05,
+      "loss": 1.6401,
+      "step": 19330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.380038829839745e-05,
+      "loss": 1.6777,
+      "step": 19340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3797097634012306e-05,
+      "loss": 1.6723,
+      "step": 19350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3793806969627174e-05,
+      "loss": 1.7523,
+      "step": 19360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.379051630524203e-05,
+      "loss": 1.9204,
+      "step": 19370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3787225640856896e-05,
+      "loss": 1.7151,
+      "step": 19380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.378393497647175e-05,
+      "loss": 1.7065,
+      "step": 19390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.378064431208661e-05,
+      "loss": 1.7051,
+      "step": 19400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377735364770147e-05,
+      "loss": 1.7054,
+      "step": 19410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3774062983316335e-05,
+      "loss": 1.8695,
+      "step": 19420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3770772318931196e-05,
+      "loss": 1.9732,
+      "step": 19430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376748165454606e-05,
+      "loss": 1.8818,
+      "step": 19440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376419099016091e-05,
+      "loss": 1.8561,
+      "step": 19450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376090032577578e-05,
+      "loss": 1.6444,
+      "step": 19460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3757609661390634e-05,
+      "loss": 1.7354,
+      "step": 19470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37543189970055e-05,
+      "loss": 1.6618,
+      "step": 19480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375102833262036e-05,
+      "loss": 1.7069,
+      "step": 19490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374773766823522e-05,
+      "loss": 1.824,
+      "step": 19500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374444700385008e-05,
+      "loss": 1.6934,
+      "step": 19510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374115633946494e-05,
+      "loss": 1.7652,
+      "step": 19520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37378656750798e-05,
+      "loss": 1.5653,
+      "step": 19530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373457501069466e-05,
+      "loss": 1.7998,
+      "step": 19540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373128434630952e-05,
+      "loss": 1.7538,
+      "step": 19550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3727993681924386e-05,
+      "loss": 1.7112,
+      "step": 19560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.372470301753924e-05,
+      "loss": 1.7459,
+      "step": 19570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.372141235315411e-05,
+      "loss": 1.638,
+      "step": 19580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.371812168876896e-05,
+      "loss": 1.7915,
+      "step": 19590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3714831024383824e-05,
+      "loss": 1.7151,
+      "step": 19600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3711540359998685e-05,
+      "loss": 1.8781,
+      "step": 19610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3708249695613546e-05,
+      "loss": 1.5684,
+      "step": 19620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.370495903122841e-05,
+      "loss": 1.5818,
+      "step": 19630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.370166836684327e-05,
+      "loss": 1.7525,
+      "step": 19640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3698377702458123e-05,
+      "loss": 1.7249,
+      "step": 19650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.369508703807299e-05,
+      "loss": 1.6471,
+      "step": 19660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3691796373687846e-05,
+      "loss": 1.6375,
+      "step": 19670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3688505709302714e-05,
+      "loss": 1.7251,
+      "step": 19680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.368521504491757e-05,
+      "loss": 1.6398,
+      "step": 19690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.368192438053243e-05,
+      "loss": 1.7341,
+      "step": 19700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.367863371614729e-05,
+      "loss": 1.6729,
+      "step": 19710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.367534305176215e-05,
+      "loss": 1.7767,
+      "step": 19720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3672052387377014e-05,
+      "loss": 1.6266,
+      "step": 19730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3668761722991875e-05,
+      "loss": 1.6182,
+      "step": 19740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3665471058606736e-05,
+      "loss": 1.8294,
+      "step": 19750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.36621803942216e-05,
+      "loss": 1.5519,
+      "step": 19760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.365888972983645e-05,
+      "loss": 1.6965,
+      "step": 19770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.365559906545132e-05,
+      "loss": 1.9279,
+      "step": 19780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3652308401066174e-05,
+      "loss": 1.7637,
+      "step": 19790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3649017736681036e-05,
+      "loss": 1.5743,
+      "step": 19800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.36457270722959e-05,
+      "loss": 1.8663,
+      "step": 19810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.364243640791076e-05,
+      "loss": 1.6477,
+      "step": 19820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363914574352562e-05,
+      "loss": 1.8201,
+      "step": 19830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363585507914048e-05,
+      "loss": 1.7124,
+      "step": 19840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363256441475534e-05,
+      "loss": 1.9202,
+      "step": 19850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.36292737503702e-05,
+      "loss": 1.7103,
+      "step": 19860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3625983085985065e-05,
+      "loss": 1.8205,
+      "step": 19870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3622692421599926e-05,
+      "loss": 1.9933,
+      "step": 19880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.361940175721479e-05,
+      "loss": 1.7246,
+      "step": 19890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.361611109282964e-05,
+      "loss": 1.5868,
+      "step": 19900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.361282042844451e-05,
+      "loss": 1.6973,
+      "step": 19910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3609529764059364e-05,
+      "loss": 1.7042,
+      "step": 19920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3606239099674225e-05,
+      "loss": 1.6311,
+      "step": 19930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3602948435289087e-05,
+      "loss": 1.8632,
+      "step": 19940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359965777090395e-05,
+      "loss": 1.7619,
+      "step": 19950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359636710651881e-05,
+      "loss": 1.6691,
+      "step": 19960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359307644213367e-05,
+      "loss": 1.9519,
+      "step": 19970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358978577774853e-05,
+      "loss": 1.5983,
+      "step": 19980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358649511336339e-05,
+      "loss": 1.7417,
+      "step": 19990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358320444897825e-05,
+      "loss": 1.7615,
+      "step": 20000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3579913784593115e-05,
+      "loss": 1.6096,
+      "step": 20010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357662312020797e-05,
+      "loss": 1.6387,
+      "step": 20020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357333245582284e-05,
+      "loss": 1.7311,
+      "step": 20030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357004179143769e-05,
+      "loss": 1.7792,
+      "step": 20040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3566751127052554e-05,
+      "loss": 1.9812,
+      "step": 20050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3563460462667415e-05,
+      "loss": 1.5117,
+      "step": 20060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3560169798282276e-05,
+      "loss": 1.906,
+      "step": 20070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.355687913389714e-05,
+      "loss": 1.8123,
+      "step": 20080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3553588469512e-05,
+      "loss": 1.7897,
+      "step": 20090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.355029780512685e-05,
+      "loss": 1.7276,
+      "step": 20100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.354700714074172e-05,
+      "loss": 1.7697,
+      "step": 20110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3543716476356576e-05,
+      "loss": 1.6313,
+      "step": 20120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3540425811971444e-05,
+      "loss": 1.7174,
+      "step": 20130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.35371351475863e-05,
+      "loss": 1.6237,
+      "step": 20140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.353384448320116e-05,
+      "loss": 1.7247,
+      "step": 20150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.353055381881602e-05,
+      "loss": 1.848,
+      "step": 20160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.352726315443088e-05,
+      "loss": 1.7514,
+      "step": 20170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3523972490045743e-05,
+      "loss": 1.7488,
+      "step": 20180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3520681825660605e-05,
+      "loss": 1.6919,
+      "step": 20190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.351739116127546e-05,
+      "loss": 1.8402,
+      "step": 20200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.351410049689033e-05,
+      "loss": 1.6414,
+      "step": 20210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.351080983250518e-05,
+      "loss": 1.7582,
+      "step": 20220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.350751916812005e-05,
+      "loss": 1.6445,
+      "step": 20230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3504228503734904e-05,
+      "loss": 1.6999,
+      "step": 20240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3500937839349766e-05,
+      "loss": 1.6874,
+      "step": 20250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.349764717496463e-05,
+      "loss": 1.7205,
+      "step": 20260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.349435651057949e-05,
+      "loss": 1.8286,
+      "step": 20270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.349106584619435e-05,
+      "loss": 1.7246,
+      "step": 20280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.348777518180921e-05,
+      "loss": 2.1191,
+      "step": 20290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3484484517424065e-05,
+      "loss": 1.7899,
+      "step": 20300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.348119385303893e-05,
+      "loss": 1.5539,
+      "step": 20310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.347790318865379e-05,
+      "loss": 1.7408,
+      "step": 20320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3474612524268656e-05,
+      "loss": 1.8319,
+      "step": 20330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.347132185988351e-05,
+      "loss": 1.8338,
+      "step": 20340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.346803119549837e-05,
+      "loss": 1.6074,
+      "step": 20350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.346474053111323e-05,
+      "loss": 1.7856,
+      "step": 20360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3461449866728094e-05,
+      "loss": 1.616,
+      "step": 20370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3458159202342955e-05,
+      "loss": 1.7188,
+      "step": 20380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3454868537957817e-05,
+      "loss": 1.8102,
+      "step": 20390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.345157787357267e-05,
+      "loss": 1.9822,
+      "step": 20400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344828720918754e-05,
+      "loss": 1.7867,
+      "step": 20410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3444996544802394e-05,
+      "loss": 1.6424,
+      "step": 20420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344170588041726e-05,
+      "loss": 1.7378,
+      "step": 20430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3438415216032116e-05,
+      "loss": 1.6992,
+      "step": 20440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343512455164698e-05,
+      "loss": 1.7303,
+      "step": 20450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343183388726184e-05,
+      "loss": 1.6532,
+      "step": 20460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34285432228767e-05,
+      "loss": 1.6133,
+      "step": 20470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342525255849156e-05,
+      "loss": 1.7632,
+      "step": 20480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342196189410642e-05,
+      "loss": 1.7284,
+      "step": 20490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3418671229721284e-05,
+      "loss": 1.5769,
+      "step": 20500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3415380565336145e-05,
+      "loss": 1.5823,
+      "step": 20510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3412089900951006e-05,
+      "loss": 1.5361,
+      "step": 20520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340879923656587e-05,
+      "loss": 1.8125,
+      "step": 20530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340550857218072e-05,
+      "loss": 1.8339,
+      "step": 20540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340221790779558e-05,
+      "loss": 1.7018,
+      "step": 20550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3398927243410445e-05,
+      "loss": 1.7013,
+      "step": 20560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3395636579025306e-05,
+      "loss": 1.4933,
+      "step": 20570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339234591464017e-05,
+      "loss": 1.6798,
+      "step": 20580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.338905525025503e-05,
+      "loss": 1.6754,
+      "step": 20590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.338576458586989e-05,
+      "loss": 1.7457,
+      "step": 20600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.338247392148475e-05,
+      "loss": 1.6394,
+      "step": 20610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.337918325709961e-05,
+      "loss": 1.7864,
+      "step": 20620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3375892592714473e-05,
+      "loss": 1.8463,
+      "step": 20630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3372601928329335e-05,
+      "loss": 1.7553,
+      "step": 20640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.336931126394419e-05,
+      "loss": 1.6119,
+      "step": 20650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.336602059955906e-05,
+      "loss": 1.8468,
+      "step": 20660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.336272993517391e-05,
+      "loss": 1.6957,
+      "step": 20670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.335943927078878e-05,
+      "loss": 1.779,
+      "step": 20680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3356148606403634e-05,
+      "loss": 1.7831,
+      "step": 20690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3352857942018496e-05,
+      "loss": 1.5944,
+      "step": 20700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.334956727763336e-05,
+      "loss": 1.651,
+      "step": 20710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.334627661324822e-05,
+      "loss": 1.8279,
+      "step": 20720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.334298594886308e-05,
+      "loss": 1.7202,
+      "step": 20730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.333969528447794e-05,
+      "loss": 1.621,
+      "step": 20740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3336404620092795e-05,
+      "loss": 1.6901,
+      "step": 20750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.333311395570766e-05,
+      "loss": 1.9571,
+      "step": 20760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.332982329132252e-05,
+      "loss": 1.6943,
+      "step": 20770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3326532626937386e-05,
+      "loss": 1.7713,
+      "step": 20780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.332324196255224e-05,
+      "loss": 1.7157,
+      "step": 20790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.33199512981671e-05,
+      "loss": 1.7139,
+      "step": 20800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.331666063378196e-05,
+      "loss": 1.6111,
+      "step": 20810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3313369969396824e-05,
+      "loss": 1.693,
+      "step": 20820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3310079305011685e-05,
+      "loss": 1.8316,
+      "step": 20830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3306788640626546e-05,
+      "loss": 1.7638,
+      "step": 20840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.33034979762414e-05,
+      "loss": 1.677,
+      "step": 20850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.330020731185627e-05,
+      "loss": 1.7511,
+      "step": 20860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3296916647471123e-05,
+      "loss": 1.5565,
+      "step": 20870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.329362598308599e-05,
+      "loss": 1.8332,
+      "step": 20880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3290335318700846e-05,
+      "loss": 1.6315,
+      "step": 20890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.328704465431571e-05,
+      "loss": 1.8024,
+      "step": 20900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.328375398993057e-05,
+      "loss": 1.8321,
+      "step": 20910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.328046332554543e-05,
+      "loss": 1.8105,
+      "step": 20920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327717266116029e-05,
+      "loss": 1.8447,
+      "step": 20930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327388199677515e-05,
+      "loss": 1.6942,
+      "step": 20940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327059133239001e-05,
+      "loss": 1.7811,
+      "step": 20950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3267300668004875e-05,
+      "loss": 1.7966,
+      "step": 20960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326401000361973e-05,
+      "loss": 1.7999,
+      "step": 20970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32607193392346e-05,
+      "loss": 1.7683,
+      "step": 20980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325742867484945e-05,
+      "loss": 1.7687,
+      "step": 20990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325413801046431e-05,
+      "loss": 1.7992,
+      "step": 21000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3250847346079174e-05,
+      "loss": 1.8141,
+      "step": 21010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3247556681694036e-05,
+      "loss": 1.7942,
+      "step": 21020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32442660173089e-05,
+      "loss": 1.676,
+      "step": 21030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324097535292376e-05,
+      "loss": 1.6187,
+      "step": 21040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323768468853861e-05,
+      "loss": 1.6819,
+      "step": 21050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323439402415348e-05,
+      "loss": 1.8099,
+      "step": 21060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3231103359768335e-05,
+      "loss": 1.5999,
+      "step": 21070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32278126953832e-05,
+      "loss": 1.7314,
+      "step": 21080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.322452203099806e-05,
+      "loss": 1.6449,
+      "step": 21090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.322123136661292e-05,
+      "loss": 1.878,
+      "step": 21100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.321794070222778e-05,
+      "loss": 1.6958,
+      "step": 21110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.321465003784264e-05,
+      "loss": 1.7473,
+      "step": 21120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.32113593734575e-05,
+      "loss": 1.6756,
+      "step": 21130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3208068709072364e-05,
+      "loss": 1.8061,
+      "step": 21140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.320477804468722e-05,
+      "loss": 1.7518,
+      "step": 21150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.320148738030209e-05,
+      "loss": 1.8097,
+      "step": 21160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.319819671591694e-05,
+      "loss": 1.726,
+      "step": 21170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.319490605153181e-05,
+      "loss": 1.7457,
+      "step": 21180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3191615387146664e-05,
+      "loss": 1.642,
+      "step": 21190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3188324722761525e-05,
+      "loss": 1.7058,
+      "step": 21200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3185034058376386e-05,
+      "loss": 1.5491,
+      "step": 21210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.318174339399125e-05,
+      "loss": 1.6613,
+      "step": 21220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.317845272960611e-05,
+      "loss": 1.6887,
+      "step": 21230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.317516206522097e-05,
+      "loss": 1.8651,
+      "step": 21240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.317187140083583e-05,
+      "loss": 1.6345,
+      "step": 21250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.316858073645069e-05,
+      "loss": 1.6605,
+      "step": 21260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3165290072065554e-05,
+      "loss": 1.6728,
+      "step": 21270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3161999407680415e-05,
+      "loss": 1.497,
+      "step": 21280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3158708743295276e-05,
+      "loss": 1.5975,
+      "step": 21290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.315541807891013e-05,
+      "loss": 1.7015,
+      "step": 21300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.315212741452499e-05,
+      "loss": 1.6415,
+      "step": 21310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3148836750139853e-05,
+      "loss": 1.7422,
+      "step": 21320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3145546085754715e-05,
+      "loss": 1.7843,
+      "step": 21330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3142255421369576e-05,
+      "loss": 1.8029,
+      "step": 21340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.313896475698444e-05,
+      "loss": 1.8271,
+      "step": 21350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.31356740925993e-05,
+      "loss": 1.6271,
+      "step": 21360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.313238342821416e-05,
+      "loss": 1.7852,
+      "step": 21370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.312909276382902e-05,
+      "loss": 1.6899,
+      "step": 21380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.312580209944388e-05,
+      "loss": 1.6342,
+      "step": 21390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.312251143505874e-05,
+      "loss": 1.6123,
+      "step": 21400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3119220770673605e-05,
+      "loss": 1.6794,
+      "step": 21410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.311593010628846e-05,
+      "loss": 1.7917,
+      "step": 21420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.311263944190333e-05,
+      "loss": 1.5767,
+      "step": 21430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310934877751818e-05,
+      "loss": 1.7355,
+      "step": 21440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310605811313304e-05,
+      "loss": 1.8447,
+      "step": 21450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3102767448747904e-05,
+      "loss": 1.7065,
+      "step": 21460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3099476784362766e-05,
+      "loss": 1.7318,
+      "step": 21470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309618611997763e-05,
+      "loss": 1.7295,
+      "step": 21480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309289545559249e-05,
+      "loss": 1.738,
+      "step": 21490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308960479120734e-05,
+      "loss": 1.6351,
+      "step": 21500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308631412682221e-05,
+      "loss": 1.7347,
+      "step": 21510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3083023462437065e-05,
+      "loss": 1.6911,
+      "step": 21520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307973279805193e-05,
+      "loss": 1.642,
+      "step": 21530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307644213366679e-05,
+      "loss": 1.496,
+      "step": 21540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307315146928165e-05,
+      "loss": 1.5676,
+      "step": 21550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306986080489651e-05,
+      "loss": 1.7222,
+      "step": 21560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306657014051137e-05,
+      "loss": 1.7566,
+      "step": 21570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306327947612623e-05,
+      "loss": 1.8237,
+      "step": 21580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3059988811741094e-05,
+      "loss": 1.5914,
+      "step": 21590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.305669814735595e-05,
+      "loss": 1.844,
+      "step": 21600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.305340748297082e-05,
+      "loss": 1.7137,
+      "step": 21610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.305011681858567e-05,
+      "loss": 1.7593,
+      "step": 21620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.304682615420054e-05,
+      "loss": 1.8924,
+      "step": 21630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3043535489815394e-05,
+      "loss": 1.9924,
+      "step": 21640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3040244825430255e-05,
+      "loss": 1.7257,
+      "step": 21650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3036954161045116e-05,
+      "loss": 1.8378,
+      "step": 21660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.303366349665998e-05,
+      "loss": 1.6857,
+      "step": 21670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.303037283227484e-05,
+      "loss": 1.7215,
+      "step": 21680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.30270821678897e-05,
+      "loss": 1.6835,
+      "step": 21690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3023791503504554e-05,
+      "loss": 1.6788,
+      "step": 21700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.302050083911942e-05,
+      "loss": 1.6051,
+      "step": 21710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.301721017473428e-05,
+      "loss": 1.6131,
+      "step": 21720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3013919510349145e-05,
+      "loss": 1.7712,
+      "step": 21730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3010628845964e-05,
+      "loss": 1.6338,
+      "step": 21740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.300733818157886e-05,
+      "loss": 1.848,
+      "step": 21750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.300404751719372e-05,
+      "loss": 1.9499,
+      "step": 21760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.300075685280858e-05,
+      "loss": 1.6612,
+      "step": 21770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2997466188423445e-05,
+      "loss": 1.6316,
+      "step": 21780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2994175524038306e-05,
+      "loss": 1.6015,
+      "step": 21790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.299088485965316e-05,
+      "loss": 1.7484,
+      "step": 21800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.298759419526803e-05,
+      "loss": 1.7572,
+      "step": 21810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.298430353088288e-05,
+      "loss": 1.6437,
+      "step": 21820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.298101286649775e-05,
+      "loss": 1.6726,
+      "step": 21830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2977722202112605e-05,
+      "loss": 1.5781,
+      "step": 21840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2974431537727473e-05,
+      "loss": 1.7777,
+      "step": 21850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.297114087334233e-05,
+      "loss": 1.7785,
+      "step": 21860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.296785020895719e-05,
+      "loss": 1.8858,
+      "step": 21870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.296455954457205e-05,
+      "loss": 1.6461,
+      "step": 21880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.296126888018691e-05,
+      "loss": 1.931,
+      "step": 21890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.295797821580177e-05,
+      "loss": 1.9868,
+      "step": 21900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2954687551416634e-05,
+      "loss": 1.7062,
+      "step": 21910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.295139688703149e-05,
+      "loss": 1.6942,
+      "step": 21920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.294810622264636e-05,
+      "loss": 1.6477,
+      "step": 21930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.294481555826121e-05,
+      "loss": 1.5883,
+      "step": 21940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.294152489387608e-05,
+      "loss": 1.6525,
+      "step": 21950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2938234229490934e-05,
+      "loss": 1.6579,
+      "step": 21960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2934943565105795e-05,
+      "loss": 1.5387,
+      "step": 21970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2931652900720656e-05,
+      "loss": 1.6148,
+      "step": 21980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292836223633552e-05,
+      "loss": 1.7242,
+      "step": 21990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292507157195038e-05,
+      "loss": 1.5663,
+      "step": 22000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292178090756524e-05,
+      "loss": 1.5046,
+      "step": 22010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29184902431801e-05,
+      "loss": 1.7802,
+      "step": 22020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291519957879496e-05,
+      "loss": 1.6722,
+      "step": 22030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2911908914409824e-05,
+      "loss": 1.7801,
+      "step": 22040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2908618250024685e-05,
+      "loss": 1.8791,
+      "step": 22050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290532758563954e-05,
+      "loss": 1.9085,
+      "step": 22060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29020369212544e-05,
+      "loss": 1.9821,
+      "step": 22070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289874625686926e-05,
+      "loss": 1.8121,
+      "step": 22080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2895455592484124e-05,
+      "loss": 1.9264,
+      "step": 22090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2892164928098985e-05,
+      "loss": 1.8309,
+      "step": 22100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2888874263713846e-05,
+      "loss": 1.8314,
+      "step": 22110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.288558359932871e-05,
+      "loss": 1.7494,
+      "step": 22120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.288229293494357e-05,
+      "loss": 1.6883,
+      "step": 22130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.287900227055843e-05,
+      "loss": 1.7343,
+      "step": 22140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.287571160617329e-05,
+      "loss": 1.7265,
+      "step": 22150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.287242094178815e-05,
+      "loss": 1.5531,
+      "step": 22160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.286913027740301e-05,
+      "loss": 1.7262,
+      "step": 22170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2865839613017875e-05,
+      "loss": 1.6452,
+      "step": 22180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.286254894863273e-05,
+      "loss": 1.7804,
+      "step": 22190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.28592582842476e-05,
+      "loss": 1.8233,
+      "step": 22200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.285596761986245e-05,
+      "loss": 2.0238,
+      "step": 22210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.285267695547731e-05,
+      "loss": 1.8522,
+      "step": 22220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2849386291092175e-05,
+      "loss": 2.2385,
+      "step": 22230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2846095626707036e-05,
+      "loss": 1.5751,
+      "step": 22240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.28428049623219e-05,
+      "loss": 1.7847,
+      "step": 22250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.283951429793676e-05,
+      "loss": 1.7108,
+      "step": 22260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.283622363355161e-05,
+      "loss": 1.7528,
+      "step": 22270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.283293296916648e-05,
+      "loss": 1.6502,
+      "step": 22280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2829642304781335e-05,
+      "loss": 1.5635,
+      "step": 22290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2826351640396203e-05,
+      "loss": 1.5919,
+      "step": 22300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.282306097601106e-05,
+      "loss": 1.6594,
+      "step": 22310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.281977031162592e-05,
+      "loss": 1.694,
+      "step": 22320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.281647964724078e-05,
+      "loss": 1.8306,
+      "step": 22330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.281318898285564e-05,
+      "loss": 1.7038,
+      "step": 22340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.28098983184705e-05,
+      "loss": 1.8113,
+      "step": 22350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2806607654085364e-05,
+      "loss": 1.7071,
+      "step": 22360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.280331698970022e-05,
+      "loss": 1.6348,
+      "step": 22370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.280002632531509e-05,
+      "loss": 1.7318,
+      "step": 22380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.279673566092994e-05,
+      "loss": 1.6908,
+      "step": 22390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.279344499654481e-05,
+      "loss": 1.7616,
+      "step": 22400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2790154332159664e-05,
+      "loss": 1.773,
+      "step": 22410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2786863667774525e-05,
+      "loss": 1.5924,
+      "step": 22420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2783573003389386e-05,
+      "loss": 1.7628,
+      "step": 22430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.278028233900425e-05,
+      "loss": 1.6535,
+      "step": 22440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277699167461911e-05,
+      "loss": 1.6601,
+      "step": 22450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277370101023397e-05,
+      "loss": 1.7209,
+      "step": 22460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2770410345848825e-05,
+      "loss": 1.735,
+      "step": 22470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276711968146369e-05,
+      "loss": 1.6781,
+      "step": 22480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276382901707855e-05,
+      "loss": 1.7893,
+      "step": 22490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2760538352693415e-05,
+      "loss": 1.5794,
+      "step": 22500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275724768830827e-05,
+      "loss": 1.6888,
+      "step": 22510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275395702392313e-05,
+      "loss": 1.7338,
+      "step": 22520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275066635953799e-05,
+      "loss": 1.732,
+      "step": 22530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2747375695152854e-05,
+      "loss": 1.6511,
+      "step": 22540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2744085030767715e-05,
+      "loss": 1.7725,
+      "step": 22550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2740794366382576e-05,
+      "loss": 1.8406,
+      "step": 22560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273750370199743e-05,
+      "loss": 1.704,
+      "step": 22570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27342130376123e-05,
+      "loss": 1.6598,
+      "step": 22580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273092237322715e-05,
+      "loss": 1.7784,
+      "step": 22590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272763170884202e-05,
+      "loss": 1.7126,
+      "step": 22600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2724341044456876e-05,
+      "loss": 1.5912,
+      "step": 22610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.272105038007174e-05,
+      "loss": 1.8047,
+      "step": 22620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.27177597156866e-05,
+      "loss": 1.6651,
+      "step": 22630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.271446905130146e-05,
+      "loss": 1.6359,
+      "step": 22640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.271117838691632e-05,
+      "loss": 1.726,
+      "step": 22650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.270788772253118e-05,
+      "loss": 1.7298,
+      "step": 22660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2704597058146036e-05,
+      "loss": 1.8135,
+      "step": 22670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2701306393760904e-05,
+      "loss": 1.7244,
+      "step": 22680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.269801572937576e-05,
+      "loss": 1.6156,
+      "step": 22690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.269472506499063e-05,
+      "loss": 1.6439,
+      "step": 22700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.269143440060548e-05,
+      "loss": 1.8636,
+      "step": 22710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.268814373622034e-05,
+      "loss": 1.6496,
+      "step": 22720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2684853071835204e-05,
+      "loss": 1.6926,
+      "step": 22730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2681562407450065e-05,
+      "loss": 1.6096,
+      "step": 22740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2678271743064927e-05,
+      "loss": 1.6202,
+      "step": 22750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.267498107867979e-05,
+      "loss": 1.6713,
+      "step": 22760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.267169041429465e-05,
+      "loss": 1.6496,
+      "step": 22770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.266839974990951e-05,
+      "loss": 1.7181,
+      "step": 22780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.266510908552437e-05,
+      "loss": 1.655,
+      "step": 22790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.266181842113923e-05,
+      "loss": 1.636,
+      "step": 22800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2658527756754094e-05,
+      "loss": 1.6893,
+      "step": 22810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.265523709236895e-05,
+      "loss": 1.6209,
+      "step": 22820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.265194642798381e-05,
+      "loss": 1.5864,
+      "step": 22830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.264865576359867e-05,
+      "loss": 1.6747,
+      "step": 22840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.264536509921353e-05,
+      "loss": 1.7829,
+      "step": 22850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2642074434828394e-05,
+      "loss": 1.7843,
+      "step": 22860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2638783770443255e-05,
+      "loss": 1.6913,
+      "step": 22870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2635493106058116e-05,
+      "loss": 1.8728,
+      "step": 22880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.263220244167298e-05,
+      "loss": 1.6963,
+      "step": 22890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.262891177728784e-05,
+      "loss": 1.8285,
+      "step": 22900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.26256211129027e-05,
+      "loss": 1.6823,
+      "step": 22910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2622330448517555e-05,
+      "loss": 1.805,
+      "step": 22920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.261903978413242e-05,
+      "loss": 1.6808,
+      "step": 22930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.261574911974728e-05,
+      "loss": 1.7221,
+      "step": 22940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2612458455362145e-05,
+      "loss": 1.6664,
+      "step": 22950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2609167790977e-05,
+      "loss": 1.8371,
+      "step": 22960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260587712659186e-05,
+      "loss": 1.6513,
+      "step": 22970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260258646220672e-05,
+      "loss": 1.5952,
+      "step": 22980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2599295797821583e-05,
+      "loss": 1.7777,
+      "step": 22990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2596005133436445e-05,
+      "loss": 1.7333,
+      "step": 23000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2592714469051306e-05,
+      "loss": 1.6972,
+      "step": 23010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258942380466616e-05,
+      "loss": 1.6171,
+      "step": 23020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258613314028103e-05,
+      "loss": 1.7025,
+      "step": 23030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258284247589588e-05,
+      "loss": 1.6056,
+      "step": 23040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257955181151075e-05,
+      "loss": 1.6771,
+      "step": 23050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2576261147125606e-05,
+      "loss": 1.6701,
+      "step": 23060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257297048274047e-05,
+      "loss": 1.6234,
+      "step": 23070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256967981835533e-05,
+      "loss": 1.5778,
+      "step": 23080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256638915397019e-05,
+      "loss": 1.8765,
+      "step": 23090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256309848958505e-05,
+      "loss": 1.4866,
+      "step": 23100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255980782519991e-05,
+      "loss": 1.7806,
+      "step": 23110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2556517160814766e-05,
+      "loss": 1.7059,
+      "step": 23120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2553226496429634e-05,
+      "loss": 1.7675,
+      "step": 23130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.254993583204449e-05,
+      "loss": 1.6895,
+      "step": 23140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.254664516765936e-05,
+      "loss": 1.8446,
+      "step": 23150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.254335450327421e-05,
+      "loss": 1.7243,
+      "step": 23160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.254006383888907e-05,
+      "loss": 1.7267,
+      "step": 23170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2536773174503934e-05,
+      "loss": 1.7087,
+      "step": 23180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2533482510118795e-05,
+      "loss": 1.6049,
+      "step": 23190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2530191845733657e-05,
+      "loss": 1.7012,
+      "step": 23200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.252690118134852e-05,
+      "loss": 1.6425,
+      "step": 23210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.252361051696337e-05,
+      "loss": 1.6369,
+      "step": 23220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.252031985257824e-05,
+      "loss": 1.7177,
+      "step": 23230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2517029188193095e-05,
+      "loss": 1.7227,
+      "step": 23240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.251373852380796e-05,
+      "loss": 1.4892,
+      "step": 23250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.251044785942282e-05,
+      "loss": 1.5936,
+      "step": 23260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.250715719503768e-05,
+      "loss": 1.6537,
+      "step": 23270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.250386653065254e-05,
+      "loss": 1.8002,
+      "step": 23280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.25005758662674e-05,
+      "loss": 1.8045,
+      "step": 23290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.249728520188226e-05,
+      "loss": 1.6057,
+      "step": 23300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2493994537497124e-05,
+      "loss": 1.6019,
+      "step": 23310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.249070387311198e-05,
+      "loss": 1.6584,
+      "step": 23320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2487413208726846e-05,
+      "loss": 1.5271,
+      "step": 23330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24841225443417e-05,
+      "loss": 1.6008,
+      "step": 23340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.248083187995657e-05,
+      "loss": 1.7168,
+      "step": 23350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.247754121557142e-05,
+      "loss": 1.7074,
+      "step": 23360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2474250551186285e-05,
+      "loss": 1.6589,
+      "step": 23370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2470959886801146e-05,
+      "loss": 2.0956,
+      "step": 23380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.246766922241601e-05,
+      "loss": 1.7943,
+      "step": 23390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.246437855803087e-05,
+      "loss": 1.7001,
+      "step": 23400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.246108789364573e-05,
+      "loss": 1.7901,
+      "step": 23410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2457797229260584e-05,
+      "loss": 1.8076,
+      "step": 23420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.245450656487545e-05,
+      "loss": 1.7779,
+      "step": 23430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2451215900490307e-05,
+      "loss": 1.8406,
+      "step": 23440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2447925236105175e-05,
+      "loss": 1.7165,
+      "step": 23450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.244463457172003e-05,
+      "loss": 1.7313,
+      "step": 23460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.244134390733489e-05,
+      "loss": 1.5793,
+      "step": 23470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243805324294975e-05,
+      "loss": 1.7668,
+      "step": 23480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243476257856461e-05,
+      "loss": 1.5975,
+      "step": 23490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2431471914179474e-05,
+      "loss": 1.6202,
+      "step": 23500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2428181249794335e-05,
+      "loss": 1.6153,
+      "step": 23510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24248905854092e-05,
+      "loss": 1.8033,
+      "step": 23520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242159992102406e-05,
+      "loss": 1.7528,
+      "step": 23530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241830925663892e-05,
+      "loss": 1.6662,
+      "step": 23540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241501859225378e-05,
+      "loss": 1.7386,
+      "step": 23550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241172792786864e-05,
+      "loss": 1.7499,
+      "step": 23560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2408437263483496e-05,
+      "loss": 1.8489,
+      "step": 23570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240514659909836e-05,
+      "loss": 1.7604,
+      "step": 23580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240185593471322e-05,
+      "loss": 1.6546,
+      "step": 23590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239856527032808e-05,
+      "loss": 1.8291,
+      "step": 23600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239527460594294e-05,
+      "loss": 1.6172,
+      "step": 23610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23919839415578e-05,
+      "loss": 1.6188,
+      "step": 23620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2388693277172664e-05,
+      "loss": 1.8087,
+      "step": 23630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2385402612787525e-05,
+      "loss": 1.6877,
+      "step": 23640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2382111948402386e-05,
+      "loss": 1.6918,
+      "step": 23650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.237882128401725e-05,
+      "loss": 1.7,
+      "step": 23660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.23755306196321e-05,
+      "loss": 1.7466,
+      "step": 23670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.237223995524697e-05,
+      "loss": 1.6294,
+      "step": 23680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2368949290861825e-05,
+      "loss": 1.7345,
+      "step": 23690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.236565862647669e-05,
+      "loss": 1.7198,
+      "step": 23700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.236236796209155e-05,
+      "loss": 1.7474,
+      "step": 23710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.235907729770641e-05,
+      "loss": 1.7753,
+      "step": 23720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.235578663332127e-05,
+      "loss": 1.632,
+      "step": 23730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.235249596893613e-05,
+      "loss": 1.7922,
+      "step": 23740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.234920530455099e-05,
+      "loss": 1.747,
+      "step": 23750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2345914640165854e-05,
+      "loss": 1.5631,
+      "step": 23760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.234262397578071e-05,
+      "loss": 1.7279,
+      "step": 23770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2339333311395576e-05,
+      "loss": 1.7273,
+      "step": 23780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.233604264701043e-05,
+      "loss": 1.6271,
+      "step": 23790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.23327519826253e-05,
+      "loss": 1.6479,
+      "step": 23800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.232946131824015e-05,
+      "loss": 1.72,
+      "step": 23810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2326170653855014e-05,
+      "loss": 1.775,
+      "step": 23820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2322879989469876e-05,
+      "loss": 1.6904,
+      "step": 23830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.231958932508474e-05,
+      "loss": 1.6307,
+      "step": 23840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.23162986606996e-05,
+      "loss": 1.7889,
+      "step": 23850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.231300799631446e-05,
+      "loss": 1.8437,
+      "step": 23860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2309717331929314e-05,
+      "loss": 1.9238,
+      "step": 23870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.230642666754418e-05,
+      "loss": 1.6758,
+      "step": 23880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2303136003159037e-05,
+      "loss": 1.6121,
+      "step": 23890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2299845338773905e-05,
+      "loss": 1.6949,
+      "step": 23900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.229655467438876e-05,
+      "loss": 1.5969,
+      "step": 23910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.229326401000362e-05,
+      "loss": 1.7954,
+      "step": 23920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.228997334561848e-05,
+      "loss": 1.6102,
+      "step": 23930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.228668268123334e-05,
+      "loss": 1.7921,
+      "step": 23940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2283392016848204e-05,
+      "loss": 1.6199,
+      "step": 23950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2280101352463065e-05,
+      "loss": 1.6404,
+      "step": 23960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227681068807792e-05,
+      "loss": 1.6632,
+      "step": 23970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227352002369279e-05,
+      "loss": 1.7355,
+      "step": 23980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227022935930764e-05,
+      "loss": 1.6519,
+      "step": 23990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226693869492251e-05,
+      "loss": 1.7318,
+      "step": 24000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2263648030537365e-05,
+      "loss": 1.6495,
+      "step": 24010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2260357366152226e-05,
+      "loss": 1.8225,
+      "step": 24020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225706670176709e-05,
+      "loss": 1.6588,
+      "step": 24030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225377603738195e-05,
+      "loss": 1.8015,
+      "step": 24040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225048537299681e-05,
+      "loss": 1.6602,
+      "step": 24050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224719470861167e-05,
+      "loss": 1.5242,
+      "step": 24060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2243904044226526e-05,
+      "loss": 1.7222,
+      "step": 24070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2240613379841394e-05,
+      "loss": 1.7758,
+      "step": 24080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223732271545625e-05,
+      "loss": 1.5825,
+      "step": 24090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2234032051071116e-05,
+      "loss": 1.6797,
+      "step": 24100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223074138668597e-05,
+      "loss": 1.804,
+      "step": 24110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222745072230083e-05,
+      "loss": 1.8783,
+      "step": 24120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2224160057915693e-05,
+      "loss": 1.8072,
+      "step": 24130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2220869393530555e-05,
+      "loss": 1.6419,
+      "step": 24140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2217578729145416e-05,
+      "loss": 1.6279,
+      "step": 24150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.221428806476028e-05,
+      "loss": 1.768,
+      "step": 24160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.221099740037514e-05,
+      "loss": 1.6431,
+      "step": 24170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.220770673599e-05,
+      "loss": 1.6244,
+      "step": 24180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2204416071604854e-05,
+      "loss": 1.6685,
+      "step": 24190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.220112540721972e-05,
+      "loss": 1.665,
+      "step": 24200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.219783474283458e-05,
+      "loss": 1.5724,
+      "step": 24210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.219454407844944e-05,
+      "loss": 1.6525,
+      "step": 24220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.21912534140643e-05,
+      "loss": 1.6452,
+      "step": 24230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.218796274967916e-05,
+      "loss": 1.5676,
+      "step": 24240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.218467208529402e-05,
+      "loss": 1.6928,
+      "step": 24250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.218138142090888e-05,
+      "loss": 1.6561,
+      "step": 24260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2178090756523744e-05,
+      "loss": 1.4343,
+      "step": 24270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2174800092138606e-05,
+      "loss": 1.8648,
+      "step": 24280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.217150942775347e-05,
+      "loss": 1.6648,
+      "step": 24290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.216821876336833e-05,
+      "loss": 2.1325,
+      "step": 24300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.216492809898319e-05,
+      "loss": 1.6582,
+      "step": 24310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2161637434598044e-05,
+      "loss": 1.6441,
+      "step": 24320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.215834677021291e-05,
+      "loss": 1.7461,
+      "step": 24330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2155056105827766e-05,
+      "loss": 1.6581,
+      "step": 24340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.215176544144263e-05,
+      "loss": 1.5546,
+      "step": 24350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214847477705749e-05,
+      "loss": 1.5865,
+      "step": 24360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214518411267235e-05,
+      "loss": 1.533,
+      "step": 24370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214189344828721e-05,
+      "loss": 1.88,
+      "step": 24380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.213860278390207e-05,
+      "loss": 1.8424,
+      "step": 24390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2135312119516934e-05,
+      "loss": 1.9128,
+      "step": 24400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2132021455131795e-05,
+      "loss": 1.7117,
+      "step": 24410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.212873079074665e-05,
+      "loss": 1.8082,
+      "step": 24420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.212544012636152e-05,
+      "loss": 1.737,
+      "step": 24430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.212214946197637e-05,
+      "loss": 1.6944,
+      "step": 24440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.211885879759124e-05,
+      "loss": 1.6518,
+      "step": 24450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2115568133206095e-05,
+      "loss": 1.739,
+      "step": 24460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2112277468820956e-05,
+      "loss": 1.6797,
+      "step": 24470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210898680443582e-05,
+      "loss": 1.5898,
+      "step": 24480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210569614005068e-05,
+      "loss": 1.8503,
+      "step": 24490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210240547566554e-05,
+      "loss": 1.7679,
+      "step": 24500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20991148112804e-05,
+      "loss": 1.6266,
+      "step": 24510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2095824146895256e-05,
+      "loss": 1.5677,
+      "step": 24520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2092533482510124e-05,
+      "loss": 1.7126,
+      "step": 24530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208924281812498e-05,
+      "loss": 1.7006,
+      "step": 24540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2085952153739846e-05,
+      "loss": 1.7259,
+      "step": 24550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20826614893547e-05,
+      "loss": 1.8085,
+      "step": 24560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207937082496956e-05,
+      "loss": 1.5618,
+      "step": 24570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207608016058442e-05,
+      "loss": 1.706,
+      "step": 24580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2072789496199285e-05,
+      "loss": 1.9261,
+      "step": 24590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2069498831814146e-05,
+      "loss": 1.9139,
+      "step": 24600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206620816742901e-05,
+      "loss": 1.6303,
+      "step": 24610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206291750304386e-05,
+      "loss": 1.6859,
+      "step": 24620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205962683865873e-05,
+      "loss": 1.7848,
+      "step": 24630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2056336174273584e-05,
+      "loss": 1.7827,
+      "step": 24640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.205304550988845e-05,
+      "loss": 1.7762,
+      "step": 24650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.204975484550331e-05,
+      "loss": 1.7412,
+      "step": 24660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2046464181118175e-05,
+      "loss": 1.6813,
+      "step": 24670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.204317351673303e-05,
+      "loss": 1.5657,
+      "step": 24680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.203988285234789e-05,
+      "loss": 1.7438,
+      "step": 24690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.203659218796275e-05,
+      "loss": 1.6032,
+      "step": 24700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.203330152357761e-05,
+      "loss": 1.8292,
+      "step": 24710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2030010859192474e-05,
+      "loss": 1.6315,
+      "step": 24720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2026720194807336e-05,
+      "loss": 1.6929,
+      "step": 24730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.202342953042219e-05,
+      "loss": 1.5211,
+      "step": 24740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.202013886603706e-05,
+      "loss": 1.5687,
+      "step": 24750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.201684820165191e-05,
+      "loss": 1.5711,
+      "step": 24760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.201355753726678e-05,
+      "loss": 1.6698,
+      "step": 24770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2010266872881635e-05,
+      "loss": 2.0628,
+      "step": 24780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2006976208496496e-05,
+      "loss": 1.8031,
+      "step": 24790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.200368554411136e-05,
+      "loss": 1.5922,
+      "step": 24800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.200039487972622e-05,
+      "loss": 1.7146,
+      "step": 24810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.199710421534108e-05,
+      "loss": 2.0316,
+      "step": 24820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.199381355095594e-05,
+      "loss": 1.6692,
+      "step": 24830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1990522886570796e-05,
+      "loss": 1.5527,
+      "step": 24840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1987232222185664e-05,
+      "loss": 1.6933,
+      "step": 24850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.198394155780052e-05,
+      "loss": 1.8652,
+      "step": 24860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1980650893415387e-05,
+      "loss": 1.5897,
+      "step": 24870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.197736022903024e-05,
+      "loss": 1.883,
+      "step": 24880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19740695646451e-05,
+      "loss": 1.7307,
+      "step": 24890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1970778900259964e-05,
+      "loss": 1.7877,
+      "step": 24900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1967488235874825e-05,
+      "loss": 1.6508,
+      "step": 24910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1964197571489686e-05,
+      "loss": 1.771,
+      "step": 24920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.196090690710455e-05,
+      "loss": 1.7079,
+      "step": 24930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.195761624271941e-05,
+      "loss": 1.686,
+      "step": 24940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.195432557833427e-05,
+      "loss": 1.6483,
+      "step": 24950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1951034913949124e-05,
+      "loss": 1.7994,
+      "step": 24960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194774424956399e-05,
+      "loss": 1.7206,
+      "step": 24970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194445358517885e-05,
+      "loss": 1.6564,
+      "step": 24980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194116292079371e-05,
+      "loss": 1.6274,
+      "step": 24990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193787225640857e-05,
+      "loss": 1.7677,
+      "step": 25000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193458159202343e-05,
+      "loss": 1.7483,
+      "step": 25010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193129092763829e-05,
+      "loss": 1.8273,
+      "step": 25020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192800026325315e-05,
+      "loss": 1.7594,
+      "step": 25030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1924709598868015e-05,
+      "loss": 1.5885,
+      "step": 25040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1921418934482876e-05,
+      "loss": 1.6863,
+      "step": 25050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191812827009774e-05,
+      "loss": 1.8214,
+      "step": 25060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19148376057126e-05,
+      "loss": 1.6693,
+      "step": 25070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191154694132746e-05,
+      "loss": 1.6233,
+      "step": 25080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1908256276942314e-05,
+      "loss": 1.8536,
+      "step": 25090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190496561255718e-05,
+      "loss": 1.6248,
+      "step": 25100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190167494817204e-05,
+      "loss": 1.6445,
+      "step": 25110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18983842837869e-05,
+      "loss": 1.8122,
+      "step": 25120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189509361940176e-05,
+      "loss": 1.6464,
+      "step": 25130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189180295501662e-05,
+      "loss": 1.7563,
+      "step": 25140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188851229063148e-05,
+      "loss": 1.7098,
+      "step": 25150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.188522162624634e-05,
+      "loss": 1.5649,
+      "step": 25160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1881930961861204e-05,
+      "loss": 1.844,
+      "step": 25170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1878640297476065e-05,
+      "loss": 1.5428,
+      "step": 25180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.187534963309092e-05,
+      "loss": 1.52,
+      "step": 25190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.187205896870579e-05,
+      "loss": 1.6822,
+      "step": 25200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.186876830432064e-05,
+      "loss": 1.7633,
+      "step": 25210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.186547763993551e-05,
+      "loss": 1.6711,
+      "step": 25220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1862186975550365e-05,
+      "loss": 1.709,
+      "step": 25230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1858896311165226e-05,
+      "loss": 1.5709,
+      "step": 25240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.185560564678009e-05,
+      "loss": 1.719,
+      "step": 25250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.185231498239495e-05,
+      "loss": 1.7695,
+      "step": 25260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.184902431800981e-05,
+      "loss": 1.6602,
+      "step": 25270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.184573365362467e-05,
+      "loss": 1.7718,
+      "step": 25280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1842442989239526e-05,
+      "loss": 1.6971,
+      "step": 25290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1839152324854394e-05,
+      "loss": 1.7797,
+      "step": 25300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.183586166046925e-05,
+      "loss": 1.5544,
+      "step": 25310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1832570996084116e-05,
+      "loss": 1.4836,
+      "step": 25320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.182928033169897e-05,
+      "loss": 1.8578,
+      "step": 25330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.182598966731383e-05,
+      "loss": 1.6776,
+      "step": 25340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1822699002928693e-05,
+      "loss": 1.6004,
+      "step": 25350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1819408338543555e-05,
+      "loss": 1.7141,
+      "step": 25360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1816117674158416e-05,
+      "loss": 1.6938,
+      "step": 25370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.181282700977328e-05,
+      "loss": 1.8472,
+      "step": 25380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.180953634538813e-05,
+      "loss": 1.8991,
+      "step": 25390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1806245681003e-05,
+      "loss": 1.6883,
+      "step": 25400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1802955016617854e-05,
+      "loss": 1.6922,
+      "step": 25410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.179966435223272e-05,
+      "loss": 1.7641,
+      "step": 25420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.179637368784758e-05,
+      "loss": 1.696,
+      "step": 25430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.179308302346244e-05,
+      "loss": 1.6572,
+      "step": 25440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17897923590773e-05,
+      "loss": 1.835,
+      "step": 25450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.178650169469216e-05,
+      "loss": 1.4988,
+      "step": 25460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.178321103030702e-05,
+      "loss": 1.665,
+      "step": 25470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177992036592188e-05,
+      "loss": 1.8012,
+      "step": 25480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177662970153674e-05,
+      "loss": 1.6445,
+      "step": 25490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1773339037151606e-05,
+      "loss": 1.573,
+      "step": 25500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177004837276646e-05,
+      "loss": 1.8751,
+      "step": 25510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176675770838133e-05,
+      "loss": 1.7303,
+      "step": 25520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176346704399618e-05,
+      "loss": 1.7605,
+      "step": 25530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1760176379611044e-05,
+      "loss": 1.8404,
+      "step": 25540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1756885715225905e-05,
+      "loss": 1.805,
+      "step": 25550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1753595050840767e-05,
+      "loss": 1.6663,
+      "step": 25560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175030438645563e-05,
+      "loss": 1.6735,
+      "step": 25570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174701372207049e-05,
+      "loss": 1.7088,
+      "step": 25580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1743723057685344e-05,
+      "loss": 1.824,
+      "step": 25590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174043239330021e-05,
+      "loss": 1.7051,
+      "step": 25600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1737141728915066e-05,
+      "loss": 1.5874,
+      "step": 25610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1733851064529934e-05,
+      "loss": 1.6796,
+      "step": 25620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173056040014479e-05,
+      "loss": 1.6644,
+      "step": 25630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172726973575965e-05,
+      "loss": 1.594,
+      "step": 25640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172397907137451e-05,
+      "loss": 1.6459,
+      "step": 25650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172068840698937e-05,
+      "loss": 1.6396,
+      "step": 25660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1717397742604234e-05,
+      "loss": 1.6198,
+      "step": 25670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1714107078219095e-05,
+      "loss": 1.9808,
+      "step": 25680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1710816413833956e-05,
+      "loss": 1.8325,
+      "step": 25690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.170752574944882e-05,
+      "loss": 1.7876,
+      "step": 25700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.170423508506367e-05,
+      "loss": 1.72,
+      "step": 25710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.170094442067854e-05,
+      "loss": 1.7142,
+      "step": 25720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1697653756293395e-05,
+      "loss": 1.8107,
+      "step": 25730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1694363091908256e-05,
+      "loss": 1.9219,
+      "step": 25740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.169107242752312e-05,
+      "loss": 1.6811,
+      "step": 25750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.168778176313798e-05,
+      "loss": 1.7011,
+      "step": 25760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.168449109875284e-05,
+      "loss": 1.7245,
+      "step": 25770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.16812004343677e-05,
+      "loss": 1.6316,
+      "step": 25780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.167790976998256e-05,
+      "loss": 1.7323,
+      "step": 25790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1674619105597423e-05,
+      "loss": 1.843,
+      "step": 25800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1671328441212285e-05,
+      "loss": 1.7388,
+      "step": 25810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1668037776827146e-05,
+      "loss": 1.6475,
+      "step": 25820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.166474711244201e-05,
+      "loss": 2.1498,
+      "step": 25830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.166145644805686e-05,
+      "loss": 1.6304,
+      "step": 25840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.165816578367173e-05,
+      "loss": 1.779,
+      "step": 25850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1654875119286584e-05,
+      "loss": 1.6998,
+      "step": 25860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1651584454901446e-05,
+      "loss": 1.7622,
+      "step": 25870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.164829379051631e-05,
+      "loss": 1.8166,
+      "step": 25880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.164500312613117e-05,
+      "loss": 1.6529,
+      "step": 25890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.164171246174603e-05,
+      "loss": 1.8202,
+      "step": 25900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.163842179736089e-05,
+      "loss": 1.6753,
+      "step": 25910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.163513113297575e-05,
+      "loss": 1.7157,
+      "step": 25920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.163184046859061e-05,
+      "loss": 1.8153,
+      "step": 25930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.162854980420547e-05,
+      "loss": 1.6476,
+      "step": 25940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1625259139820336e-05,
+      "loss": 1.5632,
+      "step": 25950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.162196847543519e-05,
+      "loss": 1.7376,
+      "step": 25960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.161867781105006e-05,
+      "loss": 1.5841,
+      "step": 25970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.161538714666491e-05,
+      "loss": 1.7792,
+      "step": 25980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1612096482279774e-05,
+      "loss": 1.6899,
+      "step": 25990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1608805817894635e-05,
+      "loss": 1.6325,
+      "step": 26000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1605515153509496e-05,
+      "loss": 1.6991,
+      "step": 26010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160222448912436e-05,
+      "loss": 1.6769,
+      "step": 26020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159893382473922e-05,
+      "loss": 1.7971,
+      "step": 26030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1595643160354074e-05,
+      "loss": 1.5831,
+      "step": 26040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159235249596894e-05,
+      "loss": 1.7775,
+      "step": 26050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1589061831583796e-05,
+      "loss": 1.8781,
+      "step": 26060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1585771167198664e-05,
+      "loss": 1.8134,
+      "step": 26070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158248050281352e-05,
+      "loss": 1.6958,
+      "step": 26080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157918983842838e-05,
+      "loss": 1.6172,
+      "step": 26090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157589917404324e-05,
+      "loss": 1.7117,
+      "step": 26100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15726085096581e-05,
+      "loss": 1.5456,
+      "step": 26110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1569317845272964e-05,
+      "loss": 1.603,
+      "step": 26120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1566027180887825e-05,
+      "loss": 1.8173,
+      "step": 26130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156273651650268e-05,
+      "loss": 1.725,
+      "step": 26140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155944585211755e-05,
+      "loss": 1.5643,
+      "step": 26150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15561551877324e-05,
+      "loss": 1.6586,
+      "step": 26160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.155286452334727e-05,
+      "loss": 1.6601,
+      "step": 26170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1549573858962124e-05,
+      "loss": 1.5216,
+      "step": 26180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1546283194576986e-05,
+      "loss": 1.6452,
+      "step": 26190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.154299253019185e-05,
+      "loss": 1.6798,
+      "step": 26200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.153970186580671e-05,
+      "loss": 1.6714,
+      "step": 26210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.153641120142157e-05,
+      "loss": 2.0014,
+      "step": 26220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.153312053703643e-05,
+      "loss": 1.7061,
+      "step": 26230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1529829872651285e-05,
+      "loss": 1.7642,
+      "step": 26240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.152653920826615e-05,
+      "loss": 1.5704,
+      "step": 26250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.152324854388101e-05,
+      "loss": 1.7868,
+      "step": 26260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1519957879495876e-05,
+      "loss": 1.7078,
+      "step": 26270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.151666721511073e-05,
+      "loss": 1.6219,
+      "step": 26280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.151337655072559e-05,
+      "loss": 1.5624,
+      "step": 26290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.151008588634045e-05,
+      "loss": 1.6481,
+      "step": 26300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1506795221955314e-05,
+      "loss": 1.585,
+      "step": 26310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1503504557570175e-05,
+      "loss": 1.8122,
+      "step": 26320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.150021389318504e-05,
+      "loss": 1.8364,
+      "step": 26330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.149692322879989e-05,
+      "loss": 1.8798,
+      "step": 26340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.149363256441476e-05,
+      "loss": 1.6208,
+      "step": 26350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1490341900029614e-05,
+      "loss": 1.8033,
+      "step": 26360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.148705123564448e-05,
+      "loss": 1.8763,
+      "step": 26370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1483760571259336e-05,
+      "loss": 1.6492,
+      "step": 26380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.14804699068742e-05,
+      "loss": 1.6392,
+      "step": 26390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.147717924248906e-05,
+      "loss": 1.7475,
+      "step": 26400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.147388857810392e-05,
+      "loss": 1.9071,
+      "step": 26410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.147059791371878e-05,
+      "loss": 1.5992,
+      "step": 26420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.146730724933364e-05,
+      "loss": 1.5564,
+      "step": 26430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1464016584948504e-05,
+      "loss": 1.838,
+      "step": 26440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1460725920563365e-05,
+      "loss": 1.7289,
+      "step": 26450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1457435256178226e-05,
+      "loss": 1.5374,
+      "step": 26460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.145414459179309e-05,
+      "loss": 1.9013,
+      "step": 26470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.145085392740794e-05,
+      "loss": 1.8337,
+      "step": 26480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1447563263022803e-05,
+      "loss": 1.6367,
+      "step": 26490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1444272598637665e-05,
+      "loss": 1.7502,
+      "step": 26500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1440981934252526e-05,
+      "loss": 1.8402,
+      "step": 26510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143769126986739e-05,
+      "loss": 1.7321,
+      "step": 26520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143440060548225e-05,
+      "loss": 1.6563,
+      "step": 26530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143110994109711e-05,
+      "loss": 1.7581,
+      "step": 26540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142781927671197e-05,
+      "loss": 1.7026,
+      "step": 26550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142452861232683e-05,
+      "loss": 1.5975,
+      "step": 26560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1421237947941694e-05,
+      "loss": 1.6073,
+      "step": 26570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1417947283556555e-05,
+      "loss": 1.6392,
+      "step": 26580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141465661917141e-05,
+      "loss": 1.7115,
+      "step": 26590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141136595478628e-05,
+      "loss": 1.7984,
+      "step": 26600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140807529040113e-05,
+      "loss": 1.6995,
+      "step": 26610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1404784626016e-05,
+      "loss": 1.7747,
+      "step": 26620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1401493961630854e-05,
+      "loss": 1.6426,
+      "step": 26630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1398203297245716e-05,
+      "loss": 1.5804,
+      "step": 26640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139491263286058e-05,
+      "loss": 1.7903,
+      "step": 26650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139162196847544e-05,
+      "loss": 1.7984,
+      "step": 26660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13883313040903e-05,
+      "loss": 1.6812,
+      "step": 26670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.138504063970516e-05,
+      "loss": 1.8011,
+      "step": 26680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1381749975320015e-05,
+      "loss": 1.9174,
+      "step": 26690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.137845931093488e-05,
+      "loss": 1.7312,
+      "step": 26700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.137516864654974e-05,
+      "loss": 1.556,
+      "step": 26710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1371877982164606e-05,
+      "loss": 1.5445,
+      "step": 26720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.136858731777946e-05,
+      "loss": 1.6639,
+      "step": 26730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.136529665339432e-05,
+      "loss": 1.6586,
+      "step": 26740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.136200598900918e-05,
+      "loss": 1.4812,
+      "step": 26750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1358715324624044e-05,
+      "loss": 1.6763,
+      "step": 26760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1355424660238905e-05,
+      "loss": 1.6109,
+      "step": 26770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.135213399585377e-05,
+      "loss": 1.6013,
+      "step": 26780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.134884333146862e-05,
+      "loss": 1.8793,
+      "step": 26790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.134555266708349e-05,
+      "loss": 1.6643,
+      "step": 26800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1342262002698344e-05,
+      "loss": 1.5754,
+      "step": 26810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.133897133831321e-05,
+      "loss": 1.6295,
+      "step": 26820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1335680673928066e-05,
+      "loss": 1.6054,
+      "step": 26830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.133239000954293e-05,
+      "loss": 1.8697,
+      "step": 26840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.132909934515779e-05,
+      "loss": 1.5883,
+      "step": 26850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.132580868077265e-05,
+      "loss": 1.7218,
+      "step": 26860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.132251801638751e-05,
+      "loss": 1.6354,
+      "step": 26870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.131922735200237e-05,
+      "loss": 1.7549,
+      "step": 26880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.131593668761723e-05,
+      "loss": 1.7318,
+      "step": 26890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1312646023232095e-05,
+      "loss": 1.5952,
+      "step": 26900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.130935535884695e-05,
+      "loss": 2.0802,
+      "step": 26910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.130606469446182e-05,
+      "loss": 1.5405,
+      "step": 26920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.130277403007667e-05,
+      "loss": 1.7943,
+      "step": 26930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.129948336569153e-05,
+      "loss": 1.7565,
+      "step": 26940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1296192701306395e-05,
+      "loss": 1.6223,
+      "step": 26950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1292902036921256e-05,
+      "loss": 1.6754,
+      "step": 26960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.128961137253612e-05,
+      "loss": 1.7332,
+      "step": 26970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.128632070815098e-05,
+      "loss": 1.8416,
+      "step": 26980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.128303004376583e-05,
+      "loss": 1.7856,
+      "step": 26990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12797393793807e-05,
+      "loss": 1.6528,
+      "step": 27000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1276448714995555e-05,
+      "loss": 1.7392,
+      "step": 27010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1273158050610424e-05,
+      "loss": 1.7305,
+      "step": 27020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126986738622528e-05,
+      "loss": 1.7024,
+      "step": 27030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126657672184014e-05,
+      "loss": 1.7268,
+      "step": 27040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1263286057455e-05,
+      "loss": 1.4882,
+      "step": 27050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125999539306986e-05,
+      "loss": 1.8848,
+      "step": 27060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125670472868472e-05,
+      "loss": 1.8343,
+      "step": 27070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1253414064299584e-05,
+      "loss": 1.6649,
+      "step": 27080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125012339991444e-05,
+      "loss": 1.6338,
+      "step": 27090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124683273552931e-05,
+      "loss": 1.6667,
+      "step": 27100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124354207114416e-05,
+      "loss": 1.6509,
+      "step": 27110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124025140675903e-05,
+      "loss": 1.7857,
+      "step": 27120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1236960742373884e-05,
+      "loss": 1.7983,
+      "step": 27130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1233670077988745e-05,
+      "loss": 1.7543,
+      "step": 27140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1230379413603606e-05,
+      "loss": 1.7855,
+      "step": 27150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122708874921847e-05,
+      "loss": 1.6739,
+      "step": 27160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122379808483333e-05,
+      "loss": 1.6346,
+      "step": 27170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122050742044819e-05,
+      "loss": 1.7253,
+      "step": 27180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.121721675606305e-05,
+      "loss": 1.6757,
+      "step": 27190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.121392609167791e-05,
+      "loss": 1.5691,
+      "step": 27200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1210635427292774e-05,
+      "loss": 1.5892,
+      "step": 27210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1207344762907635e-05,
+      "loss": 1.6734,
+      "step": 27220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1204054098522497e-05,
+      "loss": 1.6866,
+      "step": 27230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.120076343413735e-05,
+      "loss": 1.8203,
+      "step": 27240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.119747276975221e-05,
+      "loss": 1.603,
+      "step": 27250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1194182105367074e-05,
+      "loss": 1.7096,
+      "step": 27260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1190891440981935e-05,
+      "loss": 1.7382,
+      "step": 27270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1187600776596796e-05,
+      "loss": 1.6514,
+      "step": 27280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.118431011221166e-05,
+      "loss": 1.4671,
+      "step": 27290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.118101944782652e-05,
+      "loss": 1.5991,
+      "step": 27300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.117772878344138e-05,
+      "loss": 1.8205,
+      "step": 27310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.117443811905624e-05,
+      "loss": 1.8214,
+      "step": 27320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.11711474546711e-05,
+      "loss": 1.7266,
+      "step": 27330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.116785679028596e-05,
+      "loss": 1.6483,
+      "step": 27340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1164566125900825e-05,
+      "loss": 1.7343,
+      "step": 27350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.116127546151568e-05,
+      "loss": 1.6408,
+      "step": 27360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.115798479713055e-05,
+      "loss": 1.754,
+      "step": 27370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.11546941327454e-05,
+      "loss": 1.6871,
+      "step": 27380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.115140346836026e-05,
+      "loss": 1.5443,
+      "step": 27390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1148112803975125e-05,
+      "loss": 1.7294,
+      "step": 27400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1144822139589986e-05,
+      "loss": 1.6367,
+      "step": 27410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.114153147520485e-05,
+      "loss": 1.6281,
+      "step": 27420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.113824081081971e-05,
+      "loss": 1.785,
+      "step": 27430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.113495014643456e-05,
+      "loss": 1.5419,
+      "step": 27440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.113165948204943e-05,
+      "loss": 1.5326,
+      "step": 27450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1128368817664285e-05,
+      "loss": 1.6703,
+      "step": 27460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1125078153279153e-05,
+      "loss": 2.0075,
+      "step": 27470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.112178748889401e-05,
+      "loss": 1.6766,
+      "step": 27480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1118496824508876e-05,
+      "loss": 1.5692,
+      "step": 27490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.111520616012373e-05,
+      "loss": 1.8307,
+      "step": 27500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.111191549573859e-05,
+      "loss": 1.7057,
+      "step": 27510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110862483135345e-05,
+      "loss": 1.7772,
+      "step": 27520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1105334166968314e-05,
+      "loss": 1.7184,
+      "step": 27530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1102043502583176e-05,
+      "loss": 1.6901,
+      "step": 27540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109875283819804e-05,
+      "loss": 1.8049,
+      "step": 27550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109546217381289e-05,
+      "loss": 1.8419,
+      "step": 27560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109217150942776e-05,
+      "loss": 1.7302,
+      "step": 27570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1088880845042614e-05,
+      "loss": 1.5803,
+      "step": 27580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108559018065748e-05,
+      "loss": 1.5366,
+      "step": 27590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1082299516272336e-05,
+      "loss": 1.5757,
+      "step": 27600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10790088518872e-05,
+      "loss": 1.6533,
+      "step": 27610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107571818750206e-05,
+      "loss": 1.6452,
+      "step": 27620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107242752311692e-05,
+      "loss": 1.726,
+      "step": 27630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106913685873178e-05,
+      "loss": 1.6116,
+      "step": 27640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106584619434664e-05,
+      "loss": 1.7543,
+      "step": 27650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10625555299615e-05,
+      "loss": 1.6678,
+      "step": 27660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1059264865576365e-05,
+      "loss": 1.4752,
+      "step": 27670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105597420119122e-05,
+      "loss": 1.7023,
+      "step": 27680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105268353680609e-05,
+      "loss": 1.6479,
+      "step": 27690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.104939287242094e-05,
+      "loss": 1.6172,
+      "step": 27700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1046102208035804e-05,
+      "loss": 1.616,
+      "step": 27710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1042811543650665e-05,
+      "loss": 1.9909,
+      "step": 27720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1039520879265526e-05,
+      "loss": 1.5631,
+      "step": 27730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.103623021488039e-05,
+      "loss": 1.6406,
+      "step": 27740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.103293955049525e-05,
+      "loss": 1.634,
+      "step": 27750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.10296488861101e-05,
+      "loss": 1.6866,
+      "step": 27760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.102635822172497e-05,
+      "loss": 1.8759,
+      "step": 27770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1023067557339826e-05,
+      "loss": 1.7923,
+      "step": 27780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1019776892954694e-05,
+      "loss": 1.7514,
+      "step": 27790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.101648622856955e-05,
+      "loss": 1.5867,
+      "step": 27800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.101319556418441e-05,
+      "loss": 1.7288,
+      "step": 27810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.100990489979927e-05,
+      "loss": 1.6537,
+      "step": 27820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.100661423541413e-05,
+      "loss": 1.5485,
+      "step": 27830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.100332357102899e-05,
+      "loss": 1.6488,
+      "step": 27840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1000032906643855e-05,
+      "loss": 1.7699,
+      "step": 27850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.099674224225871e-05,
+      "loss": 1.6928,
+      "step": 27860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.099345157787358e-05,
+      "loss": 1.6961,
+      "step": 27870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.099016091348843e-05,
+      "loss": 1.6022,
+      "step": 27880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.09868702491033e-05,
+      "loss": 1.6469,
+      "step": 27890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0983579584718154e-05,
+      "loss": 1.7232,
+      "step": 27900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0980288920333015e-05,
+      "loss": 1.6951,
+      "step": 27910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0976998255947877e-05,
+      "loss": 1.7775,
+      "step": 27920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.097370759156274e-05,
+      "loss": 1.5704,
+      "step": 27930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.09704169271776e-05,
+      "loss": 1.6124,
+      "step": 27940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.096712626279246e-05,
+      "loss": 1.6307,
+      "step": 27950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.096383559840732e-05,
+      "loss": 1.7164,
+      "step": 27960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.096054493402218e-05,
+      "loss": 1.7255,
+      "step": 27970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0957254269637044e-05,
+      "loss": 1.5646,
+      "step": 27980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0953963605251905e-05,
+      "loss": 1.7012,
+      "step": 27990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.095067294086676e-05,
+      "loss": 1.8389,
+      "step": 28000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.094738227648162e-05,
+      "loss": 1.6466,
+      "step": 28010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.094409161209648e-05,
+      "loss": 1.7886,
+      "step": 28020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0940800947711344e-05,
+      "loss": 1.6648,
+      "step": 28030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0937510283326205e-05,
+      "loss": 1.6901,
+      "step": 28040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0934219618941066e-05,
+      "loss": 1.7453,
+      "step": 28050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093092895455593e-05,
+      "loss": 1.7316,
+      "step": 28060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092763829017079e-05,
+      "loss": 1.9332,
+      "step": 28070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092434762578565e-05,
+      "loss": 1.7352,
+      "step": 28080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092105696140051e-05,
+      "loss": 1.7004,
+      "step": 28090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091776629701537e-05,
+      "loss": 1.6444,
+      "step": 28100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091447563263023e-05,
+      "loss": 1.6838,
+      "step": 28110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0911184968245095e-05,
+      "loss": 1.7391,
+      "step": 28120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090789430385995e-05,
+      "loss": 1.7273,
+      "step": 28130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090460363947482e-05,
+      "loss": 1.5764,
+      "step": 28140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090131297508967e-05,
+      "loss": 1.6137,
+      "step": 28150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0898022310704533e-05,
+      "loss": 1.8203,
+      "step": 28160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0894731646319395e-05,
+      "loss": 1.6528,
+      "step": 28170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0891440981934256e-05,
+      "loss": 1.7121,
+      "step": 28180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088815031754912e-05,
+      "loss": 1.6662,
+      "step": 28190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088485965316398e-05,
+      "loss": 1.5651,
+      "step": 28200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.088156898877883e-05,
+      "loss": 1.6509,
+      "step": 28210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.08782783243937e-05,
+      "loss": 1.6315,
+      "step": 28220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0874987660008556e-05,
+      "loss": 1.6211,
+      "step": 28230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0871696995623424e-05,
+      "loss": 1.6045,
+      "step": 28240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.086840633123828e-05,
+      "loss": 1.7184,
+      "step": 28250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.086511566685314e-05,
+      "loss": 1.6672,
+      "step": 28260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0861825002468e-05,
+      "loss": 1.8036,
+      "step": 28270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.085853433808286e-05,
+      "loss": 1.5753,
+      "step": 28280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.085524367369772e-05,
+      "loss": 1.6991,
+      "step": 28290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0851953009312584e-05,
+      "loss": 1.7641,
+      "step": 28300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.084866234492744e-05,
+      "loss": 1.6941,
+      "step": 28310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.084537168054231e-05,
+      "loss": 1.8502,
+      "step": 28320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.084208101615716e-05,
+      "loss": 1.8006,
+      "step": 28330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.083879035177203e-05,
+      "loss": 1.6899,
+      "step": 28340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0835499687386884e-05,
+      "loss": 1.8429,
+      "step": 28350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0832209023001745e-05,
+      "loss": 1.749,
+      "step": 28360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0828918358616607e-05,
+      "loss": 1.6726,
+      "step": 28370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.082562769423147e-05,
+      "loss": 1.9566,
+      "step": 28380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.082233702984633e-05,
+      "loss": 1.744,
+      "step": 28390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.081904636546119e-05,
+      "loss": 1.5041,
+      "step": 28400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0815755701076045e-05,
+      "loss": 1.5774,
+      "step": 28410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.081246503669091e-05,
+      "loss": 1.6413,
+      "step": 28420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.080917437230577e-05,
+      "loss": 1.6615,
+      "step": 28430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0805883707920635e-05,
+      "loss": 1.564,
+      "step": 28440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.080259304353549e-05,
+      "loss": 1.7145,
+      "step": 28450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.079930237915035e-05,
+      "loss": 1.3958,
+      "step": 28460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.079601171476521e-05,
+      "loss": 1.5184,
+      "step": 28470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0792721050380074e-05,
+      "loss": 1.6005,
+      "step": 28480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0789430385994935e-05,
+      "loss": 1.6457,
+      "step": 28490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0786139721609796e-05,
+      "loss": 1.6564,
+      "step": 28500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.078284905722465e-05,
+      "loss": 1.8209,
+      "step": 28510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077955839283952e-05,
+      "loss": 1.5669,
+      "step": 28520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077626772845437e-05,
+      "loss": 1.5933,
+      "step": 28530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077297706406924e-05,
+      "loss": 1.659,
+      "step": 28540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0769686399684096e-05,
+      "loss": 1.7039,
+      "step": 28550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076639573529896e-05,
+      "loss": 2.0473,
+      "step": 28560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076310507091382e-05,
+      "loss": 1.659,
+      "step": 28570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075981440652868e-05,
+      "loss": 1.6782,
+      "step": 28580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075652374214354e-05,
+      "loss": 1.7165,
+      "step": 28590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07532330777584e-05,
+      "loss": 1.6423,
+      "step": 28600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074994241337326e-05,
+      "loss": 1.6562,
+      "step": 28610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0746651748988125e-05,
+      "loss": 1.7015,
+      "step": 28620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074336108460298e-05,
+      "loss": 1.6398,
+      "step": 28630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074007042021785e-05,
+      "loss": 1.6454,
+      "step": 28640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07367797558327e-05,
+      "loss": 1.4425,
+      "step": 28650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073348909144756e-05,
+      "loss": 1.5359,
+      "step": 28660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0730198427062424e-05,
+      "loss": 1.6364,
+      "step": 28670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0726907762677285e-05,
+      "loss": 1.8886,
+      "step": 28680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072361709829215e-05,
+      "loss": 2.0922,
+      "step": 28690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072032643390701e-05,
+      "loss": 1.6513,
+      "step": 28700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071703576952187e-05,
+      "loss": 1.6899,
+      "step": 28710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071374510513673e-05,
+      "loss": 1.6449,
+      "step": 28720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071045444075159e-05,
+      "loss": 1.6276,
+      "step": 28730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070716377636645e-05,
+      "loss": 1.4624,
+      "step": 28740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0703873111981314e-05,
+      "loss": 1.6616,
+      "step": 28750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070058244759617e-05,
+      "loss": 1.6525,
+      "step": 28760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.069729178321103e-05,
+      "loss": 1.8547,
+      "step": 28770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.069400111882589e-05,
+      "loss": 1.632,
+      "step": 28780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.069071045444075e-05,
+      "loss": 1.7706,
+      "step": 28790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0687419790055614e-05,
+      "loss": 1.5307,
+      "step": 28800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0684129125670475e-05,
+      "loss": 1.5468,
+      "step": 28810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0680838461285336e-05,
+      "loss": 1.7228,
+      "step": 28820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06775477969002e-05,
+      "loss": 1.7211,
+      "step": 28830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.067425713251506e-05,
+      "loss": 1.8136,
+      "step": 28840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.067096646812992e-05,
+      "loss": 1.7829,
+      "step": 28850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0667675803744775e-05,
+      "loss": 1.7751,
+      "step": 28860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066438513935964e-05,
+      "loss": 1.6392,
+      "step": 28870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06610944749745e-05,
+      "loss": 1.5901,
+      "step": 28880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0657803810589365e-05,
+      "loss": 1.7897,
+      "step": 28890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.065451314620422e-05,
+      "loss": 1.7018,
+      "step": 28900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.065122248181908e-05,
+      "loss": 1.5537,
+      "step": 28910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064793181743394e-05,
+      "loss": 1.7356,
+      "step": 28920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0644641153048804e-05,
+      "loss": 1.5445,
+      "step": 28930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0641350488663665e-05,
+      "loss": 1.6684,
+      "step": 28940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0638059824278526e-05,
+      "loss": 1.7112,
+      "step": 28950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063476915989338e-05,
+      "loss": 1.4512,
+      "step": 28960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063147849550825e-05,
+      "loss": 1.5671,
+      "step": 28970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06281878311231e-05,
+      "loss": 1.9787,
+      "step": 28980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062489716673797e-05,
+      "loss": 1.7052,
+      "step": 28990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0621606502352826e-05,
+      "loss": 1.7002,
+      "step": 29000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061831583796769e-05,
+      "loss": 1.7797,
+      "step": 29010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061502517358255e-05,
+      "loss": 1.6149,
+      "step": 29020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061173450919741e-05,
+      "loss": 1.6598,
+      "step": 29030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060844384481227e-05,
+      "loss": 1.7844,
+      "step": 29040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060515318042713e-05,
+      "loss": 1.8013,
+      "step": 29050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0601862516041987e-05,
+      "loss": 1.7562,
+      "step": 29060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0598571851656855e-05,
+      "loss": 1.6318,
+      "step": 29070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059528118727171e-05,
+      "loss": 1.6647,
+      "step": 29080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059199052288658e-05,
+      "loss": 1.6406,
+      "step": 29090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058869985850143e-05,
+      "loss": 1.622,
+      "step": 29100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058540919411629e-05,
+      "loss": 1.6393,
+      "step": 29110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0582118529731154e-05,
+      "loss": 1.6525,
+      "step": 29120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0578827865346015e-05,
+      "loss": 1.725,
+      "step": 29130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057553720096088e-05,
+      "loss": 1.7109,
+      "step": 29140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057224653657574e-05,
+      "loss": 1.6285,
+      "step": 29150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056895587219059e-05,
+      "loss": 1.7275,
+      "step": 29160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056566520780546e-05,
+      "loss": 1.4969,
+      "step": 29170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0562374543420315e-05,
+      "loss": 1.7534,
+      "step": 29180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055908387903518e-05,
+      "loss": 1.728,
+      "step": 29190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055579321465004e-05,
+      "loss": 1.7283,
+      "step": 29200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05525025502649e-05,
+      "loss": 1.991,
+      "step": 29210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.054921188587976e-05,
+      "loss": 1.7504,
+      "step": 29220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.054592122149462e-05,
+      "loss": 1.7293,
+      "step": 29230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.054263055710948e-05,
+      "loss": 1.6632,
+      "step": 29240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0539339892724344e-05,
+      "loss": 1.7076,
+      "step": 29250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.05360492283392e-05,
+      "loss": 1.6329,
+      "step": 29260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0532758563954066e-05,
+      "loss": 1.5375,
+      "step": 29270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.052946789956892e-05,
+      "loss": 1.6098,
+      "step": 29280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.052617723518379e-05,
+      "loss": 1.5408,
+      "step": 29290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0522886570798643e-05,
+      "loss": 1.7292,
+      "step": 29300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0519595906413505e-05,
+      "loss": 1.6095,
+      "step": 29310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0516305242028366e-05,
+      "loss": 1.7536,
+      "step": 29320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.051301457764323e-05,
+      "loss": 1.793,
+      "step": 29330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.050972391325809e-05,
+      "loss": 1.646,
+      "step": 29340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.050643324887295e-05,
+      "loss": 1.5772,
+      "step": 29350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0503142584487804e-05,
+      "loss": 1.7165,
+      "step": 29360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.049985192010267e-05,
+      "loss": 1.6213,
+      "step": 29370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.049656125571753e-05,
+      "loss": 1.711,
+      "step": 29380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0493270591332395e-05,
+      "loss": 1.7885,
+      "step": 29390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.048997992694725e-05,
+      "loss": 1.6507,
+      "step": 29400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.048668926256211e-05,
+      "loss": 1.6344,
+      "step": 29410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.048339859817697e-05,
+      "loss": 1.89,
+      "step": 29420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.048010793379183e-05,
+      "loss": 1.7543,
+      "step": 29430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0476817269406694e-05,
+      "loss": 1.5715,
+      "step": 29440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0473526605021556e-05,
+      "loss": 1.644,
+      "step": 29450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.047023594063642e-05,
+      "loss": 1.7983,
+      "step": 29460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046694527625128e-05,
+      "loss": 1.6775,
+      "step": 29470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046365461186614e-05,
+      "loss": 1.539,
+      "step": 29480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0460363947481e-05,
+      "loss": 1.5951,
+      "step": 29490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.045707328309586e-05,
+      "loss": 1.5738,
+      "step": 29500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0453782618710716e-05,
+      "loss": 1.6946,
+      "step": 29510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0450491954325585e-05,
+      "loss": 1.6939,
+      "step": 29520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.044720128994044e-05,
+      "loss": 1.6392,
+      "step": 29530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04439106255553e-05,
+      "loss": 1.7253,
+      "step": 29540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.044061996117016e-05,
+      "loss": 1.7247,
+      "step": 29550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043732929678502e-05,
+      "loss": 1.7588,
+      "step": 29560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0434038632399884e-05,
+      "loss": 1.6052,
+      "step": 29570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0430747968014745e-05,
+      "loss": 1.8863,
+      "step": 29580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042745730362961e-05,
+      "loss": 1.7488,
+      "step": 29590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042416663924447e-05,
+      "loss": 1.6199,
+      "step": 29600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042087597485932e-05,
+      "loss": 1.7712,
+      "step": 29610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041758531047419e-05,
+      "loss": 1.7168,
+      "step": 29620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0414294646089045e-05,
+      "loss": 1.6695,
+      "step": 29630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041100398170391e-05,
+      "loss": 1.571,
+      "step": 29640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040771331731877e-05,
+      "loss": 1.7809,
+      "step": 29650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040442265293363e-05,
+      "loss": 1.8678,
+      "step": 29660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040113198854849e-05,
+      "loss": 1.7586,
+      "step": 29670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039784132416335e-05,
+      "loss": 1.5819,
+      "step": 29680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039455065977821e-05,
+      "loss": 1.7008,
+      "step": 29690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0391259995393074e-05,
+      "loss": 1.621,
+      "step": 29700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038796933100793e-05,
+      "loss": 1.9098,
+      "step": 29710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0384678666622796e-05,
+      "loss": 1.7231,
+      "step": 29720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.038138800223765e-05,
+      "loss": 1.7288,
+      "step": 29730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.037809733785252e-05,
+      "loss": 1.7002,
+      "step": 29740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.037480667346737e-05,
+      "loss": 1.5201,
+      "step": 29750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0371516009082235e-05,
+      "loss": 1.6583,
+      "step": 29760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0368225344697096e-05,
+      "loss": 1.6078,
+      "step": 29770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.036493468031196e-05,
+      "loss": 1.9699,
+      "step": 29780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.036164401592682e-05,
+      "loss": 1.5024,
+      "step": 29790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.035835335154168e-05,
+      "loss": 1.8786,
+      "step": 29800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0355062687156534e-05,
+      "loss": 1.6306,
+      "step": 29810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.03517720227714e-05,
+      "loss": 1.7576,
+      "step": 29820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.034848135838626e-05,
+      "loss": 1.4209,
+      "step": 29830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0345190694001125e-05,
+      "loss": 1.7335,
+      "step": 29840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.034190002961598e-05,
+      "loss": 1.4974,
+      "step": 29850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.033860936523084e-05,
+      "loss": 1.6607,
+      "step": 29860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.03353187008457e-05,
+      "loss": 1.8861,
+      "step": 29870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.033202803646056e-05,
+      "loss": 1.8422,
+      "step": 29880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0328737372075424e-05,
+      "loss": 1.762,
+      "step": 29890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0325446707690286e-05,
+      "loss": 1.5514,
+      "step": 29900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.032215604330514e-05,
+      "loss": 1.6631,
+      "step": 29910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031886537892001e-05,
+      "loss": 1.7672,
+      "step": 29920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031557471453486e-05,
+      "loss": 1.8597,
+      "step": 29930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031228405014973e-05,
+      "loss": 1.6772,
+      "step": 29940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0308993385764585e-05,
+      "loss": 1.7142,
+      "step": 29950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0305702721379446e-05,
+      "loss": 1.7514,
+      "step": 29960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.030241205699431e-05,
+      "loss": 1.5719,
+      "step": 29970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.029912139260917e-05,
+      "loss": 1.781,
+      "step": 29980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.029583072822403e-05,
+      "loss": 1.7128,
+      "step": 29990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.029254006383889e-05,
+      "loss": 1.7333,
+      "step": 30000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0289249399453746e-05,
+      "loss": 1.9232,
+      "step": 30010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0285958735068614e-05,
+      "loss": 1.6801,
+      "step": 30020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028266807068347e-05,
+      "loss": 1.4957,
+      "step": 30030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0279377406298337e-05,
+      "loss": 1.614,
+      "step": 30040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027608674191319e-05,
+      "loss": 1.7834,
+      "step": 30050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027279607752805e-05,
+      "loss": 1.5882,
+      "step": 30060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0269505413142914e-05,
+      "loss": 1.6274,
+      "step": 30070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0266214748757775e-05,
+      "loss": 1.8606,
+      "step": 30080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0262924084372636e-05,
+      "loss": 1.7707,
+      "step": 30090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02596334199875e-05,
+      "loss": 1.5939,
+      "step": 30100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025634275560236e-05,
+      "loss": 1.6896,
+      "step": 30110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025305209121722e-05,
+      "loss": 1.5913,
+      "step": 30120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0249761426832074e-05,
+      "loss": 1.7519,
+      "step": 30130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024647076244694e-05,
+      "loss": 1.5933,
+      "step": 30140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02431800980618e-05,
+      "loss": 1.8143,
+      "step": 30150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023988943367666e-05,
+      "loss": 1.8379,
+      "step": 30160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023659876929152e-05,
+      "loss": 1.555,
+      "step": 30170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023330810490638e-05,
+      "loss": 1.4608,
+      "step": 30180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023001744052124e-05,
+      "loss": 1.6091,
+      "step": 30190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02267267761361e-05,
+      "loss": 1.6873,
+      "step": 30200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0223436111750965e-05,
+      "loss": 1.5725,
+      "step": 30210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0220145447365826e-05,
+      "loss": 1.8207,
+      "step": 30220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021685478298069e-05,
+      "loss": 1.7922,
+      "step": 30230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.021356411859555e-05,
+      "loss": 1.5577,
+      "step": 30240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.021027345421041e-05,
+      "loss": 1.6625,
+      "step": 30250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0206982789825264e-05,
+      "loss": 1.7332,
+      "step": 30260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.020369212544013e-05,
+      "loss": 1.5848,
+      "step": 30270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.020040146105499e-05,
+      "loss": 1.7983,
+      "step": 30280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.019711079666985e-05,
+      "loss": 1.5763,
+      "step": 30290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.019382013228471e-05,
+      "loss": 1.5655,
+      "step": 30300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.019052946789957e-05,
+      "loss": 1.633,
+      "step": 30310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.018723880351443e-05,
+      "loss": 1.766,
+      "step": 30320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.018394813912929e-05,
+      "loss": 1.6285,
+      "step": 30330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0180657474744154e-05,
+      "loss": 1.5384,
+      "step": 30340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0177366810359016e-05,
+      "loss": 1.6217,
+      "step": 30350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.017407614597388e-05,
+      "loss": 1.7243,
+      "step": 30360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.017078548158874e-05,
+      "loss": 1.7194,
+      "step": 30370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.016749481720359e-05,
+      "loss": 1.5702,
+      "step": 30380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.016420415281846e-05,
+      "loss": 1.6002,
+      "step": 30390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0160913488433315e-05,
+      "loss": 1.7827,
+      "step": 30400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.015762282404818e-05,
+      "loss": 1.7597,
+      "step": 30410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.015433215966304e-05,
+      "loss": 1.5782,
+      "step": 30420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.01510414952779e-05,
+      "loss": 1.6179,
+      "step": 30430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.014775083089276e-05,
+      "loss": 1.6376,
+      "step": 30440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.014446016650762e-05,
+      "loss": 1.6894,
+      "step": 30450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.014116950212248e-05,
+      "loss": 1.6702,
+      "step": 30460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0137878837737344e-05,
+      "loss": 1.5869,
+      "step": 30470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.01345881733522e-05,
+      "loss": 1.7166,
+      "step": 30480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0131297508967066e-05,
+      "loss": 1.5841,
+      "step": 30490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.012800684458192e-05,
+      "loss": 1.5763,
+      "step": 30500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.012471618019679e-05,
+      "loss": 1.8135,
+      "step": 30510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0121425515811644e-05,
+      "loss": 1.6594,
+      "step": 30520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0118134851426505e-05,
+      "loss": 1.9855,
+      "step": 30530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0114844187041366e-05,
+      "loss": 1.68,
+      "step": 30540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.011155352265623e-05,
+      "loss": 1.5777,
+      "step": 30550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010826285827109e-05,
+      "loss": 1.803,
+      "step": 30560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010497219388595e-05,
+      "loss": 1.7219,
+      "step": 30570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0101681529500804e-05,
+      "loss": 1.6894,
+      "step": 30580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009839086511567e-05,
+      "loss": 1.5767,
+      "step": 30590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009510020073053e-05,
+      "loss": 1.6488,
+      "step": 30600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0091809536345395e-05,
+      "loss": 1.7459,
+      "step": 30610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008851887196025e-05,
+      "loss": 1.5895,
+      "step": 30620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008522820757511e-05,
+      "loss": 1.6667,
+      "step": 30630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008193754318997e-05,
+      "loss": 1.7948,
+      "step": 30640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007864687880483e-05,
+      "loss": 1.5582,
+      "step": 30650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0075356214419694e-05,
+      "loss": 1.5413,
+      "step": 30660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0072065550034556e-05,
+      "loss": 1.7852,
+      "step": 30670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006877488564941e-05,
+      "loss": 1.5147,
+      "step": 30680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006548422126428e-05,
+      "loss": 1.7559,
+      "step": 30690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006219355687913e-05,
+      "loss": 1.7862,
+      "step": 30700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0058902892494e-05,
+      "loss": 1.7786,
+      "step": 30710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0055612228108855e-05,
+      "loss": 1.5388,
+      "step": 30720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0052321563723717e-05,
+      "loss": 1.7685,
+      "step": 30730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004903089933858e-05,
+      "loss": 1.528,
+      "step": 30740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.004574023495344e-05,
+      "loss": 1.6199,
+      "step": 30750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.00424495705683e-05,
+      "loss": 1.752,
+      "step": 30760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.003915890618316e-05,
+      "loss": 1.6671,
+      "step": 30770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0035868241798016e-05,
+      "loss": 1.5804,
+      "step": 30780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0032577577412884e-05,
+      "loss": 1.6981,
+      "step": 30790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.002928691302774e-05,
+      "loss": 1.5053,
+      "step": 30800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.002599624864261e-05,
+      "loss": 1.5987,
+      "step": 30810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.002270558425746e-05,
+      "loss": 1.7064,
+      "step": 30820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.001941491987232e-05,
+      "loss": 1.5999,
+      "step": 30830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0016124255487184e-05,
+      "loss": 1.6857,
+      "step": 30840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0012833591102045e-05,
+      "loss": 1.7141,
+      "step": 30850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0009542926716906e-05,
+      "loss": 1.6592,
+      "step": 30860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.000625226233177e-05,
+      "loss": 1.6038,
+      "step": 30870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.000296159794663e-05,
+      "loss": 1.6546,
+      "step": 30880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.999967093356149e-05,
+      "loss": 1.6574,
+      "step": 30890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9996380269176345e-05,
+      "loss": 1.6298,
+      "step": 30900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.999308960479121e-05,
+      "loss": 1.7193,
+      "step": 30910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.998979894040607e-05,
+      "loss": 1.5794,
+      "step": 30920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.998650827602093e-05,
+      "loss": 1.5958,
+      "step": 30930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.998321761163579e-05,
+      "loss": 1.7173,
+      "step": 30940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.997992694725065e-05,
+      "loss": 1.7229,
+      "step": 30950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.997663628286551e-05,
+      "loss": 1.589,
+      "step": 30960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9973345618480373e-05,
+      "loss": 1.7331,
+      "step": 30970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9970054954095235e-05,
+      "loss": 1.617,
+      "step": 30980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9966764289710096e-05,
+      "loss": 1.7615,
+      "step": 30990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.996347362532496e-05,
+      "loss": 1.4645,
+      "step": 31000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.996018296093982e-05,
+      "loss": 1.9515,
+      "step": 31010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.995689229655468e-05,
+      "loss": 1.5539,
+      "step": 31020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9953601632169534e-05,
+      "loss": 1.609,
+      "step": 31030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99503109677844e-05,
+      "loss": 1.7938,
+      "step": 31040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994702030339926e-05,
+      "loss": 1.8137,
+      "step": 31050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994372963901412e-05,
+      "loss": 1.655,
+      "step": 31060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994043897462898e-05,
+      "loss": 1.6144,
+      "step": 31070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993714831024384e-05,
+      "loss": 1.6592,
+      "step": 31080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99338576458587e-05,
+      "loss": 1.7225,
+      "step": 31090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993056698147356e-05,
+      "loss": 1.6538,
+      "step": 31100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9927276317088424e-05,
+      "loss": 1.6994,
+      "step": 31110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9923985652703286e-05,
+      "loss": 1.6641,
+      "step": 31120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992069498831814e-05,
+      "loss": 1.8645,
+      "step": 31130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991740432393301e-05,
+      "loss": 1.6209,
+      "step": 31140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991411365954786e-05,
+      "loss": 1.6255,
+      "step": 31150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991082299516273e-05,
+      "loss": 1.5865,
+      "step": 31160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9907532330777585e-05,
+      "loss": 1.7199,
+      "step": 31170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9904241666392447e-05,
+      "loss": 1.8514,
+      "step": 31180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990095100200731e-05,
+      "loss": 1.6856,
+      "step": 31190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989766033762217e-05,
+      "loss": 2.0367,
+      "step": 31200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989436967323703e-05,
+      "loss": 1.6519,
+      "step": 31210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989107900885189e-05,
+      "loss": 1.5594,
+      "step": 31220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9887788344466746e-05,
+      "loss": 1.7736,
+      "step": 31230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9884497680081614e-05,
+      "loss": 1.6308,
+      "step": 31240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988120701569647e-05,
+      "loss": 1.6951,
+      "step": 31250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.987791635131134e-05,
+      "loss": 1.5404,
+      "step": 31260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.987462568692619e-05,
+      "loss": 1.8397,
+      "step": 31270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.987133502254105e-05,
+      "loss": 1.6143,
+      "step": 31280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9868044358155914e-05,
+      "loss": 1.5759,
+      "step": 31290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9864753693770775e-05,
+      "loss": 1.8153,
+      "step": 31300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9861463029385636e-05,
+      "loss": 1.6577,
+      "step": 31310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.98581723650005e-05,
+      "loss": 1.6702,
+      "step": 31320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.985488170061535e-05,
+      "loss": 1.6963,
+      "step": 31330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.985159103623022e-05,
+      "loss": 1.5593,
+      "step": 31340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9848300371845074e-05,
+      "loss": 1.6529,
+      "step": 31350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.984500970745994e-05,
+      "loss": 1.7236,
+      "step": 31360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.98417190430748e-05,
+      "loss": 1.7184,
+      "step": 31370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.983842837868966e-05,
+      "loss": 1.6769,
+      "step": 31380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.983513771430452e-05,
+      "loss": 1.9756,
+      "step": 31390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.983184704991938e-05,
+      "loss": 1.7692,
+      "step": 31400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.982855638553424e-05,
+      "loss": 1.6715,
+      "step": 31410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.98252657211491e-05,
+      "loss": 1.7532,
+      "step": 31420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.982197505676396e-05,
+      "loss": 1.7201,
+      "step": 31430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9818684392378826e-05,
+      "loss": 1.7704,
+      "step": 31440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.981539372799368e-05,
+      "loss": 1.6562,
+      "step": 31450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.981210306360855e-05,
+      "loss": 1.7326,
+      "step": 31460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.98088123992234e-05,
+      "loss": 1.6695,
+      "step": 31470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9805521734838264e-05,
+      "loss": 1.7522,
+      "step": 31480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9802231070453125e-05,
+      "loss": 1.6466,
+      "step": 31490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.979894040606799e-05,
+      "loss": 1.6791,
+      "step": 31500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.979564974168285e-05,
+      "loss": 1.5409,
+      "step": 31510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.979235907729771e-05,
+      "loss": 1.8345,
+      "step": 31520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9789068412912564e-05,
+      "loss": 1.6545,
+      "step": 31530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.978577774852743e-05,
+      "loss": 1.8361,
+      "step": 31540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9782487084142286e-05,
+      "loss": 1.6542,
+      "step": 31550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9779196419757154e-05,
+      "loss": 1.6093,
+      "step": 31560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977590575537201e-05,
+      "loss": 1.7084,
+      "step": 31570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977261509098687e-05,
+      "loss": 1.6015,
+      "step": 31580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976932442660173e-05,
+      "loss": 1.7682,
+      "step": 31590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976603376221659e-05,
+      "loss": 1.5684,
+      "step": 31600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9762743097831454e-05,
+      "loss": 1.6052,
+      "step": 31610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9759452433446315e-05,
+      "loss": 1.6344,
+      "step": 31620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9756161769061176e-05,
+      "loss": 1.8031,
+      "step": 31630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975287110467604e-05,
+      "loss": 1.6566,
+      "step": 31640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974958044029089e-05,
+      "loss": 1.5002,
+      "step": 31650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974628977590576e-05,
+      "loss": 1.7029,
+      "step": 31660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9742999111520615e-05,
+      "loss": 1.8186,
+      "step": 31670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9739708447135476e-05,
+      "loss": 1.624,
+      "step": 31680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973641778275034e-05,
+      "loss": 1.8009,
+      "step": 31690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97331271183652e-05,
+      "loss": 1.8091,
+      "step": 31700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972983645398006e-05,
+      "loss": 1.6684,
+      "step": 31710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972654578959492e-05,
+      "loss": 1.7648,
+      "step": 31720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972325512520978e-05,
+      "loss": 1.7135,
+      "step": 31730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9719964460824644e-05,
+      "loss": 1.6522,
+      "step": 31740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9716673796439505e-05,
+      "loss": 1.599,
+      "step": 31750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9713383132054366e-05,
+      "loss": 1.6359,
+      "step": 31760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.971009246766923e-05,
+      "loss": 1.4387,
+      "step": 31770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.970680180328408e-05,
+      "loss": 1.6951,
+      "step": 31780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.970351113889895e-05,
+      "loss": 1.6261,
+      "step": 31790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9700220474513804e-05,
+      "loss": 1.7602,
+      "step": 31800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9696929810128666e-05,
+      "loss": 1.7518,
+      "step": 31810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.969363914574353e-05,
+      "loss": 2.0051,
+      "step": 31820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.969034848135839e-05,
+      "loss": 1.6036,
+      "step": 31830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.968705781697325e-05,
+      "loss": 1.6009,
+      "step": 31840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.968376715258811e-05,
+      "loss": 1.4053,
+      "step": 31850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.968047648820297e-05,
+      "loss": 1.6279,
+      "step": 31860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.967718582381783e-05,
+      "loss": 1.5865,
+      "step": 31870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.967389515943269e-05,
+      "loss": 1.687,
+      "step": 31880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9670604495047556e-05,
+      "loss": 1.6064,
+      "step": 31890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.966731383066241e-05,
+      "loss": 1.7132,
+      "step": 31900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.966402316627728e-05,
+      "loss": 1.6503,
+      "step": 31910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.966073250189213e-05,
+      "loss": 1.7466,
+      "step": 31920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9657441837506994e-05,
+      "loss": 1.6046,
+      "step": 31930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9654151173121855e-05,
+      "loss": 1.523,
+      "step": 31940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.965086050873672e-05,
+      "loss": 1.531,
+      "step": 31950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.964756984435158e-05,
+      "loss": 1.6622,
+      "step": 31960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.964427917996644e-05,
+      "loss": 1.5637,
+      "step": 31970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9640988515581294e-05,
+      "loss": 1.7234,
+      "step": 31980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.963769785119616e-05,
+      "loss": 1.6823,
+      "step": 31990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9634407186811016e-05,
+      "loss": 1.5447,
+      "step": 32000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9631116522425884e-05,
+      "loss": 1.8157,
+      "step": 32010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.962782585804074e-05,
+      "loss": 1.8234,
+      "step": 32020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.96245351936556e-05,
+      "loss": 1.7749,
+      "step": 32030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.962124452927046e-05,
+      "loss": 1.7341,
+      "step": 32040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.961795386488532e-05,
+      "loss": 1.6524,
+      "step": 32050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9614663200500184e-05,
+      "loss": 1.6629,
+      "step": 32060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9611372536115045e-05,
+      "loss": 1.6001,
+      "step": 32070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.96080818717299e-05,
+      "loss": 1.7157,
+      "step": 32080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960479120734477e-05,
+      "loss": 1.6131,
+      "step": 32090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960150054295962e-05,
+      "loss": 1.8567,
+      "step": 32100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959820987857449e-05,
+      "loss": 1.6501,
+      "step": 32110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9594919214189345e-05,
+      "loss": 1.7536,
+      "step": 32120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9591628549804206e-05,
+      "loss": 1.5479,
+      "step": 32130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958833788541907e-05,
+      "loss": 1.5387,
+      "step": 32140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958504722103393e-05,
+      "loss": 1.7947,
+      "step": 32150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958175655664879e-05,
+      "loss": 1.8742,
+      "step": 32160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957846589226365e-05,
+      "loss": 1.7786,
+      "step": 32170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9575175227878505e-05,
+      "loss": 1.6672,
+      "step": 32180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9571884563493374e-05,
+      "loss": 1.5287,
+      "step": 32190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956859389910823e-05,
+      "loss": 1.6446,
+      "step": 32200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9565303234723096e-05,
+      "loss": 1.6909,
+      "step": 32210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956201257033795e-05,
+      "loss": 1.6462,
+      "step": 32220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955872190595281e-05,
+      "loss": 1.5402,
+      "step": 32230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955543124156767e-05,
+      "loss": 1.6093,
+      "step": 32240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9552140577182534e-05,
+      "loss": 1.7236,
+      "step": 32250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9548849912797396e-05,
+      "loss": 1.7281,
+      "step": 32260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.954555924841226e-05,
+      "loss": 1.5103,
+      "step": 32270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.954226858402711e-05,
+      "loss": 1.5285,
+      "step": 32280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.953897791964198e-05,
+      "loss": 1.5479,
+      "step": 32290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9535687255256834e-05,
+      "loss": 1.694,
+      "step": 32300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.95323965908717e-05,
+      "loss": 1.6185,
+      "step": 32310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9529105926486556e-05,
+      "loss": 1.5546,
+      "step": 32320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.952581526210142e-05,
+      "loss": 1.847,
+      "step": 32330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.952252459771628e-05,
+      "loss": 1.5128,
+      "step": 32340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.951923393333114e-05,
+      "loss": 1.6895,
+      "step": 32350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9515943268946e-05,
+      "loss": 1.564,
+      "step": 32360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.951265260456086e-05,
+      "loss": 1.6477,
+      "step": 32370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9509361940175724e-05,
+      "loss": 1.9189,
+      "step": 32380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9506071275790585e-05,
+      "loss": 1.6523,
+      "step": 32390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9502780611405447e-05,
+      "loss": 1.8036,
+      "step": 32400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.949948994702031e-05,
+      "loss": 1.6603,
+      "step": 32410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.949619928263516e-05,
+      "loss": 1.5669,
+      "step": 32420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9492908618250024e-05,
+      "loss": 1.5624,
+      "step": 32430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9489617953864885e-05,
+      "loss": 1.7342,
+      "step": 32440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9486327289479746e-05,
+      "loss": 1.6411,
+      "step": 32450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.948303662509461e-05,
+      "loss": 1.5624,
+      "step": 32460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.947974596070947e-05,
+      "loss": 1.7478,
+      "step": 32470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.947645529632433e-05,
+      "loss": 1.6917,
+      "step": 32480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.947316463193919e-05,
+      "loss": 1.6409,
+      "step": 32490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.946987396755405e-05,
+      "loss": 1.5627,
+      "step": 32500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9466583303168914e-05,
+      "loss": 1.6262,
+      "step": 32510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9463292638783775e-05,
+      "loss": 1.7005,
+      "step": 32520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.946000197439863e-05,
+      "loss": 1.7213,
+      "step": 32530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94567113100135e-05,
+      "loss": 1.6982,
+      "step": 32540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.945342064562835e-05,
+      "loss": 1.689,
+      "step": 32550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.945012998124322e-05,
+      "loss": 1.8774,
+      "step": 32560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9446839316858075e-05,
+      "loss": 1.6901,
+      "step": 32570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9443548652472936e-05,
+      "loss": 1.7749,
+      "step": 32580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94402579880878e-05,
+      "loss": 1.8342,
+      "step": 32590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943696732370266e-05,
+      "loss": 1.655,
+      "step": 32600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943367665931752e-05,
+      "loss": 1.7555,
+      "step": 32610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943038599493238e-05,
+      "loss": 1.6324,
+      "step": 32620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9427095330547235e-05,
+      "loss": 1.8464,
+      "step": 32630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9423804666162103e-05,
+      "loss": 1.487,
+      "step": 32640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942051400177696e-05,
+      "loss": 1.6033,
+      "step": 32650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9417223337391826e-05,
+      "loss": 1.8178,
+      "step": 32660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941393267300668e-05,
+      "loss": 1.7489,
+      "step": 32670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941064200862154e-05,
+      "loss": 1.646,
+      "step": 32680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94073513442364e-05,
+      "loss": 1.6565,
+      "step": 32690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9404060679851264e-05,
+      "loss": 1.584,
+      "step": 32700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9400770015466126e-05,
+      "loss": 1.68,
+      "step": 32710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939747935108099e-05,
+      "loss": 1.5441,
+      "step": 32720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939418868669584e-05,
+      "loss": 1.8234,
+      "step": 32730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939089802231071e-05,
+      "loss": 1.4769,
+      "step": 32740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9387607357925564e-05,
+      "loss": 1.562,
+      "step": 32750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938431669354043e-05,
+      "loss": 1.5042,
+      "step": 32760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9381026029155286e-05,
+      "loss": 1.6644,
+      "step": 32770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.937773536477015e-05,
+      "loss": 1.5842,
+      "step": 32780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.937444470038501e-05,
+      "loss": 1.5426,
+      "step": 32790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.937115403599987e-05,
+      "loss": 1.4372,
+      "step": 32800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.936786337161473e-05,
+      "loss": 1.6267,
+      "step": 32810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.936457270722959e-05,
+      "loss": 1.7675,
+      "step": 32820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.936128204284445e-05,
+      "loss": 1.5923,
+      "step": 32830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9357991378459315e-05,
+      "loss": 2.0032,
+      "step": 32840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.935470071407417e-05,
+      "loss": 1.6208,
+      "step": 32850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.935141004968904e-05,
+      "loss": 1.5491,
+      "step": 32860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.934811938530389e-05,
+      "loss": 1.5985,
+      "step": 32870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9344828720918754e-05,
+      "loss": 1.7024,
+      "step": 32880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9341538056533615e-05,
+      "loss": 1.6023,
+      "step": 32890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9338247392148476e-05,
+      "loss": 1.5091,
+      "step": 32900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.933495672776334e-05,
+      "loss": 1.6867,
+      "step": 32910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.93316660633782e-05,
+      "loss": 1.699,
+      "step": 32920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.932837539899305e-05,
+      "loss": 1.674,
+      "step": 32930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.932508473460792e-05,
+      "loss": 1.4071,
+      "step": 32940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9321794070222776e-05,
+      "loss": 1.7135,
+      "step": 32950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9318503405837644e-05,
+      "loss": 1.5759,
+      "step": 32960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.93152127414525e-05,
+      "loss": 1.7025,
+      "step": 32970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.931192207706736e-05,
+      "loss": 1.8005,
+      "step": 32980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.930863141268222e-05,
+      "loss": 1.6247,
+      "step": 32990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.930534074829708e-05,
+      "loss": 1.7257,
+      "step": 33000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.930205008391194e-05,
+      "loss": 1.7441,
+      "step": 33010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9298759419526805e-05,
+      "loss": 1.5931,
+      "step": 33020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.929546875514166e-05,
+      "loss": 1.6646,
+      "step": 33030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.929217809075653e-05,
+      "loss": 1.8599,
+      "step": 33040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.928888742637138e-05,
+      "loss": 1.802,
+      "step": 33050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.928559676198625e-05,
+      "loss": 1.668,
+      "step": 33060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9282306097601104e-05,
+      "loss": 1.5114,
+      "step": 33070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9279015433215965e-05,
+      "loss": 1.51,
+      "step": 33080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927572476883083e-05,
+      "loss": 1.5941,
+      "step": 33090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927243410444569e-05,
+      "loss": 1.6189,
+      "step": 33100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926914344006055e-05,
+      "loss": 1.5407,
+      "step": 33110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926585277567541e-05,
+      "loss": 1.6043,
+      "step": 33120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926256211129027e-05,
+      "loss": 1.5609,
+      "step": 33130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925927144690513e-05,
+      "loss": 1.7931,
+      "step": 33140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9255980782519994e-05,
+      "loss": 1.7457,
+      "step": 33150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9252690118134855e-05,
+      "loss": 1.4554,
+      "step": 33160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924939945374972e-05,
+      "loss": 1.6696,
+      "step": 33170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924610878936458e-05,
+      "loss": 1.8713,
+      "step": 33180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924281812497943e-05,
+      "loss": 1.7709,
+      "step": 33190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9239527460594294e-05,
+      "loss": 1.6506,
+      "step": 33200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9236236796209155e-05,
+      "loss": 1.6698,
+      "step": 33210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9232946131824016e-05,
+      "loss": 1.541,
+      "step": 33220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922965546743888e-05,
+      "loss": 1.8124,
+      "step": 33230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922636480305374e-05,
+      "loss": 1.6743,
+      "step": 33240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92230741386686e-05,
+      "loss": 1.6634,
+      "step": 33250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921978347428346e-05,
+      "loss": 1.6414,
+      "step": 33260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921649280989832e-05,
+      "loss": 1.6085,
+      "step": 33270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9213202145513184e-05,
+      "loss": 1.6821,
+      "step": 33280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9209911481128045e-05,
+      "loss": 1.5713,
+      "step": 33290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.92066208167429e-05,
+      "loss": 1.4441,
+      "step": 33300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.920333015235777e-05,
+      "loss": 1.7872,
+      "step": 33310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.920003948797262e-05,
+      "loss": 1.7686,
+      "step": 33320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.919674882358749e-05,
+      "loss": 1.716,
+      "step": 33330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9193458159202345e-05,
+      "loss": 1.6434,
+      "step": 33340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9190167494817206e-05,
+      "loss": 1.779,
+      "step": 33350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.918687683043207e-05,
+      "loss": 1.4767,
+      "step": 33360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.918358616604693e-05,
+      "loss": 1.7853,
+      "step": 33370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.918029550166179e-05,
+      "loss": 1.6077,
+      "step": 33380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.917700483727665e-05,
+      "loss": 1.5984,
+      "step": 33390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9173714172891506e-05,
+      "loss": 1.6724,
+      "step": 33400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9170423508506374e-05,
+      "loss": 1.9184,
+      "step": 33410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.916713284412123e-05,
+      "loss": 1.5274,
+      "step": 33420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9163842179736096e-05,
+      "loss": 1.4394,
+      "step": 33430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.916055151535095e-05,
+      "loss": 1.6482,
+      "step": 33440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.915726085096581e-05,
+      "loss": 1.6068,
+      "step": 33450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.915397018658067e-05,
+      "loss": 1.5834,
+      "step": 33460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9150679522195534e-05,
+      "loss": 1.5525,
+      "step": 33470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9147388857810396e-05,
+      "loss": 1.5963,
+      "step": 33480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.914409819342526e-05,
+      "loss": 1.6606,
+      "step": 33490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.914080752904011e-05,
+      "loss": 1.7108,
+      "step": 33500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.913751686465498e-05,
+      "loss": 1.7351,
+      "step": 33510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9134226200269834e-05,
+      "loss": 1.7331,
+      "step": 33520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.91309355358847e-05,
+      "loss": 1.7515,
+      "step": 33530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9127644871499557e-05,
+      "loss": 1.7995,
+      "step": 33540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.912435420711442e-05,
+      "loss": 1.6067,
+      "step": 33550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.912106354272928e-05,
+      "loss": 1.6746,
+      "step": 33560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.911777287834414e-05,
+      "loss": 1.8332,
+      "step": 33570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9114482213959e-05,
+      "loss": 1.6994,
+      "step": 33580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.911119154957386e-05,
+      "loss": 1.6483,
+      "step": 33590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910790088518872e-05,
+      "loss": 1.5828,
+      "step": 33600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9104610220803585e-05,
+      "loss": 1.7451,
+      "step": 33610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910131955641844e-05,
+      "loss": 1.5822,
+      "step": 33620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909802889203331e-05,
+      "loss": 1.5985,
+      "step": 33630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909473822764816e-05,
+      "loss": 1.7579,
+      "step": 33640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9091447563263024e-05,
+      "loss": 1.6946,
+      "step": 33650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9088156898877885e-05,
+      "loss": 1.4078,
+      "step": 33660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9084866234492746e-05,
+      "loss": 1.5576,
+      "step": 33670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908157557010761e-05,
+      "loss": 1.5495,
+      "step": 33680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907828490572247e-05,
+      "loss": 1.4726,
+      "step": 33690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907499424133732e-05,
+      "loss": 1.7227,
+      "step": 33700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907170357695219e-05,
+      "loss": 1.6138,
+      "step": 33710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9068412912567046e-05,
+      "loss": 1.661,
+      "step": 33720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9065122248181914e-05,
+      "loss": 1.6031,
+      "step": 33730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906183158379677e-05,
+      "loss": 1.6524,
+      "step": 33740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905854091941163e-05,
+      "loss": 1.7609,
+      "step": 33750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905525025502649e-05,
+      "loss": 1.6017,
+      "step": 33760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905195959064135e-05,
+      "loss": 1.6614,
+      "step": 33770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9048668926256213e-05,
+      "loss": 1.6583,
+      "step": 33780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9045378261871075e-05,
+      "loss": 1.6161,
+      "step": 33790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.904208759748593e-05,
+      "loss": 1.6762,
+      "step": 33800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.90387969331008e-05,
+      "loss": 1.704,
+      "step": 33810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.903550626871565e-05,
+      "loss": 1.6836,
+      "step": 33820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.903221560433052e-05,
+      "loss": 1.835,
+      "step": 33830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9028924939945374e-05,
+      "loss": 1.8692,
+      "step": 33840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9025634275560236e-05,
+      "loss": 1.4791,
+      "step": 33850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.90223436111751e-05,
+      "loss": 1.7546,
+      "step": 33860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.901905294678996e-05,
+      "loss": 1.6375,
+      "step": 33870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.901576228240482e-05,
+      "loss": 1.6162,
+      "step": 33880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.901247161801968e-05,
+      "loss": 2.0035,
+      "step": 33890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.900918095363454e-05,
+      "loss": 1.8275,
+      "step": 33900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.90058902892494e-05,
+      "loss": 1.6828,
+      "step": 33910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9002599624864264e-05,
+      "loss": 1.6109,
+      "step": 33920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8999308960479126e-05,
+      "loss": 1.776,
+      "step": 33930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.899601829609398e-05,
+      "loss": 1.7571,
+      "step": 33940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.899272763170884e-05,
+      "loss": 1.6692,
+      "step": 33950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89894369673237e-05,
+      "loss": 1.7108,
+      "step": 33960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8986146302938564e-05,
+      "loss": 1.6331,
+      "step": 33970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8982855638553425e-05,
+      "loss": 1.4823,
+      "step": 33980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8979564974168286e-05,
+      "loss": 1.6796,
+      "step": 33990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.897627430978315e-05,
+      "loss": 1.612,
+      "step": 34000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.897298364539801e-05,
+      "loss": 1.7,
+      "step": 34010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.896969298101287e-05,
+      "loss": 1.862,
+      "step": 34020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.896640231662773e-05,
+      "loss": 1.5893,
+      "step": 34030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.896311165224259e-05,
+      "loss": 1.596,
+      "step": 34040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.895982098785745e-05,
+      "loss": 1.606,
+      "step": 34050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8956530323472315e-05,
+      "loss": 1.7596,
+      "step": 34060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.895323965908717e-05,
+      "loss": 1.5941,
+      "step": 34070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894994899470204e-05,
+      "loss": 1.6877,
+      "step": 34080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894665833031689e-05,
+      "loss": 1.7404,
+      "step": 34090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8943367665931754e-05,
+      "loss": 1.6026,
+      "step": 34100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8940077001546615e-05,
+      "loss": 1.6653,
+      "step": 34110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8936786337161476e-05,
+      "loss": 1.7642,
+      "step": 34120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893349567277634e-05,
+      "loss": 1.6732,
+      "step": 34130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89302050083912e-05,
+      "loss": 1.7122,
+      "step": 34140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892691434400605e-05,
+      "loss": 1.609,
+      "step": 34150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892362367962092e-05,
+      "loss": 1.5409,
+      "step": 34160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8920333015235776e-05,
+      "loss": 1.5985,
+      "step": 34170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8917042350850644e-05,
+      "loss": 1.5916,
+      "step": 34180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89137516864655e-05,
+      "loss": 1.5716,
+      "step": 34190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891046102208036e-05,
+      "loss": 1.5811,
+      "step": 34200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890717035769522e-05,
+      "loss": 1.6218,
+      "step": 34210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890387969331008e-05,
+      "loss": 1.7591,
+      "step": 34220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890058902892494e-05,
+      "loss": 1.5988,
+      "step": 34230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8897298364539805e-05,
+      "loss": 2.0313,
+      "step": 34240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889400770015466e-05,
+      "loss": 1.6034,
+      "step": 34250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889071703576953e-05,
+      "loss": 1.6158,
+      "step": 34260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888742637138438e-05,
+      "loss": 1.6176,
+      "step": 34270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888413570699925e-05,
+      "loss": 1.4884,
+      "step": 34280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8880845042614104e-05,
+      "loss": 1.536,
+      "step": 34290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8877554378228965e-05,
+      "loss": 1.8299,
+      "step": 34300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.887426371384383e-05,
+      "loss": 1.598,
+      "step": 34310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.887097304945869e-05,
+      "loss": 1.5469,
+      "step": 34320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.886768238507355e-05,
+      "loss": 1.897,
+      "step": 34330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.886439172068841e-05,
+      "loss": 1.7514,
+      "step": 34340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8861101056303265e-05,
+      "loss": 1.6318,
+      "step": 34350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.885781039191813e-05,
+      "loss": 1.6435,
+      "step": 34360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.885451972753299e-05,
+      "loss": 1.6887,
+      "step": 34370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8851229063147856e-05,
+      "loss": 1.8414,
+      "step": 34380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.884793839876271e-05,
+      "loss": 1.6903,
+      "step": 34390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.884464773437757e-05,
+      "loss": 1.4877,
+      "step": 34400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.884135706999243e-05,
+      "loss": 1.5616,
+      "step": 34410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8838066405607294e-05,
+      "loss": 1.4203,
+      "step": 34420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8834775741222155e-05,
+      "loss": 1.6256,
+      "step": 34430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8831485076837016e-05,
+      "loss": 1.7357,
+      "step": 34440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.882819441245187e-05,
+      "loss": 1.771,
+      "step": 34450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.882490374806674e-05,
+      "loss": 1.6005,
+      "step": 34460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8821613083681593e-05,
+      "loss": 1.5635,
+      "step": 34470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.881832241929646e-05,
+      "loss": 1.5095,
+      "step": 34480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8815031754911316e-05,
+      "loss": 1.6687,
+      "step": 34490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.881174109052618e-05,
+      "loss": 1.6875,
+      "step": 34500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.880845042614104e-05,
+      "loss": 1.7135,
+      "step": 34510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.88051597617559e-05,
+      "loss": 1.7623,
+      "step": 34520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.880186909737076e-05,
+      "loss": 1.887,
+      "step": 34530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.879857843298562e-05,
+      "loss": 1.7419,
+      "step": 34540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.879528776860048e-05,
+      "loss": 1.8411,
+      "step": 34550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8791997104215345e-05,
+      "loss": 1.7516,
+      "step": 34560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87887064398302e-05,
+      "loss": 1.6435,
+      "step": 34570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.878541577544507e-05,
+      "loss": 1.5946,
+      "step": 34580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.878212511105992e-05,
+      "loss": 1.646,
+      "step": 34590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877883444667478e-05,
+      "loss": 1.6455,
+      "step": 34600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8775543782289644e-05,
+      "loss": 1.6078,
+      "step": 34610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8772253117904506e-05,
+      "loss": 1.6427,
+      "step": 34620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876896245351937e-05,
+      "loss": 1.6544,
+      "step": 34630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876567178913423e-05,
+      "loss": 1.6945,
+      "step": 34640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876238112474909e-05,
+      "loss": 1.675,
+      "step": 34650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875909046036395e-05,
+      "loss": 1.5982,
+      "step": 34660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875579979597881e-05,
+      "loss": 1.871,
+      "step": 34670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875250913159367e-05,
+      "loss": 1.751,
+      "step": 34680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8749218467208535e-05,
+      "loss": 1.6391,
+      "step": 34690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874592780282339e-05,
+      "loss": 1.6602,
+      "step": 34700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874263713843825e-05,
+      "loss": 1.7173,
+      "step": 34710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873934647405311e-05,
+      "loss": 1.6034,
+      "step": 34720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873605580966797e-05,
+      "loss": 1.5401,
+      "step": 34730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8732765145282834e-05,
+      "loss": 1.5132,
+      "step": 34740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8729474480897695e-05,
+      "loss": 1.7435,
+      "step": 34750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872618381651256e-05,
+      "loss": 1.716,
+      "step": 34760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872289315212742e-05,
+      "loss": 1.6447,
+      "step": 34770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871960248774228e-05,
+      "loss": 1.5972,
+      "step": 34780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871631182335714e-05,
+      "loss": 1.8141,
+      "step": 34790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8713021158971995e-05,
+      "loss": 1.7553,
+      "step": 34800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.870973049458686e-05,
+      "loss": 1.5568,
+      "step": 34810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.870643983020172e-05,
+      "loss": 1.6822,
+      "step": 34820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8703149165816586e-05,
+      "loss": 1.5959,
+      "step": 34830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.869985850143144e-05,
+      "loss": 1.6635,
+      "step": 34840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86965678370463e-05,
+      "loss": 1.8728,
+      "step": 34850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.869327717266116e-05,
+      "loss": 1.7824,
+      "step": 34860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8689986508276024e-05,
+      "loss": 1.5961,
+      "step": 34870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8686695843890885e-05,
+      "loss": 1.5256,
+      "step": 34880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8683405179505746e-05,
+      "loss": 1.5779,
+      "step": 34890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86801145151206e-05,
+      "loss": 1.799,
+      "step": 34900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.867682385073547e-05,
+      "loss": 1.678,
+      "step": 34910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.867353318635032e-05,
+      "loss": 1.4706,
+      "step": 34920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.867024252196519e-05,
+      "loss": 1.6566,
+      "step": 34930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8666951857580046e-05,
+      "loss": 1.6978,
+      "step": 34940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.866366119319491e-05,
+      "loss": 1.5341,
+      "step": 34950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.866037052880977e-05,
+      "loss": 1.7503,
+      "step": 34960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.865707986442463e-05,
+      "loss": 1.6456,
+      "step": 34970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.865378920003949e-05,
+      "loss": 2.0237,
+      "step": 34980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.865049853565435e-05,
+      "loss": 1.7485,
+      "step": 34990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.864720787126921e-05,
+      "loss": 1.5728,
+      "step": 35000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8643917206884075e-05,
+      "loss": 1.7189,
+      "step": 35010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.864062654249893e-05,
+      "loss": 1.6611,
+      "step": 35020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86373358781138e-05,
+      "loss": 1.765,
+      "step": 35030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.863404521372865e-05,
+      "loss": 1.4831,
+      "step": 35040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.863075454934351e-05,
+      "loss": 1.7837,
+      "step": 35050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8627463884958374e-05,
+      "loss": 1.6636,
+      "step": 35060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8624173220573236e-05,
+      "loss": 1.6106,
+      "step": 35070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86208825561881e-05,
+      "loss": 1.7394,
+      "step": 35080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.861759189180296e-05,
+      "loss": 1.5349,
+      "step": 35090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.861430122741781e-05,
+      "loss": 1.6048,
+      "step": 35100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.861101056303268e-05,
+      "loss": 1.6637,
+      "step": 35110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8607719898647535e-05,
+      "loss": 1.7309,
+      "step": 35120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86044292342624e-05,
+      "loss": 1.6577,
+      "step": 35130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860113856987726e-05,
+      "loss": 1.7164,
+      "step": 35140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859784790549212e-05,
+      "loss": 1.6767,
+      "step": 35150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859455724110698e-05,
+      "loss": 1.6138,
+      "step": 35160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859126657672184e-05,
+      "loss": 1.5396,
+      "step": 35170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85879759123367e-05,
+      "loss": 1.659,
+      "step": 35180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8584685247951564e-05,
+      "loss": 1.4979,
+      "step": 35190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858139458356642e-05,
+      "loss": 1.6158,
+      "step": 35200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8578103919181287e-05,
+      "loss": 1.599,
+      "step": 35210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857481325479614e-05,
+      "loss": 1.6138,
+      "step": 35220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857152259041101e-05,
+      "loss": 1.7297,
+      "step": 35230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8568231926025864e-05,
+      "loss": 1.6626,
+      "step": 35240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8564941261640725e-05,
+      "loss": 1.6864,
+      "step": 35250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8561650597255586e-05,
+      "loss": 1.4298,
+      "step": 35260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855835993287045e-05,
+      "loss": 1.7051,
+      "step": 35270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855506926848531e-05,
+      "loss": 1.731,
+      "step": 35280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855177860410017e-05,
+      "loss": 1.6751,
+      "step": 35290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8548487939715024e-05,
+      "loss": 1.7924,
+      "step": 35300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854519727532989e-05,
+      "loss": 1.6095,
+      "step": 35310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.854190661094475e-05,
+      "loss": 1.7424,
+      "step": 35320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8538615946559615e-05,
+      "loss": 1.7001,
+      "step": 35330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.853532528217447e-05,
+      "loss": 1.6721,
+      "step": 35340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.853203461778933e-05,
+      "loss": 1.578,
+      "step": 35350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.852874395340419e-05,
+      "loss": 1.6571,
+      "step": 35360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.852545328901905e-05,
+      "loss": 1.4571,
+      "step": 35370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8522162624633915e-05,
+      "loss": 1.6926,
+      "step": 35380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8518871960248776e-05,
+      "loss": 1.6548,
+      "step": 35390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.851558129586364e-05,
+      "loss": 1.6814,
+      "step": 35400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.85122906314785e-05,
+      "loss": 1.6341,
+      "step": 35410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.850899996709336e-05,
+      "loss": 1.5672,
+      "step": 35420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.850570930270822e-05,
+      "loss": 1.5986,
+      "step": 35430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.850241863832308e-05,
+      "loss": 1.6455,
+      "step": 35440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.849912797393794e-05,
+      "loss": 1.5176,
+      "step": 35450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8495837309552805e-05,
+      "loss": 1.6708,
+      "step": 35460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.849254664516766e-05,
+      "loss": 1.657,
+      "step": 35470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.848925598078252e-05,
+      "loss": 1.5489,
+      "step": 35480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.848596531639738e-05,
+      "loss": 1.6029,
+      "step": 35490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.848267465201224e-05,
+      "loss": 1.7619,
+      "step": 35500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8479383987627104e-05,
+      "loss": 1.6645,
+      "step": 35510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8476093323241966e-05,
+      "loss": 1.5362,
+      "step": 35520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.847280265885683e-05,
+      "loss": 1.5738,
+      "step": 35530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.846951199447169e-05,
+      "loss": 1.6762,
+      "step": 35540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.846622133008654e-05,
+      "loss": 1.744,
+      "step": 35550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.846293066570141e-05,
+      "loss": 1.6885,
+      "step": 35560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8459640001316265e-05,
+      "loss": 1.6917,
+      "step": 35570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.845634933693113e-05,
+      "loss": 1.5494,
+      "step": 35580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.845305867254599e-05,
+      "loss": 1.5458,
+      "step": 35590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844976800816085e-05,
+      "loss": 1.7389,
+      "step": 35600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844647734377571e-05,
+      "loss": 1.7966,
+      "step": 35610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844318667939057e-05,
+      "loss": 1.6293,
+      "step": 35620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843989601500543e-05,
+      "loss": 1.6641,
+      "step": 35630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8436605350620294e-05,
+      "loss": 1.6702,
+      "step": 35640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843331468623515e-05,
+      "loss": 1.7387,
+      "step": 35650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8430024021850016e-05,
+      "loss": 1.6543,
+      "step": 35660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842673335746487e-05,
+      "loss": 1.6294,
+      "step": 35670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842344269307974e-05,
+      "loss": 1.6178,
+      "step": 35680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8420152028694594e-05,
+      "loss": 1.5129,
+      "step": 35690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8416861364309455e-05,
+      "loss": 1.6816,
+      "step": 35700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8413570699924316e-05,
+      "loss": 1.5338,
+      "step": 35710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841028003553918e-05,
+      "loss": 1.7152,
+      "step": 35720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840698937115404e-05,
+      "loss": 1.6276,
+      "step": 35730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84036987067689e-05,
+      "loss": 1.5922,
+      "step": 35740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8400408042383754e-05,
+      "loss": 1.6996,
+      "step": 35750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839711737799862e-05,
+      "loss": 1.7406,
+      "step": 35760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839382671361348e-05,
+      "loss": 1.5849,
+      "step": 35770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8390536049228345e-05,
+      "loss": 1.6677,
+      "step": 35780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83872453848432e-05,
+      "loss": 1.5359,
+      "step": 35790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838395472045806e-05,
+      "loss": 1.6119,
+      "step": 35800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838066405607292e-05,
+      "loss": 1.5457,
+      "step": 35810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837737339168778e-05,
+      "loss": 1.6354,
+      "step": 35820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8374082727302644e-05,
+      "loss": 1.6476,
+      "step": 35830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8370792062917506e-05,
+      "loss": 1.8843,
+      "step": 35840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.836750139853236e-05,
+      "loss": 1.5982,
+      "step": 35850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.836421073414723e-05,
+      "loss": 1.6279,
+      "step": 35860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.836092006976208e-05,
+      "loss": 1.6826,
+      "step": 35870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.835762940537695e-05,
+      "loss": 1.7414,
+      "step": 35880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8354338740991805e-05,
+      "loss": 1.737,
+      "step": 35890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8351048076606667e-05,
+      "loss": 1.4759,
+      "step": 35900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.834775741222153e-05,
+      "loss": 1.6707,
+      "step": 35910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.834446674783639e-05,
+      "loss": 1.5956,
+      "step": 35920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.834117608345125e-05,
+      "loss": 1.844,
+      "step": 35930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.833788541906611e-05,
+      "loss": 1.8089,
+      "step": 35940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8334594754680966e-05,
+      "loss": 1.4535,
+      "step": 35950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8331304090295834e-05,
+      "loss": 1.6257,
+      "step": 35960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.832801342591069e-05,
+      "loss": 1.5349,
+      "step": 35970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.832472276152556e-05,
+      "loss": 1.4892,
+      "step": 35980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.832143209714041e-05,
+      "loss": 1.7099,
+      "step": 35990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.831814143275527e-05,
+      "loss": 1.4595,
+      "step": 36000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8314850768370134e-05,
+      "loss": 1.4989,
+      "step": 36010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8311560103984995e-05,
+      "loss": 1.6457,
+      "step": 36020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8308269439599856e-05,
+      "loss": 1.752,
+      "step": 36030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830497877521472e-05,
+      "loss": 1.5666,
+      "step": 36040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830168811082958e-05,
+      "loss": 1.5528,
+      "step": 36050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829839744644444e-05,
+      "loss": 1.6983,
+      "step": 36060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8295106782059295e-05,
+      "loss": 1.6853,
+      "step": 36070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829181611767416e-05,
+      "loss": 1.5655,
+      "step": 36080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.828852545328902e-05,
+      "loss": 1.7606,
+      "step": 36090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8285234788903885e-05,
+      "loss": 1.7653,
+      "step": 36100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.828194412451874e-05,
+      "loss": 1.6291,
+      "step": 36110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82786534601336e-05,
+      "loss": 1.5526,
+      "step": 36120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827536279574846e-05,
+      "loss": 1.7015,
+      "step": 36130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8272072131363323e-05,
+      "loss": 1.7412,
+      "step": 36140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8268781466978185e-05,
+      "loss": 1.6145,
+      "step": 36150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8265490802593046e-05,
+      "loss": 1.6504,
+      "step": 36160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826220013820791e-05,
+      "loss": 1.6367,
+      "step": 36170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825890947382277e-05,
+      "loss": 1.5158,
+      "step": 36180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825561880943763e-05,
+      "loss": 1.6978,
+      "step": 36190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825232814505249e-05,
+      "loss": 1.8461,
+      "step": 36200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824903748066735e-05,
+      "loss": 1.7066,
+      "step": 36210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824574681628221e-05,
+      "loss": 1.681,
+      "step": 36220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824245615189707e-05,
+      "loss": 1.5874,
+      "step": 36230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823916548751193e-05,
+      "loss": 1.5544,
+      "step": 36240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823587482312679e-05,
+      "loss": 1.7088,
+      "step": 36250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823258415874165e-05,
+      "loss": 1.9327,
+      "step": 36260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822929349435651e-05,
+      "loss": 1.6527,
+      "step": 36270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8226002829971374e-05,
+      "loss": 1.5874,
+      "step": 36280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8222712165586236e-05,
+      "loss": 1.6513,
+      "step": 36290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82194215012011e-05,
+      "loss": 1.6426,
+      "step": 36300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821613083681596e-05,
+      "loss": 1.6611,
+      "step": 36310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821284017243081e-05,
+      "loss": 1.4429,
+      "step": 36320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820954950804568e-05,
+      "loss": 1.5765,
+      "step": 36330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8206258843660535e-05,
+      "loss": 1.4378,
+      "step": 36340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82029681792754e-05,
+      "loss": 1.652,
+      "step": 36350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819967751489026e-05,
+      "loss": 1.5454,
+      "step": 36360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819638685050512e-05,
+      "loss": 1.6621,
+      "step": 36370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819309618611998e-05,
+      "loss": 1.6413,
+      "step": 36380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818980552173484e-05,
+      "loss": 1.6098,
+      "step": 36390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81865148573497e-05,
+      "loss": 1.5187,
+      "step": 36400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8183224192964564e-05,
+      "loss": 1.5907,
+      "step": 36410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817993352857942e-05,
+      "loss": 1.7431,
+      "step": 36420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817664286419429e-05,
+      "loss": 1.5843,
+      "step": 36430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817335219980914e-05,
+      "loss": 1.6718,
+      "step": 36440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817006153542401e-05,
+      "loss": 1.912,
+      "step": 36450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8166770871038864e-05,
+      "loss": 1.6408,
+      "step": 36460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8163480206653725e-05,
+      "loss": 1.8705,
+      "step": 36470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8160189542268586e-05,
+      "loss": 1.7346,
+      "step": 36480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815689887788345e-05,
+      "loss": 1.8391,
+      "step": 36490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815360821349831e-05,
+      "loss": 1.836,
+      "step": 36500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815031754911317e-05,
+      "loss": 1.6525,
+      "step": 36510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8147026884728025e-05,
+      "loss": 1.582,
+      "step": 36520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814373622034289e-05,
+      "loss": 1.6885,
+      "step": 36530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814044555595775e-05,
+      "loss": 1.6685,
+      "step": 36540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8137154891572615e-05,
+      "loss": 1.6818,
+      "step": 36550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813386422718747e-05,
+      "loss": 1.5531,
+      "step": 36560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813057356280233e-05,
+      "loss": 1.6409,
+      "step": 36570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812728289841719e-05,
+      "loss": 1.4627,
+      "step": 36580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812399223403205e-05,
+      "loss": 1.8468,
+      "step": 36590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8120701569646915e-05,
+      "loss": 1.573,
+      "step": 36600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8117410905261776e-05,
+      "loss": 1.52,
+      "step": 36610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811412024087663e-05,
+      "loss": 1.7536,
+      "step": 36620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81108295764915e-05,
+      "loss": 1.6502,
+      "step": 36630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810753891210635e-05,
+      "loss": 1.5034,
+      "step": 36640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810424824772122e-05,
+      "loss": 1.728,
+      "step": 36650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8100957583336075e-05,
+      "loss": 1.6592,
+      "step": 36660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809766691895094e-05,
+      "loss": 1.6623,
+      "step": 36670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80943762545658e-05,
+      "loss": 1.5817,
+      "step": 36680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809108559018066e-05,
+      "loss": 1.5553,
+      "step": 36690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808779492579552e-05,
+      "loss": 1.7502,
+      "step": 36700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808450426141038e-05,
+      "loss": 1.701,
+      "step": 36710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8081213597025236e-05,
+      "loss": 1.6631,
+      "step": 36720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8077922932640104e-05,
+      "loss": 1.6915,
+      "step": 36730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807463226825496e-05,
+      "loss": 1.6,
+      "step": 36740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807134160386983e-05,
+      "loss": 1.6548,
+      "step": 36750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806805093948468e-05,
+      "loss": 1.5853,
+      "step": 36760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806476027509954e-05,
+      "loss": 1.6233,
+      "step": 36770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8061469610714404e-05,
+      "loss": 1.5569,
+      "step": 36780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8058178946329265e-05,
+      "loss": 1.5276,
+      "step": 36790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8054888281944126e-05,
+      "loss": 1.6723,
+      "step": 36800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805159761755899e-05,
+      "loss": 1.7735,
+      "step": 36810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804830695317385e-05,
+      "loss": 1.7304,
+      "step": 36820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804501628878871e-05,
+      "loss": 1.6027,
+      "step": 36830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8041725624403565e-05,
+      "loss": 1.681,
+      "step": 36840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.803843496001843e-05,
+      "loss": 1.7285,
+      "step": 36850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.803514429563329e-05,
+      "loss": 1.5591,
+      "step": 36860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.803185363124815e-05,
+      "loss": 1.6887,
+      "step": 36870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.802856296686301e-05,
+      "loss": 1.6472,
+      "step": 36880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.802527230247787e-05,
+      "loss": 1.6016,
+      "step": 36890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.802198163809273e-05,
+      "loss": 1.5928,
+      "step": 36900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8018690973707594e-05,
+      "loss": 1.5594,
+      "step": 36910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8015400309322455e-05,
+      "loss": 1.5488,
+      "step": 36920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8012109644937316e-05,
+      "loss": 1.6003,
+      "step": 36930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.800881898055218e-05,
+      "loss": 1.6686,
+      "step": 36940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.800552831616704e-05,
+      "loss": 1.6426,
+      "step": 36950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.80022376517819e-05,
+      "loss": 1.4893,
+      "step": 36960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7998946987396754e-05,
+      "loss": 1.5782,
+      "step": 36970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.799565632301162e-05,
+      "loss": 1.637,
+      "step": 36980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.799236565862648e-05,
+      "loss": 1.6227,
+      "step": 36990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.798907499424134e-05,
+      "loss": 1.7561,
+      "step": 37000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79857843298562e-05,
+      "loss": 1.8799,
+      "step": 37010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.798249366547106e-05,
+      "loss": 1.6466,
+      "step": 37020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.797920300108592e-05,
+      "loss": 1.6345,
+      "step": 37030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.797591233670078e-05,
+      "loss": 1.5595,
+      "step": 37040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7972621672315645e-05,
+      "loss": 1.6589,
+      "step": 37050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7969331007930506e-05,
+      "loss": 1.6974,
+      "step": 37060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.796604034354536e-05,
+      "loss": 1.6656,
+      "step": 37070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.796274967916023e-05,
+      "loss": 1.6781,
+      "step": 37080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.795945901477508e-05,
+      "loss": 1.6165,
+      "step": 37090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.795616835038995e-05,
+      "loss": 1.6842,
+      "step": 37100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7952877686004805e-05,
+      "loss": 1.7877,
+      "step": 37110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.794958702161967e-05,
+      "loss": 1.8202,
+      "step": 37120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.794629635723453e-05,
+      "loss": 1.647,
+      "step": 37130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.794300569284939e-05,
+      "loss": 1.5307,
+      "step": 37140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793971502846425e-05,
+      "loss": 1.6354,
+      "step": 37150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793642436407911e-05,
+      "loss": 1.563,
+      "step": 37160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7933133699693966e-05,
+      "loss": 1.6302,
+      "step": 37170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7929843035308834e-05,
+      "loss": 1.3498,
+      "step": 37180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792655237092369e-05,
+      "loss": 1.7081,
+      "step": 37190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792326170653856e-05,
+      "loss": 1.728,
+      "step": 37200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791997104215341e-05,
+      "loss": 2.4395,
+      "step": 37210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791668037776827e-05,
+      "loss": 1.6968,
+      "step": 37220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7913389713383134e-05,
+      "loss": 1.5387,
+      "step": 37230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7910099048997995e-05,
+      "loss": 2.0568,
+      "step": 37240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7906808384612856e-05,
+      "loss": 1.8264,
+      "step": 37250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790351772022772e-05,
+      "loss": 1.558,
+      "step": 37260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790022705584257e-05,
+      "loss": 1.6242,
+      "step": 37270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789693639145744e-05,
+      "loss": 1.5365,
+      "step": 37280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7893645727072295e-05,
+      "loss": 1.5429,
+      "step": 37290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789035506268716e-05,
+      "loss": 1.4891,
+      "step": 37300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788706439830202e-05,
+      "loss": 1.7571,
+      "step": 37310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788377373391688e-05,
+      "loss": 1.8201,
+      "step": 37320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788048306953174e-05,
+      "loss": 1.5977,
+      "step": 37330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78771924051466e-05,
+      "loss": 1.7077,
+      "step": 37340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.787390174076146e-05,
+      "loss": 1.74,
+      "step": 37350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7870611076376324e-05,
+      "loss": 1.6854,
+      "step": 37360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.786732041199118e-05,
+      "loss": 1.6606,
+      "step": 37370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7864029747606046e-05,
+      "loss": 1.7825,
+      "step": 37380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.78607390832209e-05,
+      "loss": 1.4312,
+      "step": 37390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.785744841883577e-05,
+      "loss": 1.71,
+      "step": 37400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.785415775445062e-05,
+      "loss": 1.6736,
+      "step": 37410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7850867090065484e-05,
+      "loss": 1.5951,
+      "step": 37420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7847576425680346e-05,
+      "loss": 1.6296,
+      "step": 37430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.784428576129521e-05,
+      "loss": 1.57,
+      "step": 37440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.784099509691007e-05,
+      "loss": 1.5605,
+      "step": 37450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.783770443252493e-05,
+      "loss": 1.7158,
+      "step": 37460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7834413768139784e-05,
+      "loss": 1.7347,
+      "step": 37470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.783112310375465e-05,
+      "loss": 1.6899,
+      "step": 37480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7827832439369506e-05,
+      "loss": 1.5456,
+      "step": 37490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7824541774984375e-05,
+      "loss": 1.6521,
+      "step": 37500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.782125111059923e-05,
+      "loss": 1.5215,
+      "step": 37510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.781796044621409e-05,
+      "loss": 1.6725,
+      "step": 37520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.781466978182895e-05,
+      "loss": 1.5941,
+      "step": 37530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.781137911744381e-05,
+      "loss": 1.6242,
+      "step": 37540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7808088453058674e-05,
+      "loss": 1.6835,
+      "step": 37550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7804797788673535e-05,
+      "loss": 1.5784,
+      "step": 37560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7801507124288397e-05,
+      "loss": 1.8038,
+      "step": 37570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.779821645990326e-05,
+      "loss": 1.5365,
+      "step": 37580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.779492579551811e-05,
+      "loss": 1.6032,
+      "step": 37590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.779163513113298e-05,
+      "loss": 1.5901,
+      "step": 37600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7788344466747835e-05,
+      "loss": 1.5795,
+      "step": 37610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7785053802362696e-05,
+      "loss": 1.5288,
+      "step": 37620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.778176313797756e-05,
+      "loss": 1.7693,
+      "step": 37630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777847247359242e-05,
+      "loss": 1.7688,
+      "step": 37640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777518180920728e-05,
+      "loss": 1.819,
+      "step": 37650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777189114482214e-05,
+      "loss": 1.685,
+      "step": 37660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7768600480437e-05,
+      "loss": 1.5775,
+      "step": 37670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7765309816051864e-05,
+      "loss": 1.7073,
+      "step": 37680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7762019151666725e-05,
+      "loss": 1.4927,
+      "step": 37690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7758728487281586e-05,
+      "loss": 1.5711,
+      "step": 37700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775543782289645e-05,
+      "loss": 1.5223,
+      "step": 37710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77521471585113e-05,
+      "loss": 1.6195,
+      "step": 37720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774885649412617e-05,
+      "loss": 1.726,
+      "step": 37730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7745565829741025e-05,
+      "loss": 1.6722,
+      "step": 37740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7742275165355886e-05,
+      "loss": 1.6207,
+      "step": 37750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773898450097075e-05,
+      "loss": 1.6087,
+      "step": 37760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773569383658561e-05,
+      "loss": 1.5934,
+      "step": 37770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773240317220047e-05,
+      "loss": 1.6247,
+      "step": 37780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772911250781533e-05,
+      "loss": 1.5531,
+      "step": 37790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772582184343019e-05,
+      "loss": 1.7064,
+      "step": 37800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7722531179045053e-05,
+      "loss": 1.4971,
+      "step": 37810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771924051465991e-05,
+      "loss": 1.7285,
+      "step": 37820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7715949850274776e-05,
+      "loss": 1.6588,
+      "step": 37830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771265918588963e-05,
+      "loss": 1.6204,
+      "step": 37840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77093685215045e-05,
+      "loss": 1.5385,
+      "step": 37850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.770607785711935e-05,
+      "loss": 1.4848,
+      "step": 37860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7702787192734214e-05,
+      "loss": 1.5813,
+      "step": 37870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7699496528349076e-05,
+      "loss": 1.6124,
+      "step": 37880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.769620586396394e-05,
+      "loss": 1.8193,
+      "step": 37890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.76929151995788e-05,
+      "loss": 1.6863,
+      "step": 37900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.768962453519366e-05,
+      "loss": 1.638,
+      "step": 37910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7686333870808514e-05,
+      "loss": 1.5496,
+      "step": 37920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.768304320642338e-05,
+      "loss": 1.6166,
+      "step": 37930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7679752542038236e-05,
+      "loss": 1.8062,
+      "step": 37940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7676461877653104e-05,
+      "loss": 1.6842,
+      "step": 37950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.767317121326796e-05,
+      "loss": 1.6859,
+      "step": 37960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.766988054888282e-05,
+      "loss": 1.648,
+      "step": 37970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.766658988449768e-05,
+      "loss": 1.4587,
+      "step": 37980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.766329922011254e-05,
+      "loss": 2.1201,
+      "step": 37990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7660008555727404e-05,
+      "loss": 1.7195,
+      "step": 38000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7656717891342265e-05,
+      "loss": 1.5206,
+      "step": 38010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.765342722695712e-05,
+      "loss": 1.5055,
+      "step": 38020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.765013656257199e-05,
+      "loss": 1.5712,
+      "step": 38030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.764684589818684e-05,
+      "loss": 1.6669,
+      "step": 38040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.764355523380171e-05,
+      "loss": 1.7565,
+      "step": 38050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7640264569416565e-05,
+      "loss": 1.5913,
+      "step": 38060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7636973905031426e-05,
+      "loss": 1.6076,
+      "step": 38070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.763368324064629e-05,
+      "loss": 1.6106,
+      "step": 38080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.763039257626115e-05,
+      "loss": 1.6673,
+      "step": 38090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.762710191187601e-05,
+      "loss": 1.6048,
+      "step": 38100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.762381124749087e-05,
+      "loss": 1.6173,
+      "step": 38110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7620520583105726e-05,
+      "loss": 1.651,
+      "step": 38120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7617229918720594e-05,
+      "loss": 1.5594,
+      "step": 38130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.761393925433545e-05,
+      "loss": 1.4149,
+      "step": 38140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7610648589950316e-05,
+      "loss": 1.7909,
+      "step": 38150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760735792556517e-05,
+      "loss": 1.7015,
+      "step": 38160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760406726118003e-05,
+      "loss": 1.7203,
+      "step": 38170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760077659679489e-05,
+      "loss": 1.6731,
+      "step": 38180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7597485932409755e-05,
+      "loss": 1.6166,
+      "step": 38190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7594195268024616e-05,
+      "loss": 1.8979,
+      "step": 38200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759090460363948e-05,
+      "loss": 1.6456,
+      "step": 38210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758761393925433e-05,
+      "loss": 1.5537,
+      "step": 38220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75843232748692e-05,
+      "loss": 1.7054,
+      "step": 38230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7581032610484054e-05,
+      "loss": 1.6863,
+      "step": 38240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757774194609892e-05,
+      "loss": 1.6825,
+      "step": 38250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757445128171378e-05,
+      "loss": 1.7352,
+      "step": 38260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757116061732864e-05,
+      "loss": 1.5049,
+      "step": 38270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75678699529435e-05,
+      "loss": 1.6787,
+      "step": 38280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756457928855836e-05,
+      "loss": 1.577,
+      "step": 38290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756128862417322e-05,
+      "loss": 1.717,
+      "step": 38300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755799795978808e-05,
+      "loss": 1.6265,
+      "step": 38310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7554707295402944e-05,
+      "loss": 1.541,
+      "step": 38320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7551416631017806e-05,
+      "loss": 1.4948,
+      "step": 38330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754812596663267e-05,
+      "loss": 1.6986,
+      "step": 38340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754483530224753e-05,
+      "loss": 1.737,
+      "step": 38350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754154463786238e-05,
+      "loss": 1.5198,
+      "step": 38360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7538253973477244e-05,
+      "loss": 1.7455,
+      "step": 38370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7534963309092105e-05,
+      "loss": 1.5652,
+      "step": 38380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7531672644706966e-05,
+      "loss": 1.5702,
+      "step": 38390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.752838198032183e-05,
+      "loss": 1.685,
+      "step": 38400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.752509131593669e-05,
+      "loss": 1.8584,
+      "step": 38410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.752180065155155e-05,
+      "loss": 1.6746,
+      "step": 38420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.751850998716641e-05,
+      "loss": 1.8432,
+      "step": 38430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.751521932278127e-05,
+      "loss": 1.6729,
+      "step": 38440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7511928658396134e-05,
+      "loss": 1.6797,
+      "step": 38450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7508637994010995e-05,
+      "loss": 1.6716,
+      "step": 38460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.750534732962585e-05,
+      "loss": 1.5869,
+      "step": 38470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.750205666524072e-05,
+      "loss": 1.7508,
+      "step": 38480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.749876600085557e-05,
+      "loss": 1.6339,
+      "step": 38490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.749547533647044e-05,
+      "loss": 1.5933,
+      "step": 38500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7492184672085295e-05,
+      "loss": 1.7444,
+      "step": 38510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7488894007700156e-05,
+      "loss": 1.6774,
+      "step": 38520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.748560334331502e-05,
+      "loss": 1.6803,
+      "step": 38530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.748231267892988e-05,
+      "loss": 1.8053,
+      "step": 38540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.747902201454474e-05,
+      "loss": 1.9969,
+      "step": 38550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.74757313501596e-05,
+      "loss": 1.5263,
+      "step": 38560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7472440685774456e-05,
+      "loss": 1.7941,
+      "step": 38570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7469150021389324e-05,
+      "loss": 1.6445,
+      "step": 38580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.746585935700418e-05,
+      "loss": 1.6891,
+      "step": 38590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7462568692619046e-05,
+      "loss": 1.6306,
+      "step": 38600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.74592780282339e-05,
+      "loss": 1.5682,
+      "step": 38610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.745598736384876e-05,
+      "loss": 1.6981,
+      "step": 38620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.745269669946362e-05,
+      "loss": 1.6329,
+      "step": 38630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7449406035078484e-05,
+      "loss": 1.6199,
+      "step": 38640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7446115370693346e-05,
+      "loss": 1.4021,
+      "step": 38650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.744282470630821e-05,
+      "loss": 1.5932,
+      "step": 38660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743953404192306e-05,
+      "loss": 1.4508,
+      "step": 38670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743624337753793e-05,
+      "loss": 1.5223,
+      "step": 38680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7432952713152784e-05,
+      "loss": 1.6261,
+      "step": 38690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742966204876765e-05,
+      "loss": 1.7454,
+      "step": 38700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7426371384382507e-05,
+      "loss": 1.7019,
+      "step": 38710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742308071999737e-05,
+      "loss": 1.8132,
+      "step": 38720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741979005561223e-05,
+      "loss": 1.4554,
+      "step": 38730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741649939122709e-05,
+      "loss": 1.6406,
+      "step": 38740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741320872684195e-05,
+      "loss": 1.6202,
+      "step": 38750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740991806245681e-05,
+      "loss": 1.6259,
+      "step": 38760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740662739807167e-05,
+      "loss": 1.6865,
+      "step": 38770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7403336733686535e-05,
+      "loss": 1.6642,
+      "step": 38780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740004606930139e-05,
+      "loss": 1.7816,
+      "step": 38790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739675540491626e-05,
+      "loss": 1.5493,
+      "step": 38800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739346474053111e-05,
+      "loss": 1.7208,
+      "step": 38810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7390174076145974e-05,
+      "loss": 1.4806,
+      "step": 38820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7386883411760835e-05,
+      "loss": 1.5951,
+      "step": 38830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7383592747375696e-05,
+      "loss": 1.687,
+      "step": 38840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738030208299056e-05,
+      "loss": 1.5857,
+      "step": 38850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737701141860542e-05,
+      "loss": 1.9658,
+      "step": 38860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737372075422028e-05,
+      "loss": 1.5743,
+      "step": 38870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.737043008983514e-05,
+      "loss": 1.6643,
+      "step": 38880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7367139425449996e-05,
+      "loss": 1.6122,
+      "step": 38890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7363848761064864e-05,
+      "loss": 1.5692,
+      "step": 38900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.736055809667972e-05,
+      "loss": 1.5268,
+      "step": 38910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7357267432294586e-05,
+      "loss": 1.5294,
+      "step": 38920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.735397676790944e-05,
+      "loss": 1.9271,
+      "step": 38930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.73506861035243e-05,
+      "loss": 1.6933,
+      "step": 38940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7347395439139163e-05,
+      "loss": 1.5582,
+      "step": 38950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7344104774754025e-05,
+      "loss": 1.4704,
+      "step": 38960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7340814110368886e-05,
+      "loss": 1.5486,
+      "step": 38970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.733752344598375e-05,
+      "loss": 1.8114,
+      "step": 38980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.73342327815986e-05,
+      "loss": 1.5705,
+      "step": 38990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.733094211721347e-05,
+      "loss": 1.7137,
+      "step": 39000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7327651452828324e-05,
+      "loss": 1.5522,
+      "step": 39010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.732436078844319e-05,
+      "loss": 1.6651,
+      "step": 39020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.732107012405805e-05,
+      "loss": 1.5855,
+      "step": 39030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.731777945967291e-05,
+      "loss": 1.6516,
+      "step": 39040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.731448879528777e-05,
+      "loss": 1.5625,
+      "step": 39050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.731119813090263e-05,
+      "loss": 1.7203,
+      "step": 39060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.730790746651749e-05,
+      "loss": 1.6591,
+      "step": 39070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.730461680213235e-05,
+      "loss": 1.4691,
+      "step": 39080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7301326137747214e-05,
+      "loss": 1.8467,
+      "step": 39090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7298035473362076e-05,
+      "loss": 1.825,
+      "step": 39100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.729474480897694e-05,
+      "loss": 1.6236,
+      "step": 39110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72914541445918e-05,
+      "loss": 1.5683,
+      "step": 39120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.728816348020665e-05,
+      "loss": 1.6629,
+      "step": 39130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7284872815821514e-05,
+      "loss": 1.4333,
+      "step": 39140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7281582151436375e-05,
+      "loss": 1.6193,
+      "step": 39150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7278291487051236e-05,
+      "loss": 1.533,
+      "step": 39160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72750008226661e-05,
+      "loss": 1.6847,
+      "step": 39170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.727171015828096e-05,
+      "loss": 1.5606,
+      "step": 39180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726841949389582e-05,
+      "loss": 1.6914,
+      "step": 39190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726512882951068e-05,
+      "loss": 1.5222,
+      "step": 39200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726183816512554e-05,
+      "loss": 1.5969,
+      "step": 39210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7258547500740404e-05,
+      "loss": 1.6436,
+      "step": 39220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7255256836355265e-05,
+      "loss": 1.4356,
+      "step": 39230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725196617197012e-05,
+      "loss": 1.7639,
+      "step": 39240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724867550758499e-05,
+      "loss": 1.5952,
+      "step": 39250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724538484319984e-05,
+      "loss": 1.5663,
+      "step": 39260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724209417881471e-05,
+      "loss": 1.6958,
+      "step": 39270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7238803514429565e-05,
+      "loss": 1.7132,
+      "step": 39280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7235512850044426e-05,
+      "loss": 1.6035,
+      "step": 39290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723222218565929e-05,
+      "loss": 1.6969,
+      "step": 39300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722893152127415e-05,
+      "loss": 1.7502,
+      "step": 39310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722564085688901e-05,
+      "loss": 1.671,
+      "step": 39320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722235019250387e-05,
+      "loss": 1.6478,
+      "step": 39330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7219059528118726e-05,
+      "loss": 1.56,
+      "step": 39340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7215768863733594e-05,
+      "loss": 1.7796,
+      "step": 39350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721247819934845e-05,
+      "loss": 1.7668,
+      "step": 39360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7209187534963316e-05,
+      "loss": 1.6859,
+      "step": 39370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720589687057817e-05,
+      "loss": 1.5819,
+      "step": 39380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.720260620619303e-05,
+      "loss": 1.7401,
+      "step": 39390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.719931554180789e-05,
+      "loss": 1.6577,
+      "step": 39400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7196024877422755e-05,
+      "loss": 1.6519,
+      "step": 39410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7192734213037616e-05,
+      "loss": 1.6959,
+      "step": 39420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.718944354865248e-05,
+      "loss": 1.5707,
+      "step": 39430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.718615288426733e-05,
+      "loss": 1.6066,
+      "step": 39440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.71828622198822e-05,
+      "loss": 1.6766,
+      "step": 39450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7179571555497054e-05,
+      "loss": 1.5986,
+      "step": 39460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.717628089111192e-05,
+      "loss": 1.6644,
+      "step": 39470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.717299022672678e-05,
+      "loss": 1.5629,
+      "step": 39480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.716969956234164e-05,
+      "loss": 1.571,
+      "step": 39490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.71664088979565e-05,
+      "loss": 1.9359,
+      "step": 39500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.716311823357136e-05,
+      "loss": 1.6647,
+      "step": 39510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.715982756918622e-05,
+      "loss": 1.6304,
+      "step": 39520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.715653690480108e-05,
+      "loss": 1.6362,
+      "step": 39530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.715324624041594e-05,
+      "loss": 1.5724,
+      "step": 39540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7149955576030806e-05,
+      "loss": 1.6606,
+      "step": 39550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.714666491164566e-05,
+      "loss": 1.5713,
+      "step": 39560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.714337424726053e-05,
+      "loss": 1.5702,
+      "step": 39570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.714008358287538e-05,
+      "loss": 1.5065,
+      "step": 39580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7136792918490244e-05,
+      "loss": 1.5687,
+      "step": 39590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7133502254105105e-05,
+      "loss": 1.6815,
+      "step": 39600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7130211589719966e-05,
+      "loss": 1.5636,
+      "step": 39610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.712692092533483e-05,
+      "loss": 1.5266,
+      "step": 39620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.712363026094969e-05,
+      "loss": 1.7449,
+      "step": 39630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7120339596564543e-05,
+      "loss": 1.7103,
+      "step": 39640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.711704893217941e-05,
+      "loss": 1.6804,
+      "step": 39650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7113758267794266e-05,
+      "loss": 1.569,
+      "step": 39660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7110467603409134e-05,
+      "loss": 1.5532,
+      "step": 39670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710717693902399e-05,
+      "loss": 1.7413,
+      "step": 39680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710388627463885e-05,
+      "loss": 1.6384,
+      "step": 39690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710059561025371e-05,
+      "loss": 1.6344,
+      "step": 39700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709730494586857e-05,
+      "loss": 1.6135,
+      "step": 39710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7094014281483434e-05,
+      "loss": 1.5136,
+      "step": 39720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7090723617098295e-05,
+      "loss": 1.5702,
+      "step": 39730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708743295271315e-05,
+      "loss": 1.5446,
+      "step": 39740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708414228832802e-05,
+      "loss": 1.6594,
+      "step": 39750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708085162394287e-05,
+      "loss": 1.6247,
+      "step": 39760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707756095955774e-05,
+      "loss": 1.6266,
+      "step": 39770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7074270295172594e-05,
+      "loss": 1.4823,
+      "step": 39780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7070979630787456e-05,
+      "loss": 1.6851,
+      "step": 39790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706768896640232e-05,
+      "loss": 1.6524,
+      "step": 39800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706439830201718e-05,
+      "loss": 1.618,
+      "step": 39810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706110763763204e-05,
+      "loss": 1.5617,
+      "step": 39820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70578169732469e-05,
+      "loss": 1.755,
+      "step": 39830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705452630886176e-05,
+      "loss": 1.9465,
+      "step": 39840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705123564447662e-05,
+      "loss": 1.556,
+      "step": 39850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7047944980091485e-05,
+      "loss": 1.6979,
+      "step": 39860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7044654315706346e-05,
+      "loss": 1.6472,
+      "step": 39870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70413636513212e-05,
+      "loss": 1.674,
+      "step": 39880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.703807298693606e-05,
+      "loss": 1.5664,
+      "step": 39890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.703478232255092e-05,
+      "loss": 1.5428,
+      "step": 39900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7031491658165784e-05,
+      "loss": 1.6228,
+      "step": 39910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7028200993780645e-05,
+      "loss": 1.758,
+      "step": 39920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.702491032939551e-05,
+      "loss": 1.4676,
+      "step": 39930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.702161966501037e-05,
+      "loss": 1.6525,
+      "step": 39940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.701832900062523e-05,
+      "loss": 1.6218,
+      "step": 39950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.701503833624009e-05,
+      "loss": 1.6752,
+      "step": 39960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.701174767185495e-05,
+      "loss": 1.6407,
+      "step": 39970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.700845700746981e-05,
+      "loss": 1.6186,
+      "step": 39980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.700516634308467e-05,
+      "loss": 1.5572,
+      "step": 39990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7001875678699536e-05,
+      "loss": 1.5124,
+      "step": 40000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.699858501431439e-05,
+      "loss": 1.7283,
+      "step": 40010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.699529434992926e-05,
+      "loss": 1.6166,
+      "step": 40020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.699200368554411e-05,
+      "loss": 1.7291,
+      "step": 40030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6988713021158974e-05,
+      "loss": 1.6348,
+      "step": 40040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6985422356773835e-05,
+      "loss": 1.6735,
+      "step": 40050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6982131692388696e-05,
+      "loss": 1.4499,
+      "step": 40060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.697884102800356e-05,
+      "loss": 1.6656,
+      "step": 40070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.697555036361842e-05,
+      "loss": 1.6189,
+      "step": 40080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.697225969923327e-05,
+      "loss": 1.6998,
+      "step": 40090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.696896903484814e-05,
+      "loss": 1.9471,
+      "step": 40100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6965678370462996e-05,
+      "loss": 1.6995,
+      "step": 40110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6962387706077864e-05,
+      "loss": 1.49,
+      "step": 40120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.695909704169272e-05,
+      "loss": 1.4639,
+      "step": 40130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.695580637730758e-05,
+      "loss": 1.6887,
+      "step": 40140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.695251571292244e-05,
+      "loss": 1.4358,
+      "step": 40150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69492250485373e-05,
+      "loss": 1.8993,
+      "step": 40160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6945934384152164e-05,
+      "loss": 1.5517,
+      "step": 40170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6942643719767025e-05,
+      "loss": 1.6816,
+      "step": 40180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693935305538188e-05,
+      "loss": 1.6986,
+      "step": 40190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693606239099675e-05,
+      "loss": 1.5422,
+      "step": 40200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69327717266116e-05,
+      "loss": 1.4746,
+      "step": 40210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692948106222647e-05,
+      "loss": 1.7202,
+      "step": 40220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6926190397841324e-05,
+      "loss": 1.5452,
+      "step": 40230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6922899733456186e-05,
+      "loss": 1.7596,
+      "step": 40240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691960906907105e-05,
+      "loss": 1.5742,
+      "step": 40250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691631840468591e-05,
+      "loss": 1.8192,
+      "step": 40260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691302774030077e-05,
+      "loss": 1.7947,
+      "step": 40270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690973707591563e-05,
+      "loss": 1.7047,
+      "step": 40280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6906446411530485e-05,
+      "loss": 1.717,
+      "step": 40290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690315574714535e-05,
+      "loss": 1.9599,
+      "step": 40300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689986508276021e-05,
+      "loss": 1.591,
+      "step": 40310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6896574418375076e-05,
+      "loss": 1.6024,
+      "step": 40320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689328375398993e-05,
+      "loss": 1.526,
+      "step": 40330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688999308960479e-05,
+      "loss": 1.6534,
+      "step": 40340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688670242521965e-05,
+      "loss": 1.9069,
+      "step": 40350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6883411760834514e-05,
+      "loss": 1.6271,
+      "step": 40360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6880121096449375e-05,
+      "loss": 1.4889,
+      "step": 40370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6876830432064237e-05,
+      "loss": 1.6598,
+      "step": 40380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687353976767909e-05,
+      "loss": 1.6492,
+      "step": 40390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.687024910329396e-05,
+      "loss": 1.7238,
+      "step": 40400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6866958438908814e-05,
+      "loss": 1.9865,
+      "step": 40410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.686366777452368e-05,
+      "loss": 1.86,
+      "step": 40420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6860377110138536e-05,
+      "loss": 1.7254,
+      "step": 40430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.68570864457534e-05,
+      "loss": 1.5778,
+      "step": 40440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.685379578136826e-05,
+      "loss": 1.7292,
+      "step": 40450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.685050511698312e-05,
+      "loss": 1.5548,
+      "step": 40460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.684721445259798e-05,
+      "loss": 1.6435,
+      "step": 40470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.684392378821284e-05,
+      "loss": 1.5953,
+      "step": 40480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.68406331238277e-05,
+      "loss": 1.64,
+      "step": 40490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6837342459442565e-05,
+      "loss": 1.6846,
+      "step": 40500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.683405179505742e-05,
+      "loss": 1.7359,
+      "step": 40510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.683076113067229e-05,
+      "loss": 1.5686,
+      "step": 40520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.682747046628714e-05,
+      "loss": 1.6447,
+      "step": 40530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6824179801902e-05,
+      "loss": 1.6513,
+      "step": 40540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6820889137516865e-05,
+      "loss": 1.6128,
+      "step": 40550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6817598473131726e-05,
+      "loss": 1.6929,
+      "step": 40560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.681430780874659e-05,
+      "loss": 1.5366,
+      "step": 40570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.681101714436145e-05,
+      "loss": 1.7975,
+      "step": 40580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.680772647997631e-05,
+      "loss": 1.4555,
+      "step": 40590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.680443581559117e-05,
+      "loss": 1.6621,
+      "step": 40600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.680114515120603e-05,
+      "loss": 1.5364,
+      "step": 40610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6797854486820893e-05,
+      "loss": 1.6689,
+      "step": 40620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6794563822435755e-05,
+      "loss": 1.662,
+      "step": 40630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.679127315805061e-05,
+      "loss": 1.5593,
+      "step": 40640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.678798249366547e-05,
+      "loss": 1.7086,
+      "step": 40650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.678469182928033e-05,
+      "loss": 1.613,
+      "step": 40660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.678140116489519e-05,
+      "loss": 1.5861,
+      "step": 40670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6778110500510054e-05,
+      "loss": 1.7109,
+      "step": 40680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6774819836124916e-05,
+      "loss": 1.6734,
+      "step": 40690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677152917173978e-05,
+      "loss": 1.6281,
+      "step": 40700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676823850735464e-05,
+      "loss": 1.632,
+      "step": 40710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67649478429695e-05,
+      "loss": 1.7849,
+      "step": 40720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676165717858436e-05,
+      "loss": 1.6855,
+      "step": 40730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6758366514199215e-05,
+      "loss": 1.7304,
+      "step": 40740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675507584981408e-05,
+      "loss": 1.7332,
+      "step": 40750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675178518542894e-05,
+      "loss": 1.8902,
+      "step": 40760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6748494521043806e-05,
+      "loss": 1.5794,
+      "step": 40770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674520385665866e-05,
+      "loss": 1.6855,
+      "step": 40780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674191319227352e-05,
+      "loss": 1.6458,
+      "step": 40790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673862252788838e-05,
+      "loss": 1.6083,
+      "step": 40800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6735331863503244e-05,
+      "loss": 1.5531,
+      "step": 40810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6732041199118105e-05,
+      "loss": 1.7372,
+      "step": 40820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6728750534732967e-05,
+      "loss": 1.463,
+      "step": 40830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672545987034782e-05,
+      "loss": 1.6672,
+      "step": 40840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672216920596269e-05,
+      "loss": 1.6197,
+      "step": 40850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6718878541577544e-05,
+      "loss": 1.6664,
+      "step": 40860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671558787719241e-05,
+      "loss": 1.4011,
+      "step": 40870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6712297212807266e-05,
+      "loss": 1.7207,
+      "step": 40880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670900654842213e-05,
+      "loss": 1.5426,
+      "step": 40890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670571588403699e-05,
+      "loss": 1.6899,
+      "step": 40900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.670242521965185e-05,
+      "loss": 1.698,
+      "step": 40910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669913455526671e-05,
+      "loss": 1.5222,
+      "step": 40920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669584389088157e-05,
+      "loss": 1.5757,
+      "step": 40930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669255322649643e-05,
+      "loss": 1.626,
+      "step": 40940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6689262562111295e-05,
+      "loss": 1.8142,
+      "step": 40950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.668597189772615e-05,
+      "loss": 1.8289,
+      "step": 40960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.668268123334102e-05,
+      "loss": 1.6274,
+      "step": 40970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.667939056895587e-05,
+      "loss": 1.5588,
+      "step": 40980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.667609990457073e-05,
+      "loss": 1.555,
+      "step": 40990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6672809240185595e-05,
+      "loss": 1.5611,
+      "step": 41000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6669518575800456e-05,
+      "loss": 1.6834,
+      "step": 41010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.666622791141532e-05,
+      "loss": 1.6309,
+      "step": 41020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.666293724703018e-05,
+      "loss": 1.6558,
+      "step": 41030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.665964658264503e-05,
+      "loss": 1.682,
+      "step": 41040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66563559182599e-05,
+      "loss": 1.5903,
+      "step": 41050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6653065253874755e-05,
+      "loss": 1.7744,
+      "step": 41060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664977458948962e-05,
+      "loss": 1.6687,
+      "step": 41070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664648392510448e-05,
+      "loss": 1.6982,
+      "step": 41080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664319326071934e-05,
+      "loss": 1.6582,
+      "step": 41090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66399025963342e-05,
+      "loss": 1.6563,
+      "step": 41100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663661193194906e-05,
+      "loss": 1.4649,
+      "step": 41110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663332126756392e-05,
+      "loss": 1.725,
+      "step": 41120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6630030603178784e-05,
+      "loss": 1.6874,
+      "step": 41130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.662673993879364e-05,
+      "loss": 1.6533,
+      "step": 41140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.662344927440851e-05,
+      "loss": 1.5359,
+      "step": 41150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.662015861002336e-05,
+      "loss": 1.5548,
+      "step": 41160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661686794563823e-05,
+      "loss": 1.6365,
+      "step": 41170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6613577281253084e-05,
+      "loss": 1.6496,
+      "step": 41180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6610286616867945e-05,
+      "loss": 1.6737,
+      "step": 41190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6606995952482806e-05,
+      "loss": 1.4987,
+      "step": 41200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660370528809767e-05,
+      "loss": 1.5288,
+      "step": 41210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660041462371253e-05,
+      "loss": 1.8197,
+      "step": 41220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659712395932739e-05,
+      "loss": 1.8458,
+      "step": 41230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659383329494225e-05,
+      "loss": 1.474,
+      "step": 41240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659054263055711e-05,
+      "loss": 1.7122,
+      "step": 41250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658725196617197e-05,
+      "loss": 1.4866,
+      "step": 41260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6583961301786835e-05,
+      "loss": 1.719,
+      "step": 41270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658067063740169e-05,
+      "loss": 1.717,
+      "step": 41280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657737997301655e-05,
+      "loss": 1.5565,
+      "step": 41290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657408930863141e-05,
+      "loss": 1.5867,
+      "step": 41300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6570798644246273e-05,
+      "loss": 1.6331,
+      "step": 41310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6567507979861135e-05,
+      "loss": 1.5244,
+      "step": 41320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6564217315475996e-05,
+      "loss": 1.6039,
+      "step": 41330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656092665109086e-05,
+      "loss": 1.5308,
+      "step": 41340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655763598670572e-05,
+      "loss": 1.7839,
+      "step": 41350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655434532232058e-05,
+      "loss": 1.591,
+      "step": 41360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655105465793544e-05,
+      "loss": 1.525,
+      "step": 41370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65477639935503e-05,
+      "loss": 1.6208,
+      "step": 41380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654447332916516e-05,
+      "loss": 1.5841,
+      "step": 41390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6541182664780025e-05,
+      "loss": 1.5999,
+      "step": 41400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653789200039488e-05,
+      "loss": 1.6659,
+      "step": 41410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.653460133600974e-05,
+      "loss": 1.8623,
+      "step": 41420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.65313106716246e-05,
+      "loss": 1.7329,
+      "step": 41430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.652802000723946e-05,
+      "loss": 1.5873,
+      "step": 41440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6524729342854324e-05,
+      "loss": 1.464,
+      "step": 41450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6521438678469186e-05,
+      "loss": 1.623,
+      "step": 41460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.651814801408405e-05,
+      "loss": 1.7254,
+      "step": 41470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.651485734969891e-05,
+      "loss": 1.5181,
+      "step": 41480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.651156668531376e-05,
+      "loss": 1.5688,
+      "step": 41490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.650827602092863e-05,
+      "loss": 1.4915,
+      "step": 41500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6504985356543485e-05,
+      "loss": 1.6984,
+      "step": 41510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.650169469215835e-05,
+      "loss": 1.6045,
+      "step": 41520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.649840402777321e-05,
+      "loss": 1.605,
+      "step": 41530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.649511336338807e-05,
+      "loss": 1.7093,
+      "step": 41540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.649182269900293e-05,
+      "loss": 1.6407,
+      "step": 41550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648853203461779e-05,
+      "loss": 1.7062,
+      "step": 41560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648524137023265e-05,
+      "loss": 1.5042,
+      "step": 41570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6481950705847514e-05,
+      "loss": 1.6143,
+      "step": 41580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647866004146237e-05,
+      "loss": 1.6816,
+      "step": 41590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647536937707724e-05,
+      "loss": 1.5571,
+      "step": 41600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647207871269209e-05,
+      "loss": 1.5174,
+      "step": 41610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.646878804830696e-05,
+      "loss": 1.5432,
+      "step": 41620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6465497383921814e-05,
+      "loss": 1.5938,
+      "step": 41630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6462206719536675e-05,
+      "loss": 1.4966,
+      "step": 41640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6458916055151536e-05,
+      "loss": 1.6038,
+      "step": 41650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64556253907664e-05,
+      "loss": 1.6275,
+      "step": 41660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.645233472638126e-05,
+      "loss": 1.5942,
+      "step": 41670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644904406199612e-05,
+      "loss": 1.4124,
+      "step": 41680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644575339761098e-05,
+      "loss": 1.6576,
+      "step": 41690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644246273322584e-05,
+      "loss": 1.6188,
+      "step": 41700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64391720688407e-05,
+      "loss": 1.6634,
+      "step": 41710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6435881404455565e-05,
+      "loss": 1.6068,
+      "step": 41720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643259074007042e-05,
+      "loss": 1.6987,
+      "step": 41730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642930007568529e-05,
+      "loss": 1.7788,
+      "step": 41740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642600941130014e-05,
+      "loss": 1.6353,
+      "step": 41750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6422718746915003e-05,
+      "loss": 1.587,
+      "step": 41760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6419428082529865e-05,
+      "loss": 1.5511,
+      "step": 41770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6416137418144726e-05,
+      "loss": 1.739,
+      "step": 41780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641284675375959e-05,
+      "loss": 1.4688,
+      "step": 41790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640955608937445e-05,
+      "loss": 1.6315,
+      "step": 41800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64062654249893e-05,
+      "loss": 1.6355,
+      "step": 41810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640297476060417e-05,
+      "loss": 1.4577,
+      "step": 41820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6399684096219025e-05,
+      "loss": 1.6887,
+      "step": 41830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6396393431833894e-05,
+      "loss": 1.5423,
+      "step": 41840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639310276744875e-05,
+      "loss": 1.6581,
+      "step": 41850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638981210306361e-05,
+      "loss": 1.7337,
+      "step": 41860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638652143867847e-05,
+      "loss": 1.505,
+      "step": 41870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638323077429333e-05,
+      "loss": 1.5781,
+      "step": 41880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637994010990819e-05,
+      "loss": 1.7698,
+      "step": 41890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6376649445523054e-05,
+      "loss": 1.6415,
+      "step": 41900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637335878113791e-05,
+      "loss": 1.7163,
+      "step": 41910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637006811675278e-05,
+      "loss": 1.595,
+      "step": 41920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.636677745236763e-05,
+      "loss": 1.6812,
+      "step": 41930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.63634867879825e-05,
+      "loss": 1.6871,
+      "step": 41940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6360196123597354e-05,
+      "loss": 1.6926,
+      "step": 41950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6356905459212215e-05,
+      "loss": 1.873,
+      "step": 41960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6353614794827076e-05,
+      "loss": 1.6989,
+      "step": 41970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.635032413044194e-05,
+      "loss": 1.67,
+      "step": 41980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.63470334660568e-05,
+      "loss": 1.6943,
+      "step": 41990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.634374280167166e-05,
+      "loss": 1.5546,
+      "step": 42000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6340452137286515e-05,
+      "loss": 1.4191,
+      "step": 42010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.633716147290138e-05,
+      "loss": 1.6848,
+      "step": 42020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.633387080851624e-05,
+      "loss": 1.623,
+      "step": 42030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6330580144131105e-05,
+      "loss": 1.6208,
+      "step": 42040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.632728947974596e-05,
+      "loss": 1.5798,
+      "step": 42050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.632399881536082e-05,
+      "loss": 1.5582,
+      "step": 42060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.632070815097568e-05,
+      "loss": 1.6583,
+      "step": 42070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6317417486590544e-05,
+      "loss": 1.5638,
+      "step": 42080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6314126822205405e-05,
+      "loss": 1.6574,
+      "step": 42090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6310836157820266e-05,
+      "loss": 1.6522,
+      "step": 42100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630754549343513e-05,
+      "loss": 1.4971,
+      "step": 42110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630425482904999e-05,
+      "loss": 1.6529,
+      "step": 42120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630096416466485e-05,
+      "loss": 1.6012,
+      "step": 42130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629767350027971e-05,
+      "loss": 1.5106,
+      "step": 42140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629438283589457e-05,
+      "loss": 1.5981,
+      "step": 42150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629109217150943e-05,
+      "loss": 1.6182,
+      "step": 42160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.628780150712429e-05,
+      "loss": 1.7028,
+      "step": 42170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.628451084273915e-05,
+      "loss": 1.6015,
+      "step": 42180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.628122017835401e-05,
+      "loss": 1.6127,
+      "step": 42190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627792951396887e-05,
+      "loss": 1.5525,
+      "step": 42200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627463884958373e-05,
+      "loss": 1.4499,
+      "step": 42210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6271348185198595e-05,
+      "loss": 1.5979,
+      "step": 42220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6268057520813456e-05,
+      "loss": 1.5959,
+      "step": 42230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626476685642832e-05,
+      "loss": 1.6113,
+      "step": 42240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626147619204318e-05,
+      "loss": 1.5588,
+      "step": 42250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625818552765803e-05,
+      "loss": 1.5503,
+      "step": 42260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62548948632729e-05,
+      "loss": 1.6232,
+      "step": 42270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6251604198887755e-05,
+      "loss": 1.6538,
+      "step": 42280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6248313534502623e-05,
+      "loss": 1.6369,
+      "step": 42290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624502287011748e-05,
+      "loss": 1.6064,
+      "step": 42300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624173220573234e-05,
+      "loss": 1.4714,
+      "step": 42310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62384415413472e-05,
+      "loss": 1.6,
+      "step": 42320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623515087696206e-05,
+      "loss": 1.6517,
+      "step": 42330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623186021257692e-05,
+      "loss": 1.7397,
+      "step": 42340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6228569548191784e-05,
+      "loss": 1.5965,
+      "step": 42350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622527888380664e-05,
+      "loss": 1.6426,
+      "step": 42360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622198821942151e-05,
+      "loss": 1.761,
+      "step": 42370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621869755503636e-05,
+      "loss": 1.6881,
+      "step": 42380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621540689065123e-05,
+      "loss": 1.6369,
+      "step": 42390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6212116226266084e-05,
+      "loss": 1.5579,
+      "step": 42400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6208825561880945e-05,
+      "loss": 1.6694,
+      "step": 42410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6205534897495806e-05,
+      "loss": 1.6649,
+      "step": 42420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620224423311067e-05,
+      "loss": 1.6795,
+      "step": 42430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.619895356872553e-05,
+      "loss": 1.7108,
+      "step": 42440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.619566290434039e-05,
+      "loss": 1.6751,
+      "step": 42450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6192372239955245e-05,
+      "loss": 1.7363,
+      "step": 42460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.618908157557011e-05,
+      "loss": 1.647,
+      "step": 42470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.618579091118497e-05,
+      "loss": 1.7544,
+      "step": 42480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6182500246799835e-05,
+      "loss": 1.5576,
+      "step": 42490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.617920958241469e-05,
+      "loss": 1.6968,
+      "step": 42500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.617591891802955e-05,
+      "loss": 1.5602,
+      "step": 42510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.617262825364441e-05,
+      "loss": 1.6652,
+      "step": 42520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6169337589259274e-05,
+      "loss": 1.5926,
+      "step": 42530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6166046924874135e-05,
+      "loss": 1.6195,
+      "step": 42540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6162756260488996e-05,
+      "loss": 1.5773,
+      "step": 42550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615946559610385e-05,
+      "loss": 1.4256,
+      "step": 42560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615617493171872e-05,
+      "loss": 1.8381,
+      "step": 42570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615288426733357e-05,
+      "loss": 1.643,
+      "step": 42580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.614959360294844e-05,
+      "loss": 1.6552,
+      "step": 42590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6146302938563296e-05,
+      "loss": 1.682,
+      "step": 42600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.614301227417816e-05,
+      "loss": 1.4974,
+      "step": 42610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.613972160979302e-05,
+      "loss": 1.6332,
+      "step": 42620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.613643094540788e-05,
+      "loss": 1.7331,
+      "step": 42630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.613314028102274e-05,
+      "loss": 1.7193,
+      "step": 42640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.61298496166376e-05,
+      "loss": 1.6777,
+      "step": 42650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6126558952252456e-05,
+      "loss": 1.5569,
+      "step": 42660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6123268287867325e-05,
+      "loss": 1.537,
+      "step": 42670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.611997762348218e-05,
+      "loss": 1.5757,
+      "step": 42680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.611668695909705e-05,
+      "loss": 1.5912,
+      "step": 42690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.61133962947119e-05,
+      "loss": 1.5434,
+      "step": 42700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.611010563032676e-05,
+      "loss": 1.7615,
+      "step": 42710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6106814965941624e-05,
+      "loss": 1.7133,
+      "step": 42720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6103524301556485e-05,
+      "loss": 1.7114,
+      "step": 42730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610023363717135e-05,
+      "loss": 1.7897,
+      "step": 42740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609694297278621e-05,
+      "loss": 1.5345,
+      "step": 42750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609365230840107e-05,
+      "loss": 1.5964,
+      "step": 42760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609036164401593e-05,
+      "loss": 1.5737,
+      "step": 42770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6087070979630785e-05,
+      "loss": 1.4998,
+      "step": 42780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608378031524565e-05,
+      "loss": 1.4774,
+      "step": 42790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608048965086051e-05,
+      "loss": 1.6671,
+      "step": 42800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607719898647537e-05,
+      "loss": 1.6337,
+      "step": 42810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607390832209023e-05,
+      "loss": 1.7381,
+      "step": 42820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607061765770509e-05,
+      "loss": 1.6821,
+      "step": 42830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606732699331995e-05,
+      "loss": 1.777,
+      "step": 42840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6064036328934814e-05,
+      "loss": 1.6163,
+      "step": 42850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6060745664549675e-05,
+      "loss": 1.5541,
+      "step": 42860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6057455000164536e-05,
+      "loss": 1.6849,
+      "step": 42870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60541643357794e-05,
+      "loss": 1.5704,
+      "step": 42880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605087367139426e-05,
+      "loss": 1.6752,
+      "step": 42890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604758300700912e-05,
+      "loss": 1.4911,
+      "step": 42900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6044292342623975e-05,
+      "loss": 1.5011,
+      "step": 42910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604100167823884e-05,
+      "loss": 1.4972,
+      "step": 42920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60377110138537e-05,
+      "loss": 1.6419,
+      "step": 42930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.603442034946856e-05,
+      "loss": 1.5614,
+      "step": 42940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.603112968508342e-05,
+      "loss": 1.5894,
+      "step": 42950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.602783902069828e-05,
+      "loss": 1.5169,
+      "step": 42960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.602454835631314e-05,
+      "loss": 1.7547,
+      "step": 42970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6021257691928003e-05,
+      "loss": 1.718,
+      "step": 42980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6017967027542865e-05,
+      "loss": 1.5377,
+      "step": 42990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6014676363157726e-05,
+      "loss": 1.6597,
+      "step": 43000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.601138569877258e-05,
+      "loss": 1.6078,
+      "step": 43010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.600809503438745e-05,
+      "loss": 1.6813,
+      "step": 43020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.60048043700023e-05,
+      "loss": 1.6329,
+      "step": 43030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.600151370561717e-05,
+      "loss": 1.5804,
+      "step": 43040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5998223041232026e-05,
+      "loss": 1.4294,
+      "step": 43050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.599493237684689e-05,
+      "loss": 1.5248,
+      "step": 43060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.599164171246175e-05,
+      "loss": 1.6339,
+      "step": 43070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598835104807661e-05,
+      "loss": 1.5565,
+      "step": 43080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598506038369147e-05,
+      "loss": 1.5377,
+      "step": 43090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598176971930633e-05,
+      "loss": 1.6498,
+      "step": 43100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5978479054921186e-05,
+      "loss": 1.6441,
+      "step": 43110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5975188390536054e-05,
+      "loss": 1.7007,
+      "step": 43120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.597189772615091e-05,
+      "loss": 1.703,
+      "step": 43130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.596860706176578e-05,
+      "loss": 1.7145,
+      "step": 43140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.596531639738063e-05,
+      "loss": 1.5373,
+      "step": 43150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.596202573299549e-05,
+      "loss": 1.612,
+      "step": 43160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5958735068610354e-05,
+      "loss": 1.513,
+      "step": 43170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5955444404225215e-05,
+      "loss": 1.6548,
+      "step": 43180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5952153739840077e-05,
+      "loss": 1.7558,
+      "step": 43190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.594886307545494e-05,
+      "loss": 1.49,
+      "step": 43200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.594557241106979e-05,
+      "loss": 1.5692,
+      "step": 43210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.594228174668466e-05,
+      "loss": 1.5207,
+      "step": 43220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5938991082299515e-05,
+      "loss": 1.6538,
+      "step": 43230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593570041791438e-05,
+      "loss": 1.5653,
+      "step": 43240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593240975352924e-05,
+      "loss": 1.5632,
+      "step": 43250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59291190891441e-05,
+      "loss": 1.6411,
+      "step": 43260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592582842475896e-05,
+      "loss": 1.5438,
+      "step": 43270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592253776037382e-05,
+      "loss": 1.6386,
+      "step": 43280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591924709598868e-05,
+      "loss": 1.6915,
+      "step": 43290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5915956431603544e-05,
+      "loss": 1.7085,
+      "step": 43300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59126657672184e-05,
+      "loss": 1.5467,
+      "step": 43310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5909375102833266e-05,
+      "loss": 1.6221,
+      "step": 43320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590608443844812e-05,
+      "loss": 1.514,
+      "step": 43330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590279377406299e-05,
+      "loss": 1.5182,
+      "step": 43340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589950310967784e-05,
+      "loss": 1.607,
+      "step": 43350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5896212445292705e-05,
+      "loss": 1.5346,
+      "step": 43360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5892921780907566e-05,
+      "loss": 1.5857,
+      "step": 43370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588963111652243e-05,
+      "loss": 1.6008,
+      "step": 43380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588634045213729e-05,
+      "loss": 1.79,
+      "step": 43390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588304978775215e-05,
+      "loss": 1.4808,
+      "step": 43400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5879759123367004e-05,
+      "loss": 1.5558,
+      "step": 43410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587646845898187e-05,
+      "loss": 1.5795,
+      "step": 43420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587317779459673e-05,
+      "loss": 1.7134,
+      "step": 43430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5869887130211595e-05,
+      "loss": 1.545,
+      "step": 43440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.586659646582645e-05,
+      "loss": 1.5008,
+      "step": 43450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.586330580144131e-05,
+      "loss": 1.7241,
+      "step": 43460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.586001513705617e-05,
+      "loss": 1.7632,
+      "step": 43470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.585672447267103e-05,
+      "loss": 1.5465,
+      "step": 43480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5853433808285894e-05,
+      "loss": 1.6568,
+      "step": 43490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5850143143900756e-05,
+      "loss": 1.6211,
+      "step": 43500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.584685247951562e-05,
+      "loss": 1.635,
+      "step": 43510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.584356181513048e-05,
+      "loss": 1.8488,
+      "step": 43520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.584027115074533e-05,
+      "loss": 1.7635,
+      "step": 43530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.58369804863602e-05,
+      "loss": 1.5284,
+      "step": 43540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5833689821975055e-05,
+      "loss": 1.6074,
+      "step": 43550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5830399157589916e-05,
+      "loss": 1.6981,
+      "step": 43560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.582710849320478e-05,
+      "loss": 1.6964,
+      "step": 43570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.582381782881964e-05,
+      "loss": 1.5411,
+      "step": 43580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.58205271644345e-05,
+      "loss": 1.9173,
+      "step": 43590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.581723650004936e-05,
+      "loss": 1.485,
+      "step": 43600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.581394583566422e-05,
+      "loss": 1.7415,
+      "step": 43610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5810655171279084e-05,
+      "loss": 1.5826,
+      "step": 43620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5807364506893945e-05,
+      "loss": 1.642,
+      "step": 43630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5804073842508806e-05,
+      "loss": 1.4035,
+      "step": 43640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.580078317812367e-05,
+      "loss": 1.7475,
+      "step": 43650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.579749251373852e-05,
+      "loss": 1.6519,
+      "step": 43660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.579420184935339e-05,
+      "loss": 1.5818,
+      "step": 43670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5790911184968245e-05,
+      "loss": 1.6986,
+      "step": 43680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5787620520583106e-05,
+      "loss": 1.733,
+      "step": 43690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.578432985619797e-05,
+      "loss": 1.7114,
+      "step": 43700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.578103919181283e-05,
+      "loss": 1.4511,
+      "step": 43710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577774852742769e-05,
+      "loss": 1.4398,
+      "step": 43720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577445786304255e-05,
+      "loss": 1.6858,
+      "step": 43730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577116719865741e-05,
+      "loss": 1.6046,
+      "step": 43740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5767876534272274e-05,
+      "loss": 1.6154,
+      "step": 43750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576458586988713e-05,
+      "loss": 1.6498,
+      "step": 43760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5761295205501996e-05,
+      "loss": 1.6483,
+      "step": 43770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575800454111685e-05,
+      "loss": 1.6358,
+      "step": 43780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575471387673172e-05,
+      "loss": 1.6242,
+      "step": 43790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575142321234657e-05,
+      "loss": 1.6839,
+      "step": 43800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5748132547961434e-05,
+      "loss": 1.725,
+      "step": 43810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5744841883576296e-05,
+      "loss": 1.5332,
+      "step": 43820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574155121919116e-05,
+      "loss": 1.6972,
+      "step": 43830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573826055480602e-05,
+      "loss": 1.5557,
+      "step": 43840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573496989042088e-05,
+      "loss": 1.7295,
+      "step": 43850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5731679226035734e-05,
+      "loss": 1.5245,
+      "step": 43860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57283885616506e-05,
+      "loss": 1.4939,
+      "step": 43870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5725097897265457e-05,
+      "loss": 1.7071,
+      "step": 43880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5721807232880325e-05,
+      "loss": 1.5785,
+      "step": 43890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571851656849518e-05,
+      "loss": 1.5787,
+      "step": 43900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571522590411004e-05,
+      "loss": 1.6027,
+      "step": 43910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57119352397249e-05,
+      "loss": 1.5205,
+      "step": 43920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570864457533976e-05,
+      "loss": 1.643,
+      "step": 43930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5705353910954624e-05,
+      "loss": 1.6268,
+      "step": 43940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5702063246569485e-05,
+      "loss": 1.552,
+      "step": 43950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.569877258218434e-05,
+      "loss": 1.4669,
+      "step": 43960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.569548191779921e-05,
+      "loss": 1.6869,
+      "step": 43970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.569219125341406e-05,
+      "loss": 1.5872,
+      "step": 43980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.568890058902893e-05,
+      "loss": 1.6868,
+      "step": 43990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5685609924643785e-05,
+      "loss": 1.5297,
+      "step": 44000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5682319260258646e-05,
+      "loss": 1.5147,
+      "step": 44010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.567902859587351e-05,
+      "loss": 1.7357,
+      "step": 44020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.567573793148837e-05,
+      "loss": 1.6595,
+      "step": 44030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.567244726710323e-05,
+      "loss": 1.5585,
+      "step": 44040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.566915660271809e-05,
+      "loss": 1.6639,
+      "step": 44050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5665865938332946e-05,
+      "loss": 1.6112,
+      "step": 44060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5662575273947814e-05,
+      "loss": 1.6958,
+      "step": 44070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.565928460956267e-05,
+      "loss": 1.819,
+      "step": 44080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5655993945177536e-05,
+      "loss": 1.5089,
+      "step": 44090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.565270328079239e-05,
+      "loss": 1.656,
+      "step": 44100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.564941261640725e-05,
+      "loss": 1.5371,
+      "step": 44110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5646121952022113e-05,
+      "loss": 1.6958,
+      "step": 44120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5642831287636975e-05,
+      "loss": 1.5087,
+      "step": 44130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5639540623251836e-05,
+      "loss": 1.7626,
+      "step": 44140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.56362499588667e-05,
+      "loss": 1.5988,
+      "step": 44150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.563295929448155e-05,
+      "loss": 1.4393,
+      "step": 44160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.562966863009642e-05,
+      "loss": 1.6864,
+      "step": 44170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5626377965711274e-05,
+      "loss": 1.6121,
+      "step": 44180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.562308730132614e-05,
+      "loss": 1.687,
+      "step": 44190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5619796636941e-05,
+      "loss": 1.6658,
+      "step": 44200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.561650597255586e-05,
+      "loss": 1.7614,
+      "step": 44210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.561321530817072e-05,
+      "loss": 1.6564,
+      "step": 44220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560992464378558e-05,
+      "loss": 1.7528,
+      "step": 44230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560663397940044e-05,
+      "loss": 1.5751,
+      "step": 44240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.56033433150153e-05,
+      "loss": 1.6371,
+      "step": 44250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5600052650630164e-05,
+      "loss": 1.7586,
+      "step": 44260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5596761986245026e-05,
+      "loss": 1.7191,
+      "step": 44270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559347132185989e-05,
+      "loss": 1.4782,
+      "step": 44280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559018065747475e-05,
+      "loss": 1.5175,
+      "step": 44290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55868899930896e-05,
+      "loss": 1.6616,
+      "step": 44300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5583599328704464e-05,
+      "loss": 1.6759,
+      "step": 44310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5580308664319325e-05,
+      "loss": 1.5405,
+      "step": 44320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5577017999934187e-05,
+      "loss": 1.5501,
+      "step": 44330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557372733554905e-05,
+      "loss": 1.4762,
+      "step": 44340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557043667116391e-05,
+      "loss": 1.6666,
+      "step": 44350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556714600677877e-05,
+      "loss": 1.6644,
+      "step": 44360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556385534239363e-05,
+      "loss": 1.7227,
+      "step": 44370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556056467800849e-05,
+      "loss": 1.6965,
+      "step": 44380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5557274013623354e-05,
+      "loss": 1.7038,
+      "step": 44390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5553983349238215e-05,
+      "loss": 1.5769,
+      "step": 44400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555069268485307e-05,
+      "loss": 1.7212,
+      "step": 44410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554740202046794e-05,
+      "loss": 1.6238,
+      "step": 44420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554411135608279e-05,
+      "loss": 1.5404,
+      "step": 44430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554082069169766e-05,
+      "loss": 1.7052,
+      "step": 44440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5537530027312515e-05,
+      "loss": 1.7241,
+      "step": 44450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5534239362927376e-05,
+      "loss": 1.6454,
+      "step": 44460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.553094869854224e-05,
+      "loss": 1.5844,
+      "step": 44470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55276580341571e-05,
+      "loss": 1.6279,
+      "step": 44480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.552436736977196e-05,
+      "loss": 1.6366,
+      "step": 44490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.552107670538682e-05,
+      "loss": 1.6438,
+      "step": 44500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5517786041001676e-05,
+      "loss": 1.5648,
+      "step": 44510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5514495376616544e-05,
+      "loss": 1.6578,
+      "step": 44520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55112047122314e-05,
+      "loss": 1.798,
+      "step": 44530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5507914047846266e-05,
+      "loss": 1.6924,
+      "step": 44540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.550462338346112e-05,
+      "loss": 1.6958,
+      "step": 44550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.550133271907599e-05,
+      "loss": 1.5713,
+      "step": 44560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.549804205469084e-05,
+      "loss": 1.7552,
+      "step": 44570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5494751390305705e-05,
+      "loss": 1.5798,
+      "step": 44580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5491460725920566e-05,
+      "loss": 1.555,
+      "step": 44590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548817006153543e-05,
+      "loss": 1.7193,
+      "step": 44600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548487939715029e-05,
+      "loss": 1.5466,
+      "step": 44610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548158873276515e-05,
+      "loss": 1.6592,
+      "step": 44620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5478298068380004e-05,
+      "loss": 1.4618,
+      "step": 44630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.547500740399487e-05,
+      "loss": 1.6994,
+      "step": 44640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.547171673960973e-05,
+      "loss": 1.6032,
+      "step": 44650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5468426075224595e-05,
+      "loss": 1.4609,
+      "step": 44660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546513541083945e-05,
+      "loss": 1.6239,
+      "step": 44670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546184474645431e-05,
+      "loss": 1.54,
+      "step": 44680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.545855408206917e-05,
+      "loss": 1.5928,
+      "step": 44690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.545526341768403e-05,
+      "loss": 1.7048,
+      "step": 44700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5451972753298894e-05,
+      "loss": 1.4088,
+      "step": 44710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5448682088913756e-05,
+      "loss": 1.5433,
+      "step": 44720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.544539142452861e-05,
+      "loss": 1.7016,
+      "step": 44730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.544210076014348e-05,
+      "loss": 1.5531,
+      "step": 44740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543881009575833e-05,
+      "loss": 1.7789,
+      "step": 44750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54355194313732e-05,
+      "loss": 1.6586,
+      "step": 44760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5432228766988055e-05,
+      "loss": 1.7288,
+      "step": 44770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5428938102602916e-05,
+      "loss": 1.6006,
+      "step": 44780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542564743821778e-05,
+      "loss": 1.601,
+      "step": 44790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542235677383264e-05,
+      "loss": 1.5011,
+      "step": 44800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54190661094475e-05,
+      "loss": 1.5391,
+      "step": 44810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541577544506236e-05,
+      "loss": 1.8276,
+      "step": 44820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5412484780677216e-05,
+      "loss": 1.6677,
+      "step": 44830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5409194116292084e-05,
+      "loss": 1.6099,
+      "step": 44840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540590345190694e-05,
+      "loss": 1.5137,
+      "step": 44850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5402612787521807e-05,
+      "loss": 1.4905,
+      "step": 44860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539932212313666e-05,
+      "loss": 1.5795,
+      "step": 44870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539603145875152e-05,
+      "loss": 1.5338,
+      "step": 44880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5392740794366384e-05,
+      "loss": 1.6667,
+      "step": 44890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5389450129981245e-05,
+      "loss": 1.7369,
+      "step": 44900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5386159465596106e-05,
+      "loss": 1.6351,
+      "step": 44910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538286880121097e-05,
+      "loss": 1.5956,
+      "step": 44920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537957813682582e-05,
+      "loss": 1.6039,
+      "step": 44930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537628747244069e-05,
+      "loss": 1.6233,
+      "step": 44940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5372996808055544e-05,
+      "loss": 1.4982,
+      "step": 44950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536970614367041e-05,
+      "loss": 1.8452,
+      "step": 44960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536641547928527e-05,
+      "loss": 1.5872,
+      "step": 44970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.536312481490013e-05,
+      "loss": 1.5126,
+      "step": 44980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.535983415051499e-05,
+      "loss": 1.465,
+      "step": 44990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.535654348612985e-05,
+      "loss": 1.6943,
+      "step": 45000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.535325282174471e-05,
+      "loss": 1.4945,
+      "step": 45010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.534996215735957e-05,
+      "loss": 1.4903,
+      "step": 45020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5346671492974435e-05,
+      "loss": 1.5764,
+      "step": 45030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5343380828589296e-05,
+      "loss": 1.7111,
+      "step": 45040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.534009016420416e-05,
+      "loss": 1.6074,
+      "step": 45050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.533679949981902e-05,
+      "loss": 1.5221,
+      "step": 45060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.533350883543387e-05,
+      "loss": 1.5693,
+      "step": 45070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5330218171048734e-05,
+      "loss": 1.5679,
+      "step": 45080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5326927506663595e-05,
+      "loss": 1.6326,
+      "step": 45090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.532363684227846e-05,
+      "loss": 1.7767,
+      "step": 45100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.532034617789332e-05,
+      "loss": 1.6261,
+      "step": 45110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.531705551350818e-05,
+      "loss": 1.7665,
+      "step": 45120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.531376484912304e-05,
+      "loss": 1.542,
+      "step": 45130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.53104741847379e-05,
+      "loss": 1.3399,
+      "step": 45140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.530718352035276e-05,
+      "loss": 1.4683,
+      "step": 45150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5303892855967624e-05,
+      "loss": 1.6813,
+      "step": 45160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5300602191582486e-05,
+      "loss": 1.5805,
+      "step": 45170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.529731152719734e-05,
+      "loss": 1.5639,
+      "step": 45180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.529402086281221e-05,
+      "loss": 1.4997,
+      "step": 45190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.529073019842706e-05,
+      "loss": 1.5545,
+      "step": 45200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.528743953404193e-05,
+      "loss": 1.6824,
+      "step": 45210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5284148869656785e-05,
+      "loss": 1.6388,
+      "step": 45220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5280858205271646e-05,
+      "loss": 1.4747,
+      "step": 45230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527756754088651e-05,
+      "loss": 1.7289,
+      "step": 45240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527427687650137e-05,
+      "loss": 1.7439,
+      "step": 45250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527098621211623e-05,
+      "loss": 1.6446,
+      "step": 45260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526769554773109e-05,
+      "loss": 1.5805,
+      "step": 45270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5264404883345946e-05,
+      "loss": 1.5193,
+      "step": 45280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5261114218960814e-05,
+      "loss": 1.7818,
+      "step": 45290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525782355457567e-05,
+      "loss": 1.7981,
+      "step": 45300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5254532890190537e-05,
+      "loss": 1.8498,
+      "step": 45310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525124222580539e-05,
+      "loss": 1.6489,
+      "step": 45320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524795156142025e-05,
+      "loss": 1.6007,
+      "step": 45330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5244660897035114e-05,
+      "loss": 1.7304,
+      "step": 45340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5241370232649975e-05,
+      "loss": 1.6246,
+      "step": 45350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5238079568264836e-05,
+      "loss": 1.5998,
+      "step": 45360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52347889038797e-05,
+      "loss": 1.665,
+      "step": 45370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523149823949455e-05,
+      "loss": 1.6895,
+      "step": 45380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522820757510942e-05,
+      "loss": 1.652,
+      "step": 45390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5224916910724274e-05,
+      "loss": 1.5845,
+      "step": 45400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522162624633914e-05,
+      "loss": 1.5342,
+      "step": 45410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5218335581954e-05,
+      "loss": 1.6474,
+      "step": 45420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521504491756886e-05,
+      "loss": 1.699,
+      "step": 45430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521175425318372e-05,
+      "loss": 1.7888,
+      "step": 45440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520846358879858e-05,
+      "loss": 1.5928,
+      "step": 45450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520517292441344e-05,
+      "loss": 1.5533,
+      "step": 45460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52018822600283e-05,
+      "loss": 1.4408,
+      "step": 45470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.519859159564316e-05,
+      "loss": 1.6017,
+      "step": 45480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5195300931258026e-05,
+      "loss": 1.5258,
+      "step": 45490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.519201026687288e-05,
+      "loss": 1.8229,
+      "step": 45500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.518871960248775e-05,
+      "loss": 1.5947,
+      "step": 45510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.51854289381026e-05,
+      "loss": 1.6223,
+      "step": 45520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5182138273717464e-05,
+      "loss": 1.5967,
+      "step": 45530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5178847609332325e-05,
+      "loss": 1.5786,
+      "step": 45540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5175556944947187e-05,
+      "loss": 1.7027,
+      "step": 45550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.517226628056205e-05,
+      "loss": 1.575,
+      "step": 45560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.516897561617691e-05,
+      "loss": 1.8056,
+      "step": 45570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5165684951791764e-05,
+      "loss": 1.5615,
+      "step": 45580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.516239428740663e-05,
+      "loss": 1.5644,
+      "step": 45590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5159103623021486e-05,
+      "loss": 1.5221,
+      "step": 45600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5155812958636354e-05,
+      "loss": 1.4894,
+      "step": 45610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.515252229425121e-05,
+      "loss": 1.4621,
+      "step": 45620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.514923162986607e-05,
+      "loss": 1.6663,
+      "step": 45630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.514594096548093e-05,
+      "loss": 1.6933,
+      "step": 45640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.514265030109579e-05,
+      "loss": 1.7101,
+      "step": 45650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5139359636710654e-05,
+      "loss": 1.5135,
+      "step": 45660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5136068972325515e-05,
+      "loss": 1.8358,
+      "step": 45670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.513277830794037e-05,
+      "loss": 1.5359,
+      "step": 45680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.512948764355524e-05,
+      "loss": 1.662,
+      "step": 45690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.512619697917009e-05,
+      "loss": 1.5619,
+      "step": 45700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.512290631478496e-05,
+      "loss": 1.6622,
+      "step": 45710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5119615650399815e-05,
+      "loss": 1.5899,
+      "step": 45720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5116324986014676e-05,
+      "loss": 1.6231,
+      "step": 45730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.511303432162954e-05,
+      "loss": 1.5354,
+      "step": 45740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.51097436572444e-05,
+      "loss": 1.6726,
+      "step": 45750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510645299285926e-05,
+      "loss": 1.5948,
+      "step": 45760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510316232847412e-05,
+      "loss": 1.634,
+      "step": 45770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509987166408898e-05,
+      "loss": 1.5853,
+      "step": 45780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5096580999703843e-05,
+      "loss": 1.7518,
+      "step": 45790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5093290335318705e-05,
+      "loss": 1.6345,
+      "step": 45800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5089999670933566e-05,
+      "loss": 1.599,
+      "step": 45810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508670900654842e-05,
+      "loss": 1.5537,
+      "step": 45820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508341834216328e-05,
+      "loss": 1.5495,
+      "step": 45830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508012767777814e-05,
+      "loss": 1.6606,
+      "step": 45840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5076837013393004e-05,
+      "loss": 1.5436,
+      "step": 45850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5073546349007866e-05,
+      "loss": 1.6182,
+      "step": 45860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507025568462273e-05,
+      "loss": 1.5859,
+      "step": 45870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506696502023759e-05,
+      "loss": 1.4013,
+      "step": 45880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506367435585245e-05,
+      "loss": 1.5496,
+      "step": 45890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506038369146731e-05,
+      "loss": 1.6798,
+      "step": 45900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505709302708217e-05,
+      "loss": 1.6005,
+      "step": 45910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505380236269703e-05,
+      "loss": 1.5586,
+      "step": 45920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505051169831189e-05,
+      "loss": 1.5494,
+      "step": 45930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5047221033926756e-05,
+      "loss": 1.574,
+      "step": 45940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504393036954161e-05,
+      "loss": 1.5391,
+      "step": 45950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504063970515648e-05,
+      "loss": 1.4834,
+      "step": 45960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503734904077133e-05,
+      "loss": 1.5898,
+      "step": 45970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5034058376386194e-05,
+      "loss": 1.7635,
+      "step": 45980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5030767712001055e-05,
+      "loss": 1.4095,
+      "step": 45990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5027477047615917e-05,
+      "loss": 1.3911,
+      "step": 46000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.502418638323078e-05,
+      "loss": 1.676,
+      "step": 46010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.502089571884564e-05,
+      "loss": 1.5423,
+      "step": 46020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5017605054460494e-05,
+      "loss": 1.6829,
+      "step": 46030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.501431439007536e-05,
+      "loss": 1.7608,
+      "step": 46040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5011023725690216e-05,
+      "loss": 1.6684,
+      "step": 46050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5007733061305084e-05,
+      "loss": 1.4873,
+      "step": 46060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.500444239691994e-05,
+      "loss": 1.7751,
+      "step": 46070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.50011517325348e-05,
+      "loss": 1.5937,
+      "step": 46080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.499786106814966e-05,
+      "loss": 1.7932,
+      "step": 46090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.499457040376452e-05,
+      "loss": 1.6892,
+      "step": 46100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4991279739379384e-05,
+      "loss": 1.5126,
+      "step": 46110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4987989074994245e-05,
+      "loss": 1.7021,
+      "step": 46120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49846984106091e-05,
+      "loss": 1.6637,
+      "step": 46130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.498140774622397e-05,
+      "loss": 1.6618,
+      "step": 46140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.497811708183882e-05,
+      "loss": 1.673,
+      "step": 46150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.497482641745369e-05,
+      "loss": 1.604,
+      "step": 46160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4971535753068545e-05,
+      "loss": 1.6001,
+      "step": 46170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4968245088683406e-05,
+      "loss": 1.6195,
+      "step": 46180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.496495442429827e-05,
+      "loss": 1.6422,
+      "step": 46190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.496166375991313e-05,
+      "loss": 1.4947,
+      "step": 46200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.495837309552799e-05,
+      "loss": 1.9493,
+      "step": 46210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.495508243114285e-05,
+      "loss": 1.5619,
+      "step": 46220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4951791766757705e-05,
+      "loss": 1.6849,
+      "step": 46230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4948501102372573e-05,
+      "loss": 1.5438,
+      "step": 46240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.494521043798743e-05,
+      "loss": 1.6325,
+      "step": 46250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4941919773602296e-05,
+      "loss": 1.7585,
+      "step": 46260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493862910921715e-05,
+      "loss": 1.4089,
+      "step": 46270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493533844483201e-05,
+      "loss": 1.7039,
+      "step": 46280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493204778044687e-05,
+      "loss": 1.5698,
+      "step": 46290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4928757116061734e-05,
+      "loss": 1.5186,
+      "step": 46300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4925466451676595e-05,
+      "loss": 1.9179,
+      "step": 46310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492217578729146e-05,
+      "loss": 1.5708,
+      "step": 46320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491888512290631e-05,
+      "loss": 1.7149,
+      "step": 46330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491559445852118e-05,
+      "loss": 1.5138,
+      "step": 46340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4912303794136034e-05,
+      "loss": 1.628,
+      "step": 46350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49090131297509e-05,
+      "loss": 1.8286,
+      "step": 46360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4905722465365756e-05,
+      "loss": 1.9132,
+      "step": 46370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490243180098062e-05,
+      "loss": 1.5632,
+      "step": 46380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489914113659548e-05,
+      "loss": 1.5554,
+      "step": 46390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489585047221034e-05,
+      "loss": 1.6617,
+      "step": 46400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48925598078252e-05,
+      "loss": 1.8016,
+      "step": 46410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488926914344006e-05,
+      "loss": 1.6477,
+      "step": 46420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488597847905492e-05,
+      "loss": 1.7562,
+      "step": 46430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4882687814669785e-05,
+      "loss": 1.6045,
+      "step": 46440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487939715028464e-05,
+      "loss": 1.6508,
+      "step": 46450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487610648589951e-05,
+      "loss": 1.6581,
+      "step": 46460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487281582151436e-05,
+      "loss": 1.6137,
+      "step": 46470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4869525157129223e-05,
+      "loss": 1.5728,
+      "step": 46480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4866234492744085e-05,
+      "loss": 1.5807,
+      "step": 46490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4862943828358946e-05,
+      "loss": 1.5906,
+      "step": 46500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.485965316397381e-05,
+      "loss": 1.6188,
+      "step": 46510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.485636249958867e-05,
+      "loss": 1.7433,
+      "step": 46520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.485307183520353e-05,
+      "loss": 1.6888,
+      "step": 46530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.484978117081839e-05,
+      "loss": 1.6947,
+      "step": 46540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.484649050643325e-05,
+      "loss": 1.6558,
+      "step": 46550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4843199842048114e-05,
+      "loss": 1.6146,
+      "step": 46560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4839909177662975e-05,
+      "loss": 1.3883,
+      "step": 46570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483661851327783e-05,
+      "loss": 1.7441,
+      "step": 46580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483332784889269e-05,
+      "loss": 1.5805,
+      "step": 46590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483003718450755e-05,
+      "loss": 1.789,
+      "step": 46600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482674652012241e-05,
+      "loss": 1.6428,
+      "step": 46610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4823455855737274e-05,
+      "loss": 1.487,
+      "step": 46620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4820165191352136e-05,
+      "loss": 1.8363,
+      "step": 46630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4816874526967e-05,
+      "loss": 1.6079,
+      "step": 46640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481358386258186e-05,
+      "loss": 1.5822,
+      "step": 46650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481029319819672e-05,
+      "loss": 1.4518,
+      "step": 46660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.480700253381158e-05,
+      "loss": 1.7307,
+      "step": 46670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4803711869426435e-05,
+      "loss": 1.7832,
+      "step": 46680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.48004212050413e-05,
+      "loss": 1.4538,
+      "step": 46690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479713054065616e-05,
+      "loss": 1.9819,
+      "step": 46700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4793839876271026e-05,
+      "loss": 1.6021,
+      "step": 46710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479054921188588e-05,
+      "loss": 1.6272,
+      "step": 46720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.478725854750074e-05,
+      "loss": 1.6901,
+      "step": 46730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47839678831156e-05,
+      "loss": 1.7302,
+      "step": 46740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4780677218730464e-05,
+      "loss": 1.7064,
+      "step": 46750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4777386554345325e-05,
+      "loss": 1.6359,
+      "step": 46760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477409588996019e-05,
+      "loss": 1.5588,
+      "step": 46770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477080522557504e-05,
+      "loss": 1.3869,
+      "step": 46780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476751456118991e-05,
+      "loss": 1.5542,
+      "step": 46790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4764223896804764e-05,
+      "loss": 1.3914,
+      "step": 46800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476093323241963e-05,
+      "loss": 1.8414,
+      "step": 46810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4757642568034486e-05,
+      "loss": 1.7537,
+      "step": 46820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475435190364935e-05,
+      "loss": 1.7433,
+      "step": 46830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475106123926421e-05,
+      "loss": 1.5526,
+      "step": 46840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474777057487907e-05,
+      "loss": 1.7237,
+      "step": 46850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474447991049393e-05,
+      "loss": 1.7966,
+      "step": 46860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474118924610879e-05,
+      "loss": 1.5185,
+      "step": 46870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473789858172365e-05,
+      "loss": 1.657,
+      "step": 46880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4734607917338515e-05,
+      "loss": 1.7311,
+      "step": 46890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473131725295337e-05,
+      "loss": 1.5943,
+      "step": 46900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472802658856824e-05,
+      "loss": 1.6308,
+      "step": 46910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472473592418309e-05,
+      "loss": 1.6785,
+      "step": 46920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4721445259797953e-05,
+      "loss": 1.5589,
+      "step": 46930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4718154595412815e-05,
+      "loss": 1.8612,
+      "step": 46940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4714863931027676e-05,
+      "loss": 1.6408,
+      "step": 46950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471157326664254e-05,
+      "loss": 1.5286,
+      "step": 46960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47082826022574e-05,
+      "loss": 1.5724,
+      "step": 46970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470499193787225e-05,
+      "loss": 1.5937,
+      "step": 46980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470170127348712e-05,
+      "loss": 1.7815,
+      "step": 46990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4698410609101976e-05,
+      "loss": 1.7621,
+      "step": 47000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4695119944716844e-05,
+      "loss": 1.4347,
+      "step": 47010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.46918292803317e-05,
+      "loss": 1.7743,
+      "step": 47020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.468853861594656e-05,
+      "loss": 1.5118,
+      "step": 47030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.468524795156142e-05,
+      "loss": 1.5882,
+      "step": 47040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.468195728717628e-05,
+      "loss": 1.6686,
+      "step": 47050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.467866662279114e-05,
+      "loss": 1.6511,
+      "step": 47060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4675375958406004e-05,
+      "loss": 1.6476,
+      "step": 47070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.467208529402086e-05,
+      "loss": 1.5546,
+      "step": 47080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.466879462963573e-05,
+      "loss": 1.514,
+      "step": 47090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.466550396525058e-05,
+      "loss": 1.5287,
+      "step": 47100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.466221330086545e-05,
+      "loss": 1.5256,
+      "step": 47110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4658922636480304e-05,
+      "loss": 1.6136,
+      "step": 47120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4655631972095165e-05,
+      "loss": 1.7245,
+      "step": 47130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4652341307710026e-05,
+      "loss": 1.6311,
+      "step": 47140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464905064332489e-05,
+      "loss": 1.6953,
+      "step": 47150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464575997893975e-05,
+      "loss": 1.6154,
+      "step": 47160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464246931455461e-05,
+      "loss": 1.5552,
+      "step": 47170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.463917865016947e-05,
+      "loss": 1.5497,
+      "step": 47180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.463588798578433e-05,
+      "loss": 1.5787,
+      "step": 47190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.463259732139919e-05,
+      "loss": 1.7298,
+      "step": 47200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4629306657014055e-05,
+      "loss": 1.6686,
+      "step": 47210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.462601599262891e-05,
+      "loss": 1.4806,
+      "step": 47220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.462272532824377e-05,
+      "loss": 1.5477,
+      "step": 47230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.461943466385863e-05,
+      "loss": 1.5118,
+      "step": 47240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4616143999473494e-05,
+      "loss": 1.6186,
+      "step": 47250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4612853335088355e-05,
+      "loss": 1.4891,
+      "step": 47260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4609562670703216e-05,
+      "loss": 1.5543,
+      "step": 47270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460627200631808e-05,
+      "loss": 1.5677,
+      "step": 47280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460298134193294e-05,
+      "loss": 1.5076,
+      "step": 47290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45996906775478e-05,
+      "loss": 1.83,
+      "step": 47300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459640001316266e-05,
+      "loss": 1.635,
+      "step": 47310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459310934877752e-05,
+      "loss": 1.6609,
+      "step": 47320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458981868439238e-05,
+      "loss": 1.5404,
+      "step": 47330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4586528020007245e-05,
+      "loss": 1.6049,
+      "step": 47340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45832373556221e-05,
+      "loss": 1.7519,
+      "step": 47350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457994669123696e-05,
+      "loss": 1.4698,
+      "step": 47360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457665602685182e-05,
+      "loss": 1.7528,
+      "step": 47370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457336536246668e-05,
+      "loss": 1.6055,
+      "step": 47380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4570074698081545e-05,
+      "loss": 1.6422,
+      "step": 47390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4566784033696406e-05,
+      "loss": 1.5964,
+      "step": 47400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456349336931127e-05,
+      "loss": 1.7838,
+      "step": 47410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456020270492613e-05,
+      "loss": 1.5404,
+      "step": 47420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455691204054099e-05,
+      "loss": 1.6495,
+      "step": 47430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455362137615585e-05,
+      "loss": 1.7154,
+      "step": 47440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4550330711770705e-05,
+      "loss": 1.5578,
+      "step": 47450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4547040047385573e-05,
+      "loss": 1.6146,
+      "step": 47460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454374938300043e-05,
+      "loss": 1.6519,
+      "step": 47470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4540458718615296e-05,
+      "loss": 1.5928,
+      "step": 47480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453716805423015e-05,
+      "loss": 1.9296,
+      "step": 47490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453387738984501e-05,
+      "loss": 1.705,
+      "step": 47500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453058672545987e-05,
+      "loss": 1.5052,
+      "step": 47510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4527296061074734e-05,
+      "loss": 1.5244,
+      "step": 47520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4524005396689596e-05,
+      "loss": 1.6159,
+      "step": 47530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.452071473230446e-05,
+      "loss": 1.4801,
+      "step": 47540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.451742406791931e-05,
+      "loss": 1.5267,
+      "step": 47550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.451413340353418e-05,
+      "loss": 1.653,
+      "step": 47560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4510842739149034e-05,
+      "loss": 1.5755,
+      "step": 47570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.45075520747639e-05,
+      "loss": 1.5744,
+      "step": 47580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4504261410378756e-05,
+      "loss": 1.6278,
+      "step": 47590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.450097074599362e-05,
+      "loss": 1.4454,
+      "step": 47600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449768008160848e-05,
+      "loss": 1.5994,
+      "step": 47610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449438941722334e-05,
+      "loss": 1.4595,
+      "step": 47620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44910987528382e-05,
+      "loss": 1.5734,
+      "step": 47630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.448780808845306e-05,
+      "loss": 1.6271,
+      "step": 47640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.448451742406792e-05,
+      "loss": 1.8672,
+      "step": 47650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4481226759682785e-05,
+      "loss": 1.5791,
+      "step": 47660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.447793609529764e-05,
+      "loss": 1.7451,
+      "step": 47670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.447464543091251e-05,
+      "loss": 1.5334,
+      "step": 47680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.447135476652736e-05,
+      "loss": 1.6513,
+      "step": 47690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4468064102142224e-05,
+      "loss": 1.5953,
+      "step": 47700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4464773437757085e-05,
+      "loss": 1.6101,
+      "step": 47710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4461482773371946e-05,
+      "loss": 1.5799,
+      "step": 47720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.445819210898681e-05,
+      "loss": 1.6829,
+      "step": 47730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.445490144460167e-05,
+      "loss": 1.5641,
+      "step": 47740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.445161078021652e-05,
+      "loss": 1.5953,
+      "step": 47750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.444832011583139e-05,
+      "loss": 1.8367,
+      "step": 47760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4445029451446246e-05,
+      "loss": 1.6865,
+      "step": 47770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4441738787061114e-05,
+      "loss": 1.6297,
+      "step": 47780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443844812267597e-05,
+      "loss": 1.7688,
+      "step": 47790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443515745829083e-05,
+      "loss": 1.4331,
+      "step": 47800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443186679390569e-05,
+      "loss": 1.7972,
+      "step": 47810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442857612952055e-05,
+      "loss": 1.4565,
+      "step": 47820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442528546513541e-05,
+      "loss": 1.6752,
+      "step": 47830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4421994800750275e-05,
+      "loss": 1.438,
+      "step": 47840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441870413636513e-05,
+      "loss": 1.5097,
+      "step": 47850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441541347198e-05,
+      "loss": 1.6766,
+      "step": 47860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441212280759485e-05,
+      "loss": 1.55,
+      "step": 47870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440883214320972e-05,
+      "loss": 1.4806,
+      "step": 47880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4405541478824574e-05,
+      "loss": 1.663,
+      "step": 47890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4402250814439435e-05,
+      "loss": 1.9989,
+      "step": 47900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43989601500543e-05,
+      "loss": 1.5716,
+      "step": 47910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439566948566916e-05,
+      "loss": 1.7821,
+      "step": 47920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439237882128402e-05,
+      "loss": 1.5483,
+      "step": 47930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438908815689888e-05,
+      "loss": 1.6207,
+      "step": 47940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4385797492513735e-05,
+      "loss": 2.0449,
+      "step": 47950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43825068281286e-05,
+      "loss": 1.7122,
+      "step": 47960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437921616374346e-05,
+      "loss": 1.6298,
+      "step": 47970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4375925499358326e-05,
+      "loss": 1.7519,
+      "step": 47980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437263483497318e-05,
+      "loss": 1.5337,
+      "step": 47990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436934417058804e-05,
+      "loss": 1.6707,
+      "step": 48000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43660535062029e-05,
+      "loss": 1.585,
+      "step": 48010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4362762841817764e-05,
+      "loss": 1.7212,
+      "step": 48020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4359472177432625e-05,
+      "loss": 1.5108,
+      "step": 48030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4356181513047486e-05,
+      "loss": 1.4555,
+      "step": 48040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.435289084866235e-05,
+      "loss": 1.5908,
+      "step": 48050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.434960018427721e-05,
+      "loss": 1.4997,
+      "step": 48060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.434630951989207e-05,
+      "loss": 1.699,
+      "step": 48070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.434301885550693e-05,
+      "loss": 1.718,
+      "step": 48080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.433972819112179e-05,
+      "loss": 1.4996,
+      "step": 48090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.433643752673665e-05,
+      "loss": 1.6604,
+      "step": 48100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.433314686235151e-05,
+      "loss": 1.5234,
+      "step": 48110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432985619796637e-05,
+      "loss": 1.5731,
+      "step": 48120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432656553358123e-05,
+      "loss": 1.648,
+      "step": 48130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432327486919609e-05,
+      "loss": 1.5862,
+      "step": 48140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4319984204810954e-05,
+      "loss": 1.6544,
+      "step": 48150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4316693540425815e-05,
+      "loss": 1.6619,
+      "step": 48160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4313402876040676e-05,
+      "loss": 1.4803,
+      "step": 48170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.431011221165554e-05,
+      "loss": 1.509,
+      "step": 48180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.43068215472704e-05,
+      "loss": 1.6649,
+      "step": 48190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.430353088288525e-05,
+      "loss": 1.5956,
+      "step": 48200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.430024021850012e-05,
+      "loss": 1.6048,
+      "step": 48210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4296949554114976e-05,
+      "loss": 1.4604,
+      "step": 48220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4293658889729844e-05,
+      "loss": 1.4165,
+      "step": 48230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42903682253447e-05,
+      "loss": 1.7493,
+      "step": 48240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.428707756095956e-05,
+      "loss": 1.6681,
+      "step": 48250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.428378689657442e-05,
+      "loss": 1.5384,
+      "step": 48260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.428049623218928e-05,
+      "loss": 1.8574,
+      "step": 48270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.427720556780414e-05,
+      "loss": 1.6553,
+      "step": 48280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4273914903419004e-05,
+      "loss": 1.8318,
+      "step": 48290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.427062423903386e-05,
+      "loss": 1.4673,
+      "step": 48300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426733357464873e-05,
+      "loss": 1.6628,
+      "step": 48310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426404291026358e-05,
+      "loss": 1.5324,
+      "step": 48320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426075224587845e-05,
+      "loss": 1.5141,
+      "step": 48330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4257461581493304e-05,
+      "loss": 1.6349,
+      "step": 48340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4254170917108165e-05,
+      "loss": 1.5788,
+      "step": 48350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4250880252723027e-05,
+      "loss": 1.5232,
+      "step": 48360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424758958833789e-05,
+      "loss": 1.478,
+      "step": 48370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424429892395275e-05,
+      "loss": 1.7093,
+      "step": 48380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424100825956761e-05,
+      "loss": 1.5459,
+      "step": 48390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4237717595182465e-05,
+      "loss": 1.6045,
+      "step": 48400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423442693079733e-05,
+      "loss": 1.5928,
+      "step": 48410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423113626641219e-05,
+      "loss": 1.5972,
+      "step": 48420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4227845602027055e-05,
+      "loss": 1.7825,
+      "step": 48430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422455493764191e-05,
+      "loss": 1.5923,
+      "step": 48440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422126427325677e-05,
+      "loss": 1.6236,
+      "step": 48450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421797360887163e-05,
+      "loss": 1.5886,
+      "step": 48460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4214682944486494e-05,
+      "loss": 1.5948,
+      "step": 48470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4211392280101355e-05,
+      "loss": 1.6789,
+      "step": 48480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4208101615716216e-05,
+      "loss": 1.6701,
+      "step": 48490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420481095133107e-05,
+      "loss": 1.6491,
+      "step": 48500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420152028694594e-05,
+      "loss": 1.6054,
+      "step": 48510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419822962256079e-05,
+      "loss": 1.9039,
+      "step": 48520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.419493895817566e-05,
+      "loss": 1.637,
+      "step": 48530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4191648293790516e-05,
+      "loss": 1.5689,
+      "step": 48540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.418835762940538e-05,
+      "loss": 1.5329,
+      "step": 48550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.418506696502024e-05,
+      "loss": 1.4082,
+      "step": 48560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.41817763006351e-05,
+      "loss": 1.5961,
+      "step": 48570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.417848563624996e-05,
+      "loss": 1.5896,
+      "step": 48580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.417519497186482e-05,
+      "loss": 1.6065,
+      "step": 48590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.417190430747968e-05,
+      "loss": 1.6408,
+      "step": 48600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4168613643094545e-05,
+      "loss": 1.544,
+      "step": 48610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.41653229787094e-05,
+      "loss": 1.4793,
+      "step": 48620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.416203231432427e-05,
+      "loss": 1.6063,
+      "step": 48630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.415874164993912e-05,
+      "loss": 1.5848,
+      "step": 48640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.415545098555398e-05,
+      "loss": 1.5619,
+      "step": 48650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4152160321168844e-05,
+      "loss": 1.6098,
+      "step": 48660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4148869656783706e-05,
+      "loss": 1.6019,
+      "step": 48670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414557899239857e-05,
+      "loss": 1.4936,
+      "step": 48680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414228832801343e-05,
+      "loss": 1.4812,
+      "step": 48690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.413899766362829e-05,
+      "loss": 1.6004,
+      "step": 48700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.413570699924315e-05,
+      "loss": 1.6012,
+      "step": 48710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4132416334858005e-05,
+      "loss": 1.5415,
+      "step": 48720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.412912567047287e-05,
+      "loss": 1.723,
+      "step": 48730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.412583500608773e-05,
+      "loss": 1.5279,
+      "step": 48740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.412254434170259e-05,
+      "loss": 1.5979,
+      "step": 48750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.411925367731745e-05,
+      "loss": 1.745,
+      "step": 48760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.411596301293231e-05,
+      "loss": 1.6332,
+      "step": 48770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.411267234854717e-05,
+      "loss": 1.581,
+      "step": 48780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4109381684162034e-05,
+      "loss": 1.6839,
+      "step": 48790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4106091019776895e-05,
+      "loss": 1.6249,
+      "step": 48800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4102800355391757e-05,
+      "loss": 1.4947,
+      "step": 48810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409950969100662e-05,
+      "loss": 1.6745,
+      "step": 48820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409621902662148e-05,
+      "loss": 1.9825,
+      "step": 48830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409292836223634e-05,
+      "loss": 1.621,
+      "step": 48840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4089637697851195e-05,
+      "loss": 1.5579,
+      "step": 48850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408634703346606e-05,
+      "loss": 1.4588,
+      "step": 48860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408305636908092e-05,
+      "loss": 1.6769,
+      "step": 48870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407976570469578e-05,
+      "loss": 1.6403,
+      "step": 48880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407647504031064e-05,
+      "loss": 1.5674,
+      "step": 48890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40731843759255e-05,
+      "loss": 1.6106,
+      "step": 48900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406989371154036e-05,
+      "loss": 1.6747,
+      "step": 48910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4066603047155224e-05,
+      "loss": 1.4926,
+      "step": 48920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4063312382770085e-05,
+      "loss": 1.6637,
+      "step": 48930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4060021718384946e-05,
+      "loss": 1.6165,
+      "step": 48940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40567310539998e-05,
+      "loss": 1.6515,
+      "step": 48950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405344038961467e-05,
+      "loss": 1.5917,
+      "step": 48960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405014972522952e-05,
+      "loss": 1.6274,
+      "step": 48970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404685906084439e-05,
+      "loss": 1.5478,
+      "step": 48980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4043568396459246e-05,
+      "loss": 1.5292,
+      "step": 48990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404027773207411e-05,
+      "loss": 1.601,
+      "step": 49000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403698706768897e-05,
+      "loss": 1.5837,
+      "step": 49010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403369640330383e-05,
+      "loss": 1.4557,
+      "step": 49020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403040573891869e-05,
+      "loss": 1.585,
+      "step": 49030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.402711507453355e-05,
+      "loss": 1.5774,
+      "step": 49040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4023824410148407e-05,
+      "loss": 1.5204,
+      "step": 49050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4020533745763275e-05,
+      "loss": 1.6502,
+      "step": 49060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.401724308137813e-05,
+      "loss": 1.6025,
+      "step": 49070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4013952416993e-05,
+      "loss": 1.37,
+      "step": 49080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.401066175260785e-05,
+      "loss": 1.6525,
+      "step": 49090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.400737108822271e-05,
+      "loss": 1.8752,
+      "step": 49100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4004080423837574e-05,
+      "loss": 1.6524,
+      "step": 49110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4000789759452435e-05,
+      "loss": 1.6022,
+      "step": 49120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39974990950673e-05,
+      "loss": 1.4468,
+      "step": 49130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.399420843068216e-05,
+      "loss": 1.8204,
+      "step": 49140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.399091776629701e-05,
+      "loss": 1.5725,
+      "step": 49150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.398762710191188e-05,
+      "loss": 1.5795,
+      "step": 49160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3984336437526735e-05,
+      "loss": 1.7172,
+      "step": 49170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39810457731416e-05,
+      "loss": 1.6921,
+      "step": 49180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397775510875646e-05,
+      "loss": 1.639,
+      "step": 49190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397446444437132e-05,
+      "loss": 1.7782,
+      "step": 49200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397117377998618e-05,
+      "loss": 1.5599,
+      "step": 49210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.396788311560104e-05,
+      "loss": 1.5459,
+      "step": 49220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39645924512159e-05,
+      "loss": 1.619,
+      "step": 49230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3961301786830764e-05,
+      "loss": 1.6652,
+      "step": 49240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.395801112244562e-05,
+      "loss": 1.5243,
+      "step": 49250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3954720458060486e-05,
+      "loss": 1.568,
+      "step": 49260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.395142979367534e-05,
+      "loss": 1.5281,
+      "step": 49270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.394813912929021e-05,
+      "loss": 1.7858,
+      "step": 49280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3944848464905063e-05,
+      "loss": 1.5836,
+      "step": 49290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3941557800519925e-05,
+      "loss": 1.6042,
+      "step": 49300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3938267136134786e-05,
+      "loss": 1.698,
+      "step": 49310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393497647174965e-05,
+      "loss": 1.663,
+      "step": 49320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393168580736451e-05,
+      "loss": 1.5511,
+      "step": 49330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392839514297937e-05,
+      "loss": 1.5684,
+      "step": 49340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3925104478594224e-05,
+      "loss": 1.5974,
+      "step": 49350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392181381420909e-05,
+      "loss": 1.554,
+      "step": 49360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391852314982395e-05,
+      "loss": 1.908,
+      "step": 49370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3915232485438815e-05,
+      "loss": 1.4744,
+      "step": 49380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391194182105367e-05,
+      "loss": 1.4872,
+      "step": 49390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390865115666853e-05,
+      "loss": 1.6218,
+      "step": 49400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390536049228339e-05,
+      "loss": 1.5602,
+      "step": 49410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390206982789825e-05,
+      "loss": 1.5701,
+      "step": 49420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3898779163513114e-05,
+      "loss": 1.6746,
+      "step": 49430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3895488499127976e-05,
+      "loss": 1.7249,
+      "step": 49440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389219783474284e-05,
+      "loss": 1.9757,
+      "step": 49450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38889071703577e-05,
+      "loss": 1.6375,
+      "step": 49460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388561650597255e-05,
+      "loss": 1.3387,
+      "step": 49470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388232584158742e-05,
+      "loss": 1.5311,
+      "step": 49480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3879035177202275e-05,
+      "loss": 1.477,
+      "step": 49490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3875744512817137e-05,
+      "loss": 1.6116,
+      "step": 49500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3872453848432e-05,
+      "loss": 1.5733,
+      "step": 49510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386916318404686e-05,
+      "loss": 1.4579,
+      "step": 49520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386587251966172e-05,
+      "loss": 1.5822,
+      "step": 49530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386258185527658e-05,
+      "loss": 1.4766,
+      "step": 49540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.385929119089144e-05,
+      "loss": 1.5574,
+      "step": 49550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3856000526506304e-05,
+      "loss": 1.5043,
+      "step": 49560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3852709862121165e-05,
+      "loss": 1.6303,
+      "step": 49570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.384941919773603e-05,
+      "loss": 1.634,
+      "step": 49580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.384612853335089e-05,
+      "loss": 1.6275,
+      "step": 49590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.384283786896574e-05,
+      "loss": 1.684,
+      "step": 49600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.383954720458061e-05,
+      "loss": 1.664,
+      "step": 49610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3836256540195465e-05,
+      "loss": 1.5325,
+      "step": 49620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.383296587581033e-05,
+      "loss": 1.628,
+      "step": 49630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382967521142519e-05,
+      "loss": 1.5654,
+      "step": 49640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382638454704005e-05,
+      "loss": 1.5511,
+      "step": 49650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382309388265491e-05,
+      "loss": 1.566,
+      "step": 49660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.381980321826977e-05,
+      "loss": 1.7058,
+      "step": 49670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.381651255388463e-05,
+      "loss": 1.6513,
+      "step": 49680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3813221889499494e-05,
+      "loss": 1.6737,
+      "step": 49690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.380993122511435e-05,
+      "loss": 1.5063,
+      "step": 49700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3806640560729216e-05,
+      "loss": 1.5263,
+      "step": 49710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.380334989634407e-05,
+      "loss": 1.817,
+      "step": 49720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.380005923195894e-05,
+      "loss": 1.6914,
+      "step": 49730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3796768567573793e-05,
+      "loss": 1.491,
+      "step": 49740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3793477903188655e-05,
+      "loss": 1.7357,
+      "step": 49750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3790187238803516e-05,
+      "loss": 1.6859,
+      "step": 49760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.378689657441838e-05,
+      "loss": 1.5998,
+      "step": 49770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.378360591003324e-05,
+      "loss": 1.5507,
+      "step": 49780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37803152456481e-05,
+      "loss": 1.5511,
+      "step": 49790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3777024581262954e-05,
+      "loss": 1.5761,
+      "step": 49800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.377373391687782e-05,
+      "loss": 1.6611,
+      "step": 49810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.377044325249268e-05,
+      "loss": 1.6233,
+      "step": 49820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3767152588107545e-05,
+      "loss": 1.5693,
+      "step": 49830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37638619237224e-05,
+      "loss": 1.6018,
+      "step": 49840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376057125933726e-05,
+      "loss": 1.5063,
+      "step": 49850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375728059495212e-05,
+      "loss": 1.4998,
+      "step": 49860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375398993056698e-05,
+      "loss": 1.6095,
+      "step": 49870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3750699266181844e-05,
+      "loss": 1.5873,
+      "step": 49880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3747408601796706e-05,
+      "loss": 1.5187,
+      "step": 49890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374411793741156e-05,
+      "loss": 1.5498,
+      "step": 49900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374082727302643e-05,
+      "loss": 1.6078,
+      "step": 49910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373753660864128e-05,
+      "loss": 1.5782,
+      "step": 49920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373424594425615e-05,
+      "loss": 1.5102,
+      "step": 49930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3730955279871005e-05,
+      "loss": 1.5587,
+      "step": 49940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3727664615485866e-05,
+      "loss": 1.6292,
+      "step": 49950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372437395110073e-05,
+      "loss": 1.5405,
+      "step": 49960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372108328671559e-05,
+      "loss": 1.3254,
+      "step": 49970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371779262233045e-05,
+      "loss": 1.596,
+      "step": 49980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371450195794531e-05,
+      "loss": 1.5078,
+      "step": 49990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3711211293560166e-05,
+      "loss": 1.5944,
+      "step": 50000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3707920629175034e-05,
+      "loss": 1.6062,
+      "step": 50010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370462996478989e-05,
+      "loss": 1.5513,
+      "step": 50020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3701339300404757e-05,
+      "loss": 1.6616,
+      "step": 50030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369804863601961e-05,
+      "loss": 1.6256,
+      "step": 50040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369475797163447e-05,
+      "loss": 1.6635,
+      "step": 50050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3691467307249334e-05,
+      "loss": 1.5462,
+      "step": 50060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3688176642864195e-05,
+      "loss": 1.7641,
+      "step": 50070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3684885978479056e-05,
+      "loss": 1.5193,
+      "step": 50080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.368159531409392e-05,
+      "loss": 1.6275,
+      "step": 50090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.367830464970877e-05,
+      "loss": 1.5431,
+      "step": 50100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.367501398532364e-05,
+      "loss": 1.5641,
+      "step": 50110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3671723320938494e-05,
+      "loss": 1.5825,
+      "step": 50120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.366843265655336e-05,
+      "loss": 1.5207,
+      "step": 50130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.366514199216822e-05,
+      "loss": 1.6486,
+      "step": 50140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.366185132778308e-05,
+      "loss": 1.609,
+      "step": 50150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.365856066339794e-05,
+      "loss": 1.644,
+      "step": 50160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.36552699990128e-05,
+      "loss": 1.5935,
+      "step": 50170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.365197933462766e-05,
+      "loss": 1.5244,
+      "step": 50180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.364868867024252e-05,
+      "loss": 1.6509,
+      "step": 50190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3645398005857385e-05,
+      "loss": 1.512,
+      "step": 50200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3642107341472246e-05,
+      "loss": 1.6053,
+      "step": 50210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.363881667708711e-05,
+      "loss": 1.4673,
+      "step": 50220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.363552601270197e-05,
+      "loss": 1.8973,
+      "step": 50230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.363223534831682e-05,
+      "loss": 1.7689,
+      "step": 50240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.362894468393169e-05,
+      "loss": 1.4973,
+      "step": 50250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3625654019546545e-05,
+      "loss": 1.6035,
+      "step": 50260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.362236335516141e-05,
+      "loss": 1.5488,
+      "step": 50270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.361907269077627e-05,
+      "loss": 1.7448,
+      "step": 50280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.361578202639113e-05,
+      "loss": 1.4407,
+      "step": 50290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.361249136200599e-05,
+      "loss": 1.7291,
+      "step": 50300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.360920069762085e-05,
+      "loss": 1.5742,
+      "step": 50310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.360591003323571e-05,
+      "loss": 1.6248,
+      "step": 50320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3602619368850574e-05,
+      "loss": 1.6426,
+      "step": 50330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3599328704465436e-05,
+      "loss": 1.6203,
+      "step": 50340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35960380400803e-05,
+      "loss": 1.6443,
+      "step": 50350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359274737569516e-05,
+      "loss": 1.4324,
+      "step": 50360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358945671131001e-05,
+      "loss": 1.5592,
+      "step": 50370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358616604692488e-05,
+      "loss": 1.5697,
+      "step": 50380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3582875382539735e-05,
+      "loss": 1.7606,
+      "step": 50390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3579584718154596e-05,
+      "loss": 1.6821,
+      "step": 50400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357629405376946e-05,
+      "loss": 1.7328,
+      "step": 50410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357300338938432e-05,
+      "loss": 1.4805,
+      "step": 50420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356971272499918e-05,
+      "loss": 1.5746,
+      "step": 50430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356642206061404e-05,
+      "loss": 1.6678,
+      "step": 50440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35631313962289e-05,
+      "loss": 1.7502,
+      "step": 50450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3559840731843764e-05,
+      "loss": 1.6188,
+      "step": 50460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355655006745862e-05,
+      "loss": 1.6405,
+      "step": 50470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3553259403073487e-05,
+      "loss": 1.4666,
+      "step": 50480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354996873868834e-05,
+      "loss": 1.5844,
+      "step": 50490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354667807430321e-05,
+      "loss": 1.4778,
+      "step": 50500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3543387409918064e-05,
+      "loss": 1.5761,
+      "step": 50510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3540096745532925e-05,
+      "loss": 1.7346,
+      "step": 50520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3536806081147786e-05,
+      "loss": 1.5437,
+      "step": 50530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353351541676265e-05,
+      "loss": 1.5102,
+      "step": 50540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353022475237751e-05,
+      "loss": 1.4776,
+      "step": 50550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352693408799237e-05,
+      "loss": 1.7403,
+      "step": 50560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3523643423607224e-05,
+      "loss": 1.6105,
+      "step": 50570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.352035275922209e-05,
+      "loss": 2.0407,
+      "step": 50580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.351706209483695e-05,
+      "loss": 1.5954,
+      "step": 50590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3513771430451815e-05,
+      "loss": 1.6817,
+      "step": 50600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.351048076606667e-05,
+      "loss": 1.4273,
+      "step": 50610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.350719010168153e-05,
+      "loss": 1.5135,
+      "step": 50620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.350389943729639e-05,
+      "loss": 1.5101,
+      "step": 50630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.350060877291125e-05,
+      "loss": 1.5116,
+      "step": 50640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3497318108526115e-05,
+      "loss": 1.6155,
+      "step": 50650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3494027444140976e-05,
+      "loss": 1.4977,
+      "step": 50660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.349073677975583e-05,
+      "loss": 1.6989,
+      "step": 50670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34874461153707e-05,
+      "loss": 1.6177,
+      "step": 50680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.348415545098555e-05,
+      "loss": 1.6367,
+      "step": 50690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.348086478660042e-05,
+      "loss": 1.6492,
+      "step": 50700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3477574122215275e-05,
+      "loss": 1.4682,
+      "step": 50710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.347428345783014e-05,
+      "loss": 1.476,
+      "step": 50720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3470992793445e-05,
+      "loss": 1.4794,
+      "step": 50730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.346770212905986e-05,
+      "loss": 1.4489,
+      "step": 50740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.346441146467472e-05,
+      "loss": 1.6623,
+      "step": 50750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.346112080028958e-05,
+      "loss": 1.5784,
+      "step": 50760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3457830135904436e-05,
+      "loss": 1.6107,
+      "step": 50770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3454539471519304e-05,
+      "loss": 1.5547,
+      "step": 50780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.345124880713416e-05,
+      "loss": 1.6665,
+      "step": 50790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.344795814274903e-05,
+      "loss": 1.5086,
+      "step": 50800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.344466747836388e-05,
+      "loss": 1.4667,
+      "step": 50810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.344137681397874e-05,
+      "loss": 1.5705,
+      "step": 50820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3438086149593604e-05,
+      "loss": 1.3123,
+      "step": 50830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3434795485208465e-05,
+      "loss": 1.4931,
+      "step": 50840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3431504820823326e-05,
+      "loss": 1.3058,
+      "step": 50850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342821415643819e-05,
+      "loss": 1.4253,
+      "step": 50860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342492349205304e-05,
+      "loss": 1.4196,
+      "step": 50870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342163282766791e-05,
+      "loss": 1.329,
+      "step": 50880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3418342163282765e-05,
+      "loss": 1.0937,
+      "step": 50890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341505149889763e-05,
+      "loss": 1.3213,
+      "step": 50900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341176083451249e-05,
+      "loss": 1.373,
+      "step": 50910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340847017012735e-05,
+      "loss": 1.29,
+      "step": 50920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340517950574221e-05,
+      "loss": 1.2805,
+      "step": 50930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340188884135707e-05,
+      "loss": 1.2967,
+      "step": 50940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339859817697193e-05,
+      "loss": 1.3405,
+      "step": 50950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3395307512586793e-05,
+      "loss": 1.3,
+      "step": 50960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3392016848201655e-05,
+      "loss": 1.2753,
+      "step": 50970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3388726183816516e-05,
+      "loss": 1.3001,
+      "step": 50980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338543551943138e-05,
+      "loss": 1.236,
+      "step": 50990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338214485504624e-05,
+      "loss": 1.2615,
+      "step": 51000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337885419066109e-05,
+      "loss": 1.2648,
+      "step": 51010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3375563526275954e-05,
+      "loss": 1.2124,
+      "step": 51020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3372272861890816e-05,
+      "loss": 1.3642,
+      "step": 51030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336898219750568e-05,
+      "loss": 1.4116,
+      "step": 51040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336569153312054e-05,
+      "loss": 1.2194,
+      "step": 51050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.33624008687354e-05,
+      "loss": 1.3042,
+      "step": 51060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.335911020435026e-05,
+      "loss": 1.2859,
+      "step": 51070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.335581953996512e-05,
+      "loss": 1.246,
+      "step": 51080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.335252887557998e-05,
+      "loss": 1.205,
+      "step": 51090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3349238211194844e-05,
+      "loss": 1.1546,
+      "step": 51100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3345947546809706e-05,
+      "loss": 1.2963,
+      "step": 51110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.334265688242456e-05,
+      "loss": 1.3592,
+      "step": 51120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.333936621803943e-05,
+      "loss": 1.1648,
+      "step": 51130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.333607555365428e-05,
+      "loss": 1.0961,
+      "step": 51140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.333278488926915e-05,
+      "loss": 1.4559,
+      "step": 51150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3329494224884005e-05,
+      "loss": 1.7012,
+      "step": 51160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3326203560498867e-05,
+      "loss": 1.2067,
+      "step": 51170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.332291289611373e-05,
+      "loss": 1.3541,
+      "step": 51180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.331962223172859e-05,
+      "loss": 1.3928,
+      "step": 51190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.331633156734345e-05,
+      "loss": 1.2739,
+      "step": 51200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.331304090295831e-05,
+      "loss": 1.0739,
+      "step": 51210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3309750238573166e-05,
+      "loss": 1.2303,
+      "step": 51220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3306459574188034e-05,
+      "loss": 1.2472,
+      "step": 51230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330316890980289e-05,
+      "loss": 1.2787,
+      "step": 51240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.329987824541776e-05,
+      "loss": 1.369,
+      "step": 51250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.329658758103261e-05,
+      "loss": 1.3477,
+      "step": 51260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.329329691664747e-05,
+      "loss": 1.4011,
+      "step": 51270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3290006252262334e-05,
+      "loss": 1.3271,
+      "step": 51280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3286715587877195e-05,
+      "loss": 1.2952,
+      "step": 51290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3283424923492056e-05,
+      "loss": 1.3404,
+      "step": 51300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.328013425910692e-05,
+      "loss": 1.2977,
+      "step": 51310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327684359472177e-05,
+      "loss": 1.2824,
+      "step": 51320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327355293033664e-05,
+      "loss": 1.2774,
+      "step": 51330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3270262265951495e-05,
+      "loss": 1.2136,
+      "step": 51340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326697160156636e-05,
+      "loss": 1.2491,
+      "step": 51350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326368093718122e-05,
+      "loss": 1.2724,
+      "step": 51360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326039027279608e-05,
+      "loss": 1.1645,
+      "step": 51370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.325709960841094e-05,
+      "loss": 1.1245,
+      "step": 51380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32538089440258e-05,
+      "loss": 1.3446,
+      "step": 51390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.325051827964066e-05,
+      "loss": 1.377,
+      "step": 51400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3247227615255523e-05,
+      "loss": 1.1766,
+      "step": 51410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324393695087038e-05,
+      "loss": 1.2725,
+      "step": 51420
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3240646286485246e-05,
+      "loss": 1.2476,
+      "step": 51430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32373556221001e-05,
+      "loss": 1.205,
+      "step": 51440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323406495771497e-05,
+      "loss": 1.3195,
+      "step": 51450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323077429332982e-05,
+      "loss": 1.3464,
+      "step": 51460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3227483628944684e-05,
+      "loss": 1.3568,
+      "step": 51470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3224192964559546e-05,
+      "loss": 1.2486,
+      "step": 51480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.322090230017441e-05,
+      "loss": 1.284,
+      "step": 51490
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.321761163578927e-05,
+      "loss": 1.2707,
+      "step": 51500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.321432097140413e-05,
+      "loss": 1.2764,
+      "step": 51510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3211030307018984e-05,
+      "loss": 1.2763,
+      "step": 51520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320773964263385e-05,
+      "loss": 1.2629,
+      "step": 51530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3204448978248706e-05,
+      "loss": 1.2955,
+      "step": 51540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3201158313863574e-05,
+      "loss": 1.3228,
+      "step": 51550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319786764947843e-05,
+      "loss": 1.3,
+      "step": 51560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319457698509329e-05,
+      "loss": 1.1731,
+      "step": 51570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.319128632070815e-05,
+      "loss": 1.1885,
+      "step": 51580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.318799565632301e-05,
+      "loss": 1.3048,
+      "step": 51590
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3184704991937874e-05,
+      "loss": 1.2618,
+      "step": 51600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3181414327552735e-05,
+      "loss": 1.1569,
+      "step": 51610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.317812366316759e-05,
+      "loss": 1.3089,
+      "step": 51620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.317483299878246e-05,
+      "loss": 1.3175,
+      "step": 51630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.317154233439731e-05,
+      "loss": 1.2934,
+      "step": 51640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.316825167001218e-05,
+      "loss": 1.2375,
+      "step": 51650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3164961005627035e-05,
+      "loss": 1.2536,
+      "step": 51660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3161670341241896e-05,
+      "loss": 1.2564,
+      "step": 51670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.315837967685676e-05,
+      "loss": 1.2062,
+      "step": 51680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.315508901247162e-05,
+      "loss": 1.4254,
+      "step": 51690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.315179834808648e-05,
+      "loss": 1.3202,
+      "step": 51700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.314850768370134e-05,
+      "loss": 1.2945,
+      "step": 51710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.31452170193162e-05,
+      "loss": 1.3143,
+      "step": 51720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3141926354931064e-05,
+      "loss": 1.2403,
+      "step": 51730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3138635690545925e-05,
+      "loss": 1.218,
+      "step": 51740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3135345026160786e-05,
+      "loss": 1.2005,
+      "step": 51750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.313205436177564e-05,
+      "loss": 1.3886,
+      "step": 51760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.31287636973905e-05,
+      "loss": 1.3287,
+      "step": 51770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.312547303300536e-05,
+      "loss": 1.2999,
+      "step": 51780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3122182368620224e-05,
+      "loss": 1.2846,
+      "step": 51790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3118891704235086e-05,
+      "loss": 1.1392,
+      "step": 51800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.311560103984995e-05,
+      "loss": 1.2132,
+      "step": 51810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.311231037546481e-05,
+      "loss": 1.1448,
+      "step": 51820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.310901971107967e-05,
+      "loss": 1.2448,
+      "step": 51830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.310572904669453e-05,
+      "loss": 1.3312,
+      "step": 51840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.310243838230939e-05,
+      "loss": 1.2161,
+      "step": 51850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.309914771792425e-05,
+      "loss": 1.2997,
+      "step": 51860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.309585705353911e-05,
+      "loss": 1.3298,
+      "step": 51870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3092566389153976e-05,
+      "loss": 1.2385,
+      "step": 51880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.308927572476883e-05,
+      "loss": 1.341,
+      "step": 51890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.30859850603837e-05,
+      "loss": 1.3009,
+      "step": 51900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.308269439599855e-05,
+      "loss": 1.1952,
+      "step": 51910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3079403731613414e-05,
+      "loss": 1.4443,
+      "step": 51920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3076113067228275e-05,
+      "loss": 1.3205,
+      "step": 51930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.307282240284314e-05,
+      "loss": 1.4428,
+      "step": 51940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3069531738458e-05,
+      "loss": 1.5452,
+      "step": 51950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.306624107407286e-05,
+      "loss": 1.2475,
+      "step": 51960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3062950409687714e-05,
+      "loss": 1.1374,
+      "step": 51970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.305965974530258e-05,
+      "loss": 1.174,
+      "step": 51980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3056369080917436e-05,
+      "loss": 1.3637,
+      "step": 51990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3053078416532304e-05,
+      "loss": 1.3075,
+      "step": 52000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.304978775214716e-05,
+      "loss": 1.3324,
+      "step": 52010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.304649708776202e-05,
+      "loss": 1.2104,
+      "step": 52020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.304320642337688e-05,
+      "loss": 1.33,
+      "step": 52030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.303991575899174e-05,
+      "loss": 1.2479,
+      "step": 52040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3036625094606604e-05,
+      "loss": 1.2627,
+      "step": 52050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3033334430221465e-05,
+      "loss": 1.183,
+      "step": 52060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.303004376583632e-05,
+      "loss": 1.2613,
+      "step": 52070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.302675310145119e-05,
+      "loss": 1.234,
+      "step": 52080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.302346243706604e-05,
+      "loss": 1.2673,
+      "step": 52090
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.302017177268091e-05,
+      "loss": 1.2298,
+      "step": 52100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3016881108295765e-05,
+      "loss": 1.3523,
+      "step": 52110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3013590443910626e-05,
+      "loss": 1.2782,
+      "step": 52120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.301029977952549e-05,
+      "loss": 1.2766,
+      "step": 52130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.300700911514035e-05,
+      "loss": 1.4164,
+      "step": 52140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.300371845075521e-05,
+      "loss": 1.4143,
+      "step": 52150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.300042778637007e-05,
+      "loss": 1.339,
+      "step": 52160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2997137121984926e-05,
+      "loss": 1.3206,
+      "step": 52170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2993846457599794e-05,
+      "loss": 1.2667,
+      "step": 52180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.299055579321465e-05,
+      "loss": 1.1465,
+      "step": 52190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2987265128829516e-05,
+      "loss": 1.2559,
+      "step": 52200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.298397446444437e-05,
+      "loss": 1.2374,
+      "step": 52210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.298068380005923e-05,
+      "loss": 1.2754,
+      "step": 52220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.297739313567409e-05,
+      "loss": 1.3527,
+      "step": 52230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2974102471288954e-05,
+      "loss": 1.3185,
+      "step": 52240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2970811806903816e-05,
+      "loss": 1.3534,
+      "step": 52250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.296752114251868e-05,
+      "loss": 1.5255,
+      "step": 52260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.296423047813353e-05,
+      "loss": 1.1215,
+      "step": 52270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.29609398137484e-05,
+      "loss": 1.5448,
+      "step": 52280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2957649149363254e-05,
+      "loss": 1.2953,
+      "step": 52290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.295435848497812e-05,
+      "loss": 1.2791,
+      "step": 52300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2951067820592976e-05,
+      "loss": 1.4698,
+      "step": 52310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.294777715620784e-05,
+      "loss": 1.1954,
+      "step": 52320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.29444864918227e-05,
+      "loss": 1.1453,
+      "step": 52330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.294119582743756e-05,
+      "loss": 1.3645,
+      "step": 52340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.293790516305242e-05,
+      "loss": 1.2514,
+      "step": 52350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.293461449866728e-05,
+      "loss": 1.0486,
+      "step": 52360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.293132383428214e-05,
+      "loss": 1.3759,
+      "step": 52370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2928033169897005e-05,
+      "loss": 1.6144,
+      "step": 52380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.292474250551186e-05,
+      "loss": 1.4525,
+      "step": 52390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.292145184112673e-05,
+      "loss": 1.3311,
+      "step": 52400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.291816117674158e-05,
+      "loss": 1.2962,
+      "step": 52410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2914870512356444e-05,
+      "loss": 1.4879,
+      "step": 52420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2911579847971305e-05,
+      "loss": 1.3095,
+      "step": 52430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2908289183586166e-05,
+      "loss": 1.3089,
+      "step": 52440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.290499851920103e-05,
+      "loss": 1.1688,
+      "step": 52450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.290170785481589e-05,
+      "loss": 1.3544,
+      "step": 52460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.289841719043075e-05,
+      "loss": 1.2842,
+      "step": 52470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.289512652604561e-05,
+      "loss": 1.4376,
+      "step": 52480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.289183586166047e-05,
+      "loss": 1.2333,
+      "step": 52490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2888545197275334e-05,
+      "loss": 1.3382,
+      "step": 52500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2885254532890195e-05,
+      "loss": 1.4791,
+      "step": 52510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.288196386850505e-05,
+      "loss": 1.2709,
+      "step": 52520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.287867320411991e-05,
+      "loss": 1.3848,
+      "step": 52530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.287538253973477e-05,
+      "loss": 1.1821,
+      "step": 52540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.287209187534963e-05,
+      "loss": 1.1737,
+      "step": 52550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2868801210964495e-05,
+      "loss": 1.2825,
+      "step": 52560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2865510546579356e-05,
+      "loss": 1.1657,
+      "step": 52570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.286221988219422e-05,
+      "loss": 1.1089,
+      "step": 52580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.285892921780908e-05,
+      "loss": 1.1357,
+      "step": 52590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.285563855342394e-05,
+      "loss": 1.2294,
+      "step": 52600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.28523478890388e-05,
+      "loss": 1.3656,
+      "step": 52610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2849057224653655e-05,
+      "loss": 1.428,
+      "step": 52620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2845766560268524e-05,
+      "loss": 1.2211,
+      "step": 52630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.284247589588338e-05,
+      "loss": 1.3452,
+      "step": 52640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2839185231498246e-05,
+      "loss": 1.2846,
+      "step": 52650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.28358945671131e-05,
+      "loss": 1.2379,
+      "step": 52660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.283260390272796e-05,
+      "loss": 1.2349,
+      "step": 52670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.282931323834282e-05,
+      "loss": 1.3731,
+      "step": 52680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2826022573957684e-05,
+      "loss": 1.2044,
+      "step": 52690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2822731909572546e-05,
+      "loss": 1.4854,
+      "step": 52700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.281944124518741e-05,
+      "loss": 1.2669,
+      "step": 52710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.281615058080226e-05,
+      "loss": 1.4524,
+      "step": 52720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.281285991641713e-05,
+      "loss": 1.4277,
+      "step": 52730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2809569252031984e-05,
+      "loss": 1.1485,
+      "step": 52740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.280627858764685e-05,
+      "loss": 1.3725,
+      "step": 52750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2802987923261706e-05,
+      "loss": 1.2824,
+      "step": 52760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.279969725887657e-05,
+      "loss": 1.3389,
+      "step": 52770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.279640659449143e-05,
+      "loss": 1.2311,
+      "step": 52780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.279311593010629e-05,
+      "loss": 1.1481,
+      "step": 52790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.278982526572115e-05,
+      "loss": 1.1724,
+      "step": 52800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.278653460133601e-05,
+      "loss": 1.2317,
+      "step": 52810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.278324393695087e-05,
+      "loss": 1.2698,
+      "step": 52820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2779953272565735e-05,
+      "loss": 1.312,
+      "step": 52830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.277666260818059e-05,
+      "loss": 1.3731,
+      "step": 52840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.277337194379546e-05,
+      "loss": 1.3514,
+      "step": 52850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.277008127941031e-05,
+      "loss": 1.3614,
+      "step": 52860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2766790615025174e-05,
+      "loss": 1.2609,
+      "step": 52870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2763499950640035e-05,
+      "loss": 1.2023,
+      "step": 52880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2760209286254896e-05,
+      "loss": 1.3049,
+      "step": 52890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.275691862186976e-05,
+      "loss": 1.3744,
+      "step": 52900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.275362795748462e-05,
+      "loss": 1.2548,
+      "step": 52910
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.275033729309947e-05,
+      "loss": 1.2961,
+      "step": 52920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.274704662871434e-05,
+      "loss": 1.3698,
+      "step": 52930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2743755964329196e-05,
+      "loss": 1.2826,
+      "step": 52940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2740465299944064e-05,
+      "loss": 1.3044,
+      "step": 52950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.273717463555892e-05,
+      "loss": 1.2365,
+      "step": 52960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.273388397117378e-05,
+      "loss": 1.164,
+      "step": 52970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.273059330678864e-05,
+      "loss": 1.2819,
+      "step": 52980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.27273026424035e-05,
+      "loss": 1.2262,
+      "step": 52990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.272401197801836e-05,
+      "loss": 1.4359,
+      "step": 53000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2720721313633225e-05,
+      "loss": 1.1641,
+      "step": 53010
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.271743064924808e-05,
+      "loss": 1.0778,
+      "step": 53020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.271413998486295e-05,
+      "loss": 1.2704,
+      "step": 53030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.27108493204778e-05,
+      "loss": 1.2024,
+      "step": 53040
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.270755865609267e-05,
+      "loss": 1.2036,
+      "step": 53050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2704267991707524e-05,
+      "loss": 1.1729,
+      "step": 53060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.270097732732239e-05,
+      "loss": 1.4311,
+      "step": 53070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.269768666293725e-05,
+      "loss": 1.2262,
+      "step": 53080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.269439599855211e-05,
+      "loss": 1.228,
+      "step": 53090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.269110533416697e-05,
+      "loss": 1.3812,
+      "step": 53100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.268781466978183e-05,
+      "loss": 1.4032,
+      "step": 53110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.268452400539669e-05,
+      "loss": 1.3216,
+      "step": 53120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.268123334101155e-05,
+      "loss": 1.3557,
+      "step": 53130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.267794267662641e-05,
+      "loss": 1.326,
+      "step": 53140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2674652012241276e-05,
+      "loss": 1.1653,
+      "step": 53150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.267136134785613e-05,
+      "loss": 1.4877,
+      "step": 53160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2668070683471e-05,
+      "loss": 1.1997,
+      "step": 53170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.266478001908585e-05,
+      "loss": 1.0658,
+      "step": 53180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2661489354700714e-05,
+      "loss": 1.3498,
+      "step": 53190
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2658198690315575e-05,
+      "loss": 1.2413,
+      "step": 53200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2654908025930436e-05,
+      "loss": 1.3428,
+      "step": 53210
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.26516173615453e-05,
+      "loss": 1.3798,
+      "step": 53220
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.264832669716016e-05,
+      "loss": 1.2975,
+      "step": 53230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.264503603277502e-05,
+      "loss": 1.4648,
+      "step": 53240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.264174536838988e-05,
+      "loss": 1.2696,
+      "step": 53250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.263845470400474e-05,
+      "loss": 1.2647,
+      "step": 53260
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2635164039619604e-05,
+      "loss": 1.1146,
+      "step": 53270
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2631873375234465e-05,
+      "loss": 1.3492,
+      "step": 53280
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.262858271084932e-05,
+      "loss": 1.142,
+      "step": 53290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.262529204646418e-05,
+      "loss": 1.2609,
+      "step": 53300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.262200138207904e-05,
+      "loss": 1.2634,
+      "step": 53310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2618710717693904e-05,
+      "loss": 1.2676,
+      "step": 53320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2615420053308765e-05,
+      "loss": 1.2696,
+      "step": 53330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2612129388923626e-05,
+      "loss": 1.4422,
+      "step": 53340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.260883872453849e-05,
+      "loss": 1.1724,
+      "step": 53350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.260554806015335e-05,
+      "loss": 1.1952,
+      "step": 53360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.260225739576821e-05,
+      "loss": 1.173,
+      "step": 53370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.259896673138307e-05,
+      "loss": 1.373,
+      "step": 53380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2595676066997926e-05,
+      "loss": 1.2456,
+      "step": 53390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2592385402612794e-05,
+      "loss": 1.2484,
+      "step": 53400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.258909473822765e-05,
+      "loss": 1.362,
+      "step": 53410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2585804073842516e-05,
+      "loss": 1.2247,
+      "step": 53420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.258251340945737e-05,
+      "loss": 1.3834,
+      "step": 53430
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.257922274507223e-05,
+      "loss": 1.2854,
+      "step": 53440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.257593208068709e-05,
+      "loss": 1.3015,
+      "step": 53450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2572641416301954e-05,
+      "loss": 1.229,
+      "step": 53460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2569350751916816e-05,
+      "loss": 1.3551,
+      "step": 53470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.256606008753168e-05,
+      "loss": 1.3637,
+      "step": 53480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.256276942314653e-05,
+      "loss": 1.2533,
+      "step": 53490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.25594787587614e-05,
+      "loss": 1.5269,
+      "step": 53500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2556188094376254e-05,
+      "loss": 1.128,
+      "step": 53510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.255289742999112e-05,
+      "loss": 1.3237,
+      "step": 53520
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2549606765605977e-05,
+      "loss": 1.2201,
+      "step": 53530
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.254631610122084e-05,
+      "loss": 1.208,
+      "step": 53540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.25430254368357e-05,
+      "loss": 1.116,
+      "step": 53550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.253973477245056e-05,
+      "loss": 1.1978,
+      "step": 53560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.253644410806542e-05,
+      "loss": 1.2987,
+      "step": 53570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.253315344368028e-05,
+      "loss": 1.3826,
+      "step": 53580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.252986277929514e-05,
+      "loss": 1.3271,
+      "step": 53590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2526572114910005e-05,
+      "loss": 1.3105,
+      "step": 53600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.252328145052486e-05,
+      "loss": 1.4056,
+      "step": 53610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.251999078613973e-05,
+      "loss": 1.2796,
+      "step": 53620
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.251670012175458e-05,
+      "loss": 1.1874,
+      "step": 53630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2513409457369444e-05,
+      "loss": 1.3965,
+      "step": 53640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2510118792984305e-05,
+      "loss": 1.3419,
+      "step": 53650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2506828128599166e-05,
+      "loss": 1.2684,
+      "step": 53660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.250353746421403e-05,
+      "loss": 1.3393,
+      "step": 53670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.250024679982889e-05,
+      "loss": 1.3191,
+      "step": 53680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.249695613544374e-05,
+      "loss": 1.279,
+      "step": 53690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.249366547105861e-05,
+      "loss": 1.2565,
+      "step": 53700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2490374806673466e-05,
+      "loss": 1.3308,
+      "step": 53710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2487084142288334e-05,
+      "loss": 1.1937,
+      "step": 53720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.248379347790319e-05,
+      "loss": 1.5966,
+      "step": 53730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.248050281351805e-05,
+      "loss": 1.3148,
+      "step": 53740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.247721214913291e-05,
+      "loss": 1.3561,
+      "step": 53750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.247392148474777e-05,
+      "loss": 1.3622,
+      "step": 53760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2470630820362633e-05,
+      "loss": 1.1131,
+      "step": 53770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2467340155977495e-05,
+      "loss": 1.4038,
+      "step": 53780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.246404949159235e-05,
+      "loss": 1.2089,
+      "step": 53790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.246075882720722e-05,
+      "loss": 1.2154,
+      "step": 53800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.245746816282207e-05,
+      "loss": 1.231,
+      "step": 53810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.245417749843694e-05,
+      "loss": 1.224,
+      "step": 53820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2450886834051794e-05,
+      "loss": 1.224,
+      "step": 53830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2447596169666656e-05,
+      "loss": 1.2765,
+      "step": 53840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.244430550528152e-05,
+      "loss": 1.276,
+      "step": 53850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.244101484089638e-05,
+      "loss": 1.4468,
+      "step": 53860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.243772417651124e-05,
+      "loss": 1.3152,
+      "step": 53870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.24344335121261e-05,
+      "loss": 1.3528,
+      "step": 53880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2431142847740955e-05,
+      "loss": 1.1999,
+      "step": 53890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.242785218335582e-05,
+      "loss": 1.3191,
+      "step": 53900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.242456151897068e-05,
+      "loss": 1.1704,
+      "step": 53910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2421270854585546e-05,
+      "loss": 1.2051,
+      "step": 53920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.24179801902004e-05,
+      "loss": 1.1952,
+      "step": 53930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.241468952581526e-05,
+      "loss": 1.2563,
+      "step": 53940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.241139886143012e-05,
+      "loss": 1.0719,
+      "step": 53950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2408108197044984e-05,
+      "loss": 1.2263,
+      "step": 53960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2404817532659845e-05,
+      "loss": 1.2475,
+      "step": 53970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2401526868274707e-05,
+      "loss": 1.3719,
+      "step": 53980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.239823620388957e-05,
+      "loss": 1.236,
+      "step": 53990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.239494553950443e-05,
+      "loss": 1.2929,
+      "step": 54000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.239165487511929e-05,
+      "loss": 1.3835,
+      "step": 54010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.238836421073415e-05,
+      "loss": 1.2491,
+      "step": 54020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.238507354634901e-05,
+      "loss": 1.5226,
+      "step": 54030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.238178288196387e-05,
+      "loss": 1.1818,
+      "step": 54040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.237849221757873e-05,
+      "loss": 1.635,
+      "step": 54050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.237520155319359e-05,
+      "loss": 1.1333,
+      "step": 54060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.237191088880845e-05,
+      "loss": 1.271,
+      "step": 54070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.236862022442331e-05,
+      "loss": 1.2354,
+      "step": 54080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2365329560038174e-05,
+      "loss": 1.211,
+      "step": 54090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2362038895653035e-05,
+      "loss": 1.2821,
+      "step": 54100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2358748231267896e-05,
+      "loss": 1.2066,
+      "step": 54110
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.235545756688276e-05,
+      "loss": 1.3837,
+      "step": 54120
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.235216690249762e-05,
+      "loss": 1.2525,
+      "step": 54130
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.234887623811247e-05,
+      "loss": 1.2016,
+      "step": 54140
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.234558557372734e-05,
+      "loss": 1.283,
+      "step": 54150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2342294909342196e-05,
+      "loss": 1.1374,
+      "step": 54160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2339004244957064e-05,
+      "loss": 1.2699,
+      "step": 54170
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.233571358057192e-05,
+      "loss": 1.2351,
+      "step": 54180
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.233242291618678e-05,
+      "loss": 1.1895,
+      "step": 54190
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.232913225180164e-05,
+      "loss": 1.2806,
+      "step": 54200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.23258415874165e-05,
+      "loss": 1.405,
+      "step": 54210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.232255092303136e-05,
+      "loss": 1.3942,
+      "step": 54220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2319260258646225e-05,
+      "loss": 1.7154,
+      "step": 54230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.231596959426108e-05,
+      "loss": 1.35,
+      "step": 54240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.231267892987595e-05,
+      "loss": 1.2663,
+      "step": 54250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.23093882654908e-05,
+      "loss": 1.1971,
+      "step": 54260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.230609760110567e-05,
+      "loss": 1.2617,
+      "step": 54270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2302806936720524e-05,
+      "loss": 1.3101,
+      "step": 54280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2299516272335385e-05,
+      "loss": 1.1846,
+      "step": 54290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.229622560795025e-05,
+      "loss": 1.3116,
+      "step": 54300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.229293494356511e-05,
+      "loss": 1.2451,
+      "step": 54310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.228964427917997e-05,
+      "loss": 1.2419,
+      "step": 54320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.228635361479483e-05,
+      "loss": 1.3646,
+      "step": 54330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2283062950409685e-05,
+      "loss": 1.2169,
+      "step": 54340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.227977228602455e-05,
+      "loss": 1.2391,
+      "step": 54350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.227648162163941e-05,
+      "loss": 1.3125,
+      "step": 54360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2273190957254276e-05,
+      "loss": 1.195,
+      "step": 54370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.226990029286913e-05,
+      "loss": 1.1902,
+      "step": 54380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.226660962848399e-05,
+      "loss": 1.2543,
+      "step": 54390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.226331896409885e-05,
+      "loss": 1.2332,
+      "step": 54400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2260028299713714e-05,
+      "loss": 1.1916,
+      "step": 54410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2256737635328575e-05,
+      "loss": 1.4932,
+      "step": 54420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2253446970943436e-05,
+      "loss": 1.064,
+      "step": 54430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.225015630655829e-05,
+      "loss": 1.2496,
+      "step": 54440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.224686564217316e-05,
+      "loss": 1.252,
+      "step": 54450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2243574977788013e-05,
+      "loss": 1.1087,
+      "step": 54460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.224028431340288e-05,
+      "loss": 1.3352,
+      "step": 54470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2236993649017736e-05,
+      "loss": 1.2676,
+      "step": 54480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.22337029846326e-05,
+      "loss": 1.2854,
+      "step": 54490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.223041232024746e-05,
+      "loss": 1.2441,
+      "step": 54500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.222712165586232e-05,
+      "loss": 1.354,
+      "step": 54510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.222383099147718e-05,
+      "loss": 1.1988,
+      "step": 54520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.222054032709204e-05,
+      "loss": 1.2121,
+      "step": 54530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.22172496627069e-05,
+      "loss": 1.178,
+      "step": 54540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2213958998321765e-05,
+      "loss": 1.2545,
+      "step": 54550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.221066833393662e-05,
+      "loss": 1.3657,
+      "step": 54560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.220737766955149e-05,
+      "loss": 1.1805,
+      "step": 54570
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.220408700516634e-05,
+      "loss": 1.3042,
+      "step": 54580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.22007963407812e-05,
+      "loss": 1.5098,
+      "step": 54590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2197505676396064e-05,
+      "loss": 1.3914,
+      "step": 54600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2194215012010926e-05,
+      "loss": 1.2986,
+      "step": 54610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.219092434762579e-05,
+      "loss": 1.4309,
+      "step": 54620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.218763368324065e-05,
+      "loss": 1.3597,
+      "step": 54630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.218434301885551e-05,
+      "loss": 1.3996,
+      "step": 54640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.218105235447037e-05,
+      "loss": 1.1802,
+      "step": 54650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2177761690085225e-05,
+      "loss": 1.4589,
+      "step": 54660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.217447102570009e-05,
+      "loss": 1.3134,
+      "step": 54670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.217118036131495e-05,
+      "loss": 1.2917,
+      "step": 54680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.216788969692981e-05,
+      "loss": 1.4032,
+      "step": 54690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.216459903254467e-05,
+      "loss": 1.289,
+      "step": 54700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.216130836815953e-05,
+      "loss": 1.3995,
+      "step": 54710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.215801770377439e-05,
+      "loss": 1.3393,
+      "step": 54720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2154727039389254e-05,
+      "loss": 1.3451,
+      "step": 54730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2151436375004115e-05,
+      "loss": 1.4977,
+      "step": 54740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.214814571061898e-05,
+      "loss": 1.373,
+      "step": 54750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.214485504623384e-05,
+      "loss": 1.4414,
+      "step": 54760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.21415643818487e-05,
+      "loss": 1.1971,
+      "step": 54770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.213827371746356e-05,
+      "loss": 1.3611,
+      "step": 54780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2134983053078415e-05,
+      "loss": 1.2831,
+      "step": 54790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.213169238869328e-05,
+      "loss": 1.171,
+      "step": 54800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.212840172430814e-05,
+      "loss": 1.161,
+      "step": 54810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2125111059923e-05,
+      "loss": 1.1311,
+      "step": 54820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.212182039553786e-05,
+      "loss": 1.3235,
+      "step": 54830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.211852973115272e-05,
+      "loss": 1.2369,
+      "step": 54840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.211523906676758e-05,
+      "loss": 1.1729,
+      "step": 54850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2111948402382444e-05,
+      "loss": 1.2502,
+      "step": 54860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2108657737997305e-05,
+      "loss": 1.203,
+      "step": 54870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2105367073612166e-05,
+      "loss": 1.2984,
+      "step": 54880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.210207640922702e-05,
+      "loss": 1.3895,
+      "step": 54890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.209878574484189e-05,
+      "loss": 1.3083,
+      "step": 54900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2095495080456743e-05,
+      "loss": 1.3212,
+      "step": 54910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.209220441607161e-05,
+      "loss": 1.3204,
+      "step": 54920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2088913751686466e-05,
+      "loss": 1.2246,
+      "step": 54930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.208562308730133e-05,
+      "loss": 1.3287,
+      "step": 54940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.208233242291619e-05,
+      "loss": 1.3583,
+      "step": 54950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.207904175853105e-05,
+      "loss": 1.2729,
+      "step": 54960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.207575109414591e-05,
+      "loss": 1.4654,
+      "step": 54970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.207246042976077e-05,
+      "loss": 1.2195,
+      "step": 54980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206916976537563e-05,
+      "loss": 1.3614,
+      "step": 54990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2065879100990495e-05,
+      "loss": 1.3854,
+      "step": 55000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206258843660535e-05,
+      "loss": 1.3469,
+      "step": 55010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.205929777222022e-05,
+      "loss": 1.2315,
+      "step": 55020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.205600710783507e-05,
+      "loss": 1.3463,
+      "step": 55030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.205271644344993e-05,
+      "loss": 1.3604,
+      "step": 55040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2049425779064794e-05,
+      "loss": 1.2875,
+      "step": 55050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2046135114679656e-05,
+      "loss": 1.2715,
+      "step": 55060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.204284445029452e-05,
+      "loss": 1.2745,
+      "step": 55070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.203955378590938e-05,
+      "loss": 1.2585,
+      "step": 55080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.203626312152423e-05,
+      "loss": 1.3912,
+      "step": 55090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.20329724571391e-05,
+      "loss": 1.1514,
+      "step": 55100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2029681792753955e-05,
+      "loss": 1.4089,
+      "step": 55110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.202639112836882e-05,
+      "loss": 1.2647,
+      "step": 55120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.202310046398368e-05,
+      "loss": 1.3842,
+      "step": 55130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.201980979959854e-05,
+      "loss": 1.2064,
+      "step": 55140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.20165191352134e-05,
+      "loss": 1.1734,
+      "step": 55150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.201322847082826e-05,
+      "loss": 1.2783,
+      "step": 55160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.200993780644312e-05,
+      "loss": 1.338,
+      "step": 55170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2006647142057984e-05,
+      "loss": 1.4574,
+      "step": 55180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.200335647767284e-05,
+      "loss": 1.2392,
+      "step": 55190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.200006581328771e-05,
+      "loss": 1.2798,
+      "step": 55200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.199677514890256e-05,
+      "loss": 1.3318,
+      "step": 55210
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.199348448451743e-05,
+      "loss": 1.3923,
+      "step": 55220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1990193820132284e-05,
+      "loss": 1.3797,
+      "step": 55230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1986903155747145e-05,
+      "loss": 1.275,
+      "step": 55240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1983612491362006e-05,
+      "loss": 1.4464,
+      "step": 55250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.198032182697687e-05,
+      "loss": 1.1397,
+      "step": 55260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.197703116259173e-05,
+      "loss": 1.3378,
+      "step": 55270
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.197374049820659e-05,
+      "loss": 1.1894,
+      "step": 55280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1970449833821444e-05,
+      "loss": 1.2646,
+      "step": 55290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.196715916943631e-05,
+      "loss": 1.456,
+      "step": 55300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.196386850505117e-05,
+      "loss": 1.3242,
+      "step": 55310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1960577840666035e-05,
+      "loss": 1.313,
+      "step": 55320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.195728717628089e-05,
+      "loss": 1.2915,
+      "step": 55330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.195399651189575e-05,
+      "loss": 1.2073,
+      "step": 55340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.195070584751061e-05,
+      "loss": 1.1028,
+      "step": 55350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.194741518312547e-05,
+      "loss": 1.2693,
+      "step": 55360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1944124518740335e-05,
+      "loss": 1.3478,
+      "step": 55370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1940833854355196e-05,
+      "loss": 1.412,
+      "step": 55380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.193754318997006e-05,
+      "loss": 1.0935,
+      "step": 55390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.193425252558492e-05,
+      "loss": 1.2303,
+      "step": 55400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.193096186119977e-05,
+      "loss": 1.2751,
+      "step": 55410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.192767119681464e-05,
+      "loss": 1.2259,
+      "step": 55420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1924380532429495e-05,
+      "loss": 1.1895,
+      "step": 55430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.192108986804436e-05,
+      "loss": 1.2721,
+      "step": 55440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.191779920365922e-05,
+      "loss": 1.2469,
+      "step": 55450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.191450853927408e-05,
+      "loss": 1.3708,
+      "step": 55460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.191121787488894e-05,
+      "loss": 1.3894,
+      "step": 55470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.19079272105038e-05,
+      "loss": 1.2153,
+      "step": 55480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.190463654611866e-05,
+      "loss": 1.1498,
+      "step": 55490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1901345881733524e-05,
+      "loss": 1.3142,
+      "step": 55500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1898055217348386e-05,
+      "loss": 1.2262,
+      "step": 55510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.189476455296325e-05,
+      "loss": 1.2484,
+      "step": 55520
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.189147388857811e-05,
+      "loss": 1.2275,
+      "step": 55530
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.188818322419296e-05,
+      "loss": 1.3287,
+      "step": 55540
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.188489255980783e-05,
+      "loss": 1.3283,
+      "step": 55550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1881601895422685e-05,
+      "loss": 1.4597,
+      "step": 55560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.187831123103755e-05,
+      "loss": 1.3297,
+      "step": 55570
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.187502056665241e-05,
+      "loss": 1.3132,
+      "step": 55580
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.187172990226727e-05,
+      "loss": 1.29,
+      "step": 55590
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186843923788213e-05,
+      "loss": 1.2303,
+      "step": 55600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186514857349699e-05,
+      "loss": 1.2115,
+      "step": 55610
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186185790911185e-05,
+      "loss": 1.2366,
+      "step": 55620
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1858567244726714e-05,
+      "loss": 1.0892,
+      "step": 55630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.185527658034157e-05,
+      "loss": 1.2471,
+      "step": 55640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1851985915956437e-05,
+      "loss": 1.3174,
+      "step": 55650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.184869525157129e-05,
+      "loss": 1.2857,
+      "step": 55660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.184540458718616e-05,
+      "loss": 1.1193,
+      "step": 55670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1842113922801014e-05,
+      "loss": 1.2867,
+      "step": 55680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1838823258415875e-05,
+      "loss": 1.3221,
+      "step": 55690
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1835532594030736e-05,
+      "loss": 1.1689,
+      "step": 55700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.18322419296456e-05,
+      "loss": 1.2045,
+      "step": 55710
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.182895126526046e-05,
+      "loss": 1.2127,
+      "step": 55720
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.182566060087532e-05,
+      "loss": 1.3485,
+      "step": 55730
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1822369936490174e-05,
+      "loss": 1.2215,
+      "step": 55740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.181907927210504e-05,
+      "loss": 1.2712,
+      "step": 55750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.18157886077199e-05,
+      "loss": 1.2997,
+      "step": 55760
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1812497943334765e-05,
+      "loss": 1.3325,
+      "step": 55770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.180920727894962e-05,
+      "loss": 1.2879,
+      "step": 55780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.180591661456448e-05,
+      "loss": 1.109,
+      "step": 55790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.180262595017934e-05,
+      "loss": 1.2446,
+      "step": 55800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.17993352857942e-05,
+      "loss": 1.4047,
+      "step": 55810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1796044621409065e-05,
+      "loss": 1.3527,
+      "step": 55820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1792753957023926e-05,
+      "loss": 1.1277,
+      "step": 55830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.178946329263878e-05,
+      "loss": 1.1947,
+      "step": 55840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.178617262825365e-05,
+      "loss": 1.4143,
+      "step": 55850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.17828819638685e-05,
+      "loss": 1.1954,
+      "step": 55860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.177959129948337e-05,
+      "loss": 1.2507,
+      "step": 55870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1776300635098225e-05,
+      "loss": 1.449,
+      "step": 55880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1773009970713093e-05,
+      "loss": 1.3315,
+      "step": 55890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.176971930632795e-05,
+      "loss": 1.1679,
+      "step": 55900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.176642864194281e-05,
+      "loss": 1.283,
+      "step": 55910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.176313797755767e-05,
+      "loss": 1.2814,
+      "step": 55920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.175984731317253e-05,
+      "loss": 1.2069,
+      "step": 55930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.175655664878739e-05,
+      "loss": 1.3202,
+      "step": 55940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1753265984402254e-05,
+      "loss": 1.3851,
+      "step": 55950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174997532001711e-05,
+      "loss": 1.2057,
+      "step": 55960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174668465563198e-05,
+      "loss": 1.2679,
+      "step": 55970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174339399124683e-05,
+      "loss": 1.1699,
+      "step": 55980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.17401033268617e-05,
+      "loss": 1.3996,
+      "step": 55990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1736812662476554e-05,
+      "loss": 1.3494,
+      "step": 56000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1733521998091415e-05,
+      "loss": 1.3129,
+      "step": 56010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1730231333706276e-05,
+      "loss": 1.2343,
+      "step": 56020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.172694066932114e-05,
+      "loss": 1.2282,
+      "step": 56030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1723650004936e-05,
+      "loss": 1.1842,
+      "step": 56040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.172035934055086e-05,
+      "loss": 1.3191,
+      "step": 56050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1717068676165715e-05,
+      "loss": 1.235,
+      "step": 56060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.171377801178058e-05,
+      "loss": 1.2623,
+      "step": 56070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.171048734739544e-05,
+      "loss": 1.2293,
+      "step": 56080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1707196683010305e-05,
+      "loss": 1.4224,
+      "step": 56090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.170390601862516e-05,
+      "loss": 1.3495,
+      "step": 56100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.170061535424002e-05,
+      "loss": 1.1497,
+      "step": 56110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.169732468985488e-05,
+      "loss": 1.2633,
+      "step": 56120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1694034025469743e-05,
+      "loss": 1.4662,
+      "step": 56130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1690743361084605e-05,
+      "loss": 1.3425,
+      "step": 56140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1687452696699466e-05,
+      "loss": 1.3551,
+      "step": 56150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.168416203231433e-05,
+      "loss": 1.2343,
+      "step": 56160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.168087136792919e-05,
+      "loss": 1.3723,
+      "step": 56170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.167758070354404e-05,
+      "loss": 1.3087,
+      "step": 56180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.167429003915891e-05,
+      "loss": 1.2592,
+      "step": 56190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1670999374773766e-05,
+      "loss": 1.2496,
+      "step": 56200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.166770871038863e-05,
+      "loss": 1.3091,
+      "step": 56210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.166441804600349e-05,
+      "loss": 1.1404,
+      "step": 56220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.166112738161835e-05,
+      "loss": 1.2641,
+      "step": 56230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.165783671723321e-05,
+      "loss": 1.172,
+      "step": 56240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.165454605284807e-05,
+      "loss": 1.1626,
+      "step": 56250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.165125538846293e-05,
+      "loss": 1.2086,
+      "step": 56260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1647964724077794e-05,
+      "loss": 1.1508,
+      "step": 56270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1644674059692656e-05,
+      "loss": 1.1369,
+      "step": 56280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.164138339530752e-05,
+      "loss": 1.341,
+      "step": 56290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.163809273092238e-05,
+      "loss": 1.3199,
+      "step": 56300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.163480206653723e-05,
+      "loss": 1.2771,
+      "step": 56310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.16315114021521e-05,
+      "loss": 1.3468,
+      "step": 56320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1628220737766955e-05,
+      "loss": 1.4522,
+      "step": 56330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1624930073381817e-05,
+      "loss": 1.4464,
+      "step": 56340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.162163940899668e-05,
+      "loss": 1.3331,
+      "step": 56350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.161834874461154e-05,
+      "loss": 1.159,
+      "step": 56360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.16150580802264e-05,
+      "loss": 1.2577,
+      "step": 56370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.161176741584126e-05,
+      "loss": 1.2343,
+      "step": 56380
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.160847675145612e-05,
+      "loss": 1.2826,
+      "step": 56390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1605186087070984e-05,
+      "loss": 1.3952,
+      "step": 56400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.160189542268584e-05,
+      "loss": 1.3063,
+      "step": 56410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.159860475830071e-05,
+      "loss": 1.1905,
+      "step": 56420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.159531409391556e-05,
+      "loss": 1.2083,
+      "step": 56430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.159202342953043e-05,
+      "loss": 1.2568,
+      "step": 56440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1588732765145284e-05,
+      "loss": 1.2363,
+      "step": 56450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1585442100760145e-05,
+      "loss": 1.3804,
+      "step": 56460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1582151436375006e-05,
+      "loss": 1.3036,
+      "step": 56470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.157886077198987e-05,
+      "loss": 1.3581,
+      "step": 56480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.157557010760473e-05,
+      "loss": 1.2984,
+      "step": 56490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.157227944321959e-05,
+      "loss": 1.0651,
+      "step": 56500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1568988778834445e-05,
+      "loss": 1.3194,
+      "step": 56510
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.156569811444931e-05,
+      "loss": 1.1792,
+      "step": 56520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.156240745006417e-05,
+      "loss": 1.1988,
+      "step": 56530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1559116785679035e-05,
+      "loss": 1.1871,
+      "step": 56540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.155582612129389e-05,
+      "loss": 1.2395,
+      "step": 56550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.155253545690875e-05,
+      "loss": 1.3881,
+      "step": 56560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.154924479252361e-05,
+      "loss": 1.2517,
+      "step": 56570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1545954128138473e-05,
+      "loss": 1.1838,
+      "step": 56580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1542663463753335e-05,
+      "loss": 1.1485,
+      "step": 56590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1539372799368196e-05,
+      "loss": 1.2359,
+      "step": 56600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.153608213498305e-05,
+      "loss": 1.3607,
+      "step": 56610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.153279147059792e-05,
+      "loss": 1.2893,
+      "step": 56620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.152950080621277e-05,
+      "loss": 1.3263,
+      "step": 56630
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.152621014182764e-05,
+      "loss": 1.3735,
+      "step": 56640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1522919477442496e-05,
+      "loss": 1.1804,
+      "step": 56650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.151962881305736e-05,
+      "loss": 1.3686,
+      "step": 56660
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.151633814867222e-05,
+      "loss": 1.2319,
+      "step": 56670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.151304748428708e-05,
+      "loss": 1.3051,
+      "step": 56680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.150975681990194e-05,
+      "loss": 1.331,
+      "step": 56690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.15064661555168e-05,
+      "loss": 1.1843,
+      "step": 56700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1503175491131656e-05,
+      "loss": 1.2952,
+      "step": 56710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1499884826746524e-05,
+      "loss": 1.0247,
+      "step": 56720
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.149659416236138e-05,
+      "loss": 1.1882,
+      "step": 56730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.149330349797625e-05,
+      "loss": 1.3319,
+      "step": 56740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.14900128335911e-05,
+      "loss": 1.33,
+      "step": 56750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.148672216920596e-05,
+      "loss": 1.3497,
+      "step": 56760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1483431504820824e-05,
+      "loss": 1.3192,
+      "step": 56770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1480140840435685e-05,
+      "loss": 1.1673,
+      "step": 56780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1476850176050546e-05,
+      "loss": 1.2526,
+      "step": 56790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147355951166541e-05,
+      "loss": 1.2892,
+      "step": 56800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147026884728026e-05,
+      "loss": 1.2483,
+      "step": 56810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.146697818289513e-05,
+      "loss": 1.3469,
+      "step": 56820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1463687518509985e-05,
+      "loss": 1.2592,
+      "step": 56830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.146039685412485e-05,
+      "loss": 1.2189,
+      "step": 56840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.145710618973971e-05,
+      "loss": 1.2716,
+      "step": 56850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.145381552535457e-05,
+      "loss": 1.1952,
+      "step": 56860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.145052486096943e-05,
+      "loss": 1.318,
+      "step": 56870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.144723419658429e-05,
+      "loss": 1.3503,
+      "step": 56880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.144394353219915e-05,
+      "loss": 1.3974,
+      "step": 56890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1440652867814014e-05,
+      "loss": 1.2285,
+      "step": 56900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1437362203428875e-05,
+      "loss": 1.3846,
+      "step": 56910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1434071539043736e-05,
+      "loss": 1.2948,
+      "step": 56920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.14307808746586e-05,
+      "loss": 1.1035,
+      "step": 56930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.142749021027346e-05,
+      "loss": 1.5929,
+      "step": 56940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.142419954588831e-05,
+      "loss": 1.3552,
+      "step": 56950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1420908881503174e-05,
+      "loss": 1.2233,
+      "step": 56960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1417618217118036e-05,
+      "loss": 1.3078,
+      "step": 56970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.14143275527329e-05,
+      "loss": 1.3569,
+      "step": 56980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.141103688834776e-05,
+      "loss": 1.2413,
+      "step": 56990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.140774622396262e-05,
+      "loss": 1.1964,
+      "step": 57000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.140445555957748e-05,
+      "loss": 1.3674,
+      "step": 57010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.140116489519234e-05,
+      "loss": 1.1861,
+      "step": 57020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.13978742308072e-05,
+      "loss": 1.274,
+      "step": 57030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1394583566422065e-05,
+      "loss": 1.1387,
+      "step": 57040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1391292902036926e-05,
+      "loss": 1.2856,
+      "step": 57050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.138800223765178e-05,
+      "loss": 1.2465,
+      "step": 57060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.138471157326665e-05,
+      "loss": 1.3191,
+      "step": 57070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.13814209088815e-05,
+      "loss": 1.2774,
+      "step": 57080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137813024449637e-05,
+      "loss": 1.108,
+      "step": 57090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1374839580111225e-05,
+      "loss": 1.148,
+      "step": 57100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137154891572609e-05,
+      "loss": 1.219,
+      "step": 57110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136825825134095e-05,
+      "loss": 1.3588,
+      "step": 57120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136496758695581e-05,
+      "loss": 1.3203,
+      "step": 57130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136167692257067e-05,
+      "loss": 1.3583,
+      "step": 57140
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.135838625818553e-05,
+      "loss": 1.3483,
+      "step": 57150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1355095593800386e-05,
+      "loss": 1.1772,
+      "step": 57160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1351804929415254e-05,
+      "loss": 1.1785,
+      "step": 57170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.134851426503011e-05,
+      "loss": 1.2205,
+      "step": 57180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.134522360064498e-05,
+      "loss": 1.3077,
+      "step": 57190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.134193293625983e-05,
+      "loss": 1.4428,
+      "step": 57200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.133864227187469e-05,
+      "loss": 1.284,
+      "step": 57210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1335351607489554e-05,
+      "loss": 1.2471,
+      "step": 57220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1332060943104415e-05,
+      "loss": 1.2584,
+      "step": 57230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1328770278719276e-05,
+      "loss": 1.2178,
+      "step": 57240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.132547961433414e-05,
+      "loss": 1.3566,
+      "step": 57250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.132218894994899e-05,
+      "loss": 1.2636,
+      "step": 57260
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.131889828556386e-05,
+      "loss": 1.2312,
+      "step": 57270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1315607621178715e-05,
+      "loss": 1.2883,
+      "step": 57280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.131231695679358e-05,
+      "loss": 1.1702,
+      "step": 57290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.130902629240844e-05,
+      "loss": 1.4638,
+      "step": 57300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.13057356280233e-05,
+      "loss": 1.2097,
+      "step": 57310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.130244496363816e-05,
+      "loss": 1.2644,
+      "step": 57320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.129915429925302e-05,
+      "loss": 1.3606,
+      "step": 57330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.129586363486788e-05,
+      "loss": 1.3886,
+      "step": 57340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1292572970482744e-05,
+      "loss": 1.1919,
+      "step": 57350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.12892823060976e-05,
+      "loss": 1.1895,
+      "step": 57360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1285991641712466e-05,
+      "loss": 1.2415,
+      "step": 57370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.128270097732732e-05,
+      "loss": 1.231,
+      "step": 57380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.127941031294219e-05,
+      "loss": 1.2211,
+      "step": 57390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.127611964855704e-05,
+      "loss": 1.2105,
+      "step": 57400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1272828984171904e-05,
+      "loss": 1.0586,
+      "step": 57410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1269538319786766e-05,
+      "loss": 1.168,
+      "step": 57420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.126624765540163e-05,
+      "loss": 1.2701,
+      "step": 57430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.126295699101649e-05,
+      "loss": 1.268,
+      "step": 57440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.125966632663135e-05,
+      "loss": 1.2736,
+      "step": 57450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1256375662246204e-05,
+      "loss": 1.3061,
+      "step": 57460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.125308499786107e-05,
+      "loss": 1.2142,
+      "step": 57470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1249794333475927e-05,
+      "loss": 1.2823,
+      "step": 57480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1246503669090795e-05,
+      "loss": 1.4228,
+      "step": 57490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.124321300470565e-05,
+      "loss": 1.1549,
+      "step": 57500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.123992234032051e-05,
+      "loss": 1.378,
+      "step": 57510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.123663167593537e-05,
+      "loss": 1.2018,
+      "step": 57520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.123334101155023e-05,
+      "loss": 1.1901,
+      "step": 57530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1230050347165094e-05,
+      "loss": 1.4413,
+      "step": 57540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1226759682779955e-05,
+      "loss": 1.3395,
+      "step": 57550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.122346901839481e-05,
+      "loss": 1.2967,
+      "step": 57560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.122017835400968e-05,
+      "loss": 1.2991,
+      "step": 57570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.121688768962453e-05,
+      "loss": 1.2177,
+      "step": 57580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.12135970252394e-05,
+      "loss": 1.4262,
+      "step": 57590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1210306360854255e-05,
+      "loss": 1.2966,
+      "step": 57600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1207015696469116e-05,
+      "loss": 1.3884,
+      "step": 57610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.120372503208398e-05,
+      "loss": 1.2145,
+      "step": 57620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.120043436769884e-05,
+      "loss": 1.1468,
+      "step": 57630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.11971437033137e-05,
+      "loss": 1.2373,
+      "step": 57640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119385303892856e-05,
+      "loss": 1.4124,
+      "step": 57650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119056237454342e-05,
+      "loss": 1.3265,
+      "step": 57660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1187271710158284e-05,
+      "loss": 1.2574,
+      "step": 57670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1183981045773145e-05,
+      "loss": 1.3788,
+      "step": 57680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1180690381388006e-05,
+      "loss": 1.2563,
+      "step": 57690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.117739971700286e-05,
+      "loss": 1.4026,
+      "step": 57700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.117410905261772e-05,
+      "loss": 1.2716,
+      "step": 57710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.117081838823258e-05,
+      "loss": 1.1888,
+      "step": 57720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1167527723847445e-05,
+      "loss": 1.1884,
+      "step": 57730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1164237059462306e-05,
+      "loss": 1.2992,
+      "step": 57740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.116094639507717e-05,
+      "loss": 1.2787,
+      "step": 57750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.115765573069203e-05,
+      "loss": 1.248,
+      "step": 57760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.115436506630689e-05,
+      "loss": 1.4036,
+      "step": 57770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.115107440192175e-05,
+      "loss": 1.342,
+      "step": 57780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.114778373753661e-05,
+      "loss": 1.0821,
+      "step": 57790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1144493073151474e-05,
+      "loss": 1.2237,
+      "step": 57800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.114120240876633e-05,
+      "loss": 1.1977,
+      "step": 57810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1137911744381196e-05,
+      "loss": 1.3737,
+      "step": 57820
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.113462107999605e-05,
+      "loss": 1.1656,
+      "step": 57830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.113133041561092e-05,
+      "loss": 1.2404,
+      "step": 57840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.112803975122577e-05,
+      "loss": 1.2592,
+      "step": 57850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1124749086840634e-05,
+      "loss": 1.4902,
+      "step": 57860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1121458422455496e-05,
+      "loss": 1.1632,
+      "step": 57870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111816775807036e-05,
+      "loss": 1.2162,
+      "step": 57880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111487709368522e-05,
+      "loss": 1.2881,
+      "step": 57890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111158642930008e-05,
+      "loss": 1.1118,
+      "step": 57900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1108295764914934e-05,
+      "loss": 1.1317,
+      "step": 57910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.11050051005298e-05,
+      "loss": 1.3176,
+      "step": 57920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1101714436144656e-05,
+      "loss": 1.1891,
+      "step": 57930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1098423771759524e-05,
+      "loss": 1.251,
+      "step": 57940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.109513310737438e-05,
+      "loss": 1.4382,
+      "step": 57950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.109184244298924e-05,
+      "loss": 1.3806,
+      "step": 57960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.10885517786041e-05,
+      "loss": 1.1896,
+      "step": 57970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.108526111421896e-05,
+      "loss": 1.2041,
+      "step": 57980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1081970449833824e-05,
+      "loss": 1.2591,
+      "step": 57990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1078679785448685e-05,
+      "loss": 1.3213,
+      "step": 58000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.107538912106354e-05,
+      "loss": 1.3835,
+      "step": 58010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.107209845667841e-05,
+      "loss": 1.2003,
+      "step": 58020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.106880779229326e-05,
+      "loss": 1.188,
+      "step": 58030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.106551712790813e-05,
+      "loss": 1.3066,
+      "step": 58040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1062226463522985e-05,
+      "loss": 1.413,
+      "step": 58050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1058935799137846e-05,
+      "loss": 1.1254,
+      "step": 58060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.105564513475271e-05,
+      "loss": 1.233,
+      "step": 58070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.105235447036757e-05,
+      "loss": 1.2839,
+      "step": 58080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.104906380598243e-05,
+      "loss": 1.4631,
+      "step": 58090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.104577314159729e-05,
+      "loss": 1.246,
+      "step": 58100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1042482477212146e-05,
+      "loss": 1.2726,
+      "step": 58110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1039191812827014e-05,
+      "loss": 1.1881,
+      "step": 58120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.103590114844187e-05,
+      "loss": 1.3427,
+      "step": 58130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1032610484056736e-05,
+      "loss": 1.4211,
+      "step": 58140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102931981967159e-05,
+      "loss": 1.2236,
+      "step": 58150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102602915528645e-05,
+      "loss": 1.1525,
+      "step": 58160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102273849090131e-05,
+      "loss": 1.321,
+      "step": 58170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1019447826516175e-05,
+      "loss": 1.2174,
+      "step": 58180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1016157162131036e-05,
+      "loss": 1.2369,
+      "step": 58190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.10128664977459e-05,
+      "loss": 1.3469,
+      "step": 58200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.100957583336075e-05,
+      "loss": 1.3581,
+      "step": 58210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.100628516897562e-05,
+      "loss": 1.178,
+      "step": 58220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1002994504590474e-05,
+      "loss": 1.3675,
+      "step": 58230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.099970384020534e-05,
+      "loss": 1.3196,
+      "step": 58240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.09964131758202e-05,
+      "loss": 1.2242,
+      "step": 58250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.099312251143506e-05,
+      "loss": 1.0843,
+      "step": 58260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.098983184704992e-05,
+      "loss": 1.2163,
+      "step": 58270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.098654118266478e-05,
+      "loss": 1.172,
+      "step": 58280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.098325051827964e-05,
+      "loss": 1.3466,
+      "step": 58290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.09799598538945e-05,
+      "loss": 1.2869,
+      "step": 58300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097666918950936e-05,
+      "loss": 1.311,
+      "step": 58310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0973378525124226e-05,
+      "loss": 1.4029,
+      "step": 58320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097008786073908e-05,
+      "loss": 1.2177,
+      "step": 58330
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.096679719635395e-05,
+      "loss": 1.3408,
+      "step": 58340
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.09635065319688e-05,
+      "loss": 1.2642,
+      "step": 58350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0960215867583664e-05,
+      "loss": 1.3486,
+      "step": 58360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0956925203198525e-05,
+      "loss": 1.2743,
+      "step": 58370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0953634538813386e-05,
+      "loss": 1.2157,
+      "step": 58380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.095034387442825e-05,
+      "loss": 1.3269,
+      "step": 58390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094705321004311e-05,
+      "loss": 1.1376,
+      "step": 58400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094376254565797e-05,
+      "loss": 1.0916,
+      "step": 58410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094047188127283e-05,
+      "loss": 1.4327,
+      "step": 58420
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.093718121688769e-05,
+      "loss": 1.2189,
+      "step": 58430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0933890552502554e-05,
+      "loss": 1.2053,
+      "step": 58440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0930599888117415e-05,
+      "loss": 1.4591,
+      "step": 58450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.092730922373227e-05,
+      "loss": 1.1559,
+      "step": 58460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.092401855934713e-05,
+      "loss": 1.2057,
+      "step": 58470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.092072789496199e-05,
+      "loss": 1.2099,
+      "step": 58480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0917437230576854e-05,
+      "loss": 1.4134,
+      "step": 58490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0914146566191715e-05,
+      "loss": 1.3616,
+      "step": 58500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0910855901806576e-05,
+      "loss": 1.137,
+      "step": 58510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090756523742144e-05,
+      "loss": 1.2138,
+      "step": 58520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.09042745730363e-05,
+      "loss": 1.2203,
+      "step": 58530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090098390865116e-05,
+      "loss": 1.1901,
+      "step": 58540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.089769324426602e-05,
+      "loss": 1.0758,
+      "step": 58550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0894402579880876e-05,
+      "loss": 1.2062,
+      "step": 58560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0891111915495744e-05,
+      "loss": 1.3032,
+      "step": 58570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.08878212511106e-05,
+      "loss": 1.2815,
+      "step": 58580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0884530586725466e-05,
+      "loss": 1.3908,
+      "step": 58590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.088123992234032e-05,
+      "loss": 1.2927,
+      "step": 58600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.087794925795518e-05,
+      "loss": 1.0429,
+      "step": 58610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.087465859357004e-05,
+      "loss": 1.3421,
+      "step": 58620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0871367929184905e-05,
+      "loss": 1.4326,
+      "step": 58630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0868077264799766e-05,
+      "loss": 1.2082,
+      "step": 58640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.086478660041463e-05,
+      "loss": 1.3262,
+      "step": 58650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.086149593602948e-05,
+      "loss": 1.4165,
+      "step": 58660
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.085820527164435e-05,
+      "loss": 1.2716,
+      "step": 58670
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0854914607259204e-05,
+      "loss": 1.2522,
+      "step": 58680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.085162394287407e-05,
+      "loss": 1.3266,
+      "step": 58690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0848333278488927e-05,
+      "loss": 1.1443,
+      "step": 58700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.084504261410379e-05,
+      "loss": 1.2336,
+      "step": 58710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.084175194971865e-05,
+      "loss": 1.3633,
+      "step": 58720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.083846128533351e-05,
+      "loss": 1.1686,
+      "step": 58730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.083517062094837e-05,
+      "loss": 1.1869,
+      "step": 58740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.083187995656323e-05,
+      "loss": 1.2188,
+      "step": 58750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0828589292178094e-05,
+      "loss": 1.0974,
+      "step": 58760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0825298627792955e-05,
+      "loss": 1.1756,
+      "step": 58770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.082200796340781e-05,
+      "loss": 1.2083,
+      "step": 58780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.081871729902268e-05,
+      "loss": 1.1131,
+      "step": 58790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.081542663463753e-05,
+      "loss": 1.2857,
+      "step": 58800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.08121359702524e-05,
+      "loss": 1.274,
+      "step": 58810
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0808845305867255e-05,
+      "loss": 1.334,
+      "step": 58820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0805554641482116e-05,
+      "loss": 1.358,
+      "step": 58830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.080226397709698e-05,
+      "loss": 1.3013,
+      "step": 58840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.079897331271184e-05,
+      "loss": 1.2495,
+      "step": 58850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.07956826483267e-05,
+      "loss": 1.203,
+      "step": 58860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.079239198394156e-05,
+      "loss": 1.3196,
+      "step": 58870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0789101319556416e-05,
+      "loss": 1.0168,
+      "step": 58880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0785810655171284e-05,
+      "loss": 1.2569,
+      "step": 58890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.078251999078614e-05,
+      "loss": 1.2765,
+      "step": 58900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0779229326401006e-05,
+      "loss": 1.4292,
+      "step": 58910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.077593866201586e-05,
+      "loss": 1.2059,
+      "step": 58920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.077264799763072e-05,
+      "loss": 1.1954,
+      "step": 58930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0769357333245583e-05,
+      "loss": 1.1878,
+      "step": 58940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0766066668860445e-05,
+      "loss": 1.2716,
+      "step": 58950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0762776004475306e-05,
+      "loss": 1.1683,
+      "step": 58960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.075948534009017e-05,
+      "loss": 1.3844,
+      "step": 58970
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.075619467570502e-05,
+      "loss": 1.0943,
+      "step": 58980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.075290401131989e-05,
+      "loss": 1.2001,
+      "step": 58990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0749613346934744e-05,
+      "loss": 1.1919,
+      "step": 59000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.074632268254961e-05,
+      "loss": 1.3486,
+      "step": 59010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.074303201816447e-05,
+      "loss": 1.3478,
+      "step": 59020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073974135377933e-05,
+      "loss": 1.4249,
+      "step": 59030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073645068939419e-05,
+      "loss": 1.2124,
+      "step": 59040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073316002500905e-05,
+      "loss": 1.2465,
+      "step": 59050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.072986936062391e-05,
+      "loss": 1.4356,
+      "step": 59060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.072657869623877e-05,
+      "loss": 1.2619,
+      "step": 59070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.072328803185363e-05,
+      "loss": 1.1964,
+      "step": 59080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0719997367468496e-05,
+      "loss": 1.1909,
+      "step": 59090
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.071670670308335e-05,
+      "loss": 1.0788,
+      "step": 59100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.071341603869822e-05,
+      "loss": 1.1777,
+      "step": 59110
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.071012537431307e-05,
+      "loss": 1.2838,
+      "step": 59120
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0706834709927934e-05,
+      "loss": 1.247,
+      "step": 59130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0703544045542795e-05,
+      "loss": 1.4785,
+      "step": 59140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0700253381157657e-05,
+      "loss": 1.1936,
+      "step": 59150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.069696271677252e-05,
+      "loss": 1.1856,
+      "step": 59160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.069367205238738e-05,
+      "loss": 1.4128,
+      "step": 59170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.069038138800224e-05,
+      "loss": 1.3966,
+      "step": 59180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.06870907236171e-05,
+      "loss": 1.2755,
+      "step": 59190
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.068380005923196e-05,
+      "loss": 1.3183,
+      "step": 59200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0680509394846824e-05,
+      "loss": 1.3211,
+      "step": 59210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0677218730461685e-05,
+      "loss": 1.2238,
+      "step": 59220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.067392806607654e-05,
+      "loss": 1.4419,
+      "step": 59230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.06706374016914e-05,
+      "loss": 1.1979,
+      "step": 59240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.066734673730626e-05,
+      "loss": 1.289,
+      "step": 59250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0664056072921124e-05,
+      "loss": 1.2845,
+      "step": 59260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0660765408535985e-05,
+      "loss": 1.3264,
+      "step": 59270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0657474744150846e-05,
+      "loss": 1.1551,
+      "step": 59280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.065418407976571e-05,
+      "loss": 1.3521,
+      "step": 59290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.065089341538057e-05,
+      "loss": 1.3491,
+      "step": 59300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.064760275099543e-05,
+      "loss": 1.3071,
+      "step": 59310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.064431208661029e-05,
+      "loss": 1.2885,
+      "step": 59320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0641021422225146e-05,
+      "loss": 1.2692,
+      "step": 59330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0637730757840014e-05,
+      "loss": 1.3317,
+      "step": 59340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.063444009345487e-05,
+      "loss": 1.1629,
+      "step": 59350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0631149429069736e-05,
+      "loss": 1.3062,
+      "step": 59360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.062785876468459e-05,
+      "loss": 1.1422,
+      "step": 59370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.062456810029945e-05,
+      "loss": 1.2476,
+      "step": 59380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0621277435914313e-05,
+      "loss": 1.2888,
+      "step": 59390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0617986771529175e-05,
+      "loss": 1.2908,
+      "step": 59400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0614696107144036e-05,
+      "loss": 1.2906,
+      "step": 59410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.06114054427589e-05,
+      "loss": 1.1989,
+      "step": 59420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.060811477837375e-05,
+      "loss": 1.2967,
+      "step": 59430
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.060482411398862e-05,
+      "loss": 1.4096,
+      "step": 59440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0601533449603474e-05,
+      "loss": 1.3918,
+      "step": 59450
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059824278521834e-05,
+      "loss": 1.248,
+      "step": 59460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.05949521208332e-05,
+      "loss": 1.3575,
+      "step": 59470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059166145644806e-05,
+      "loss": 1.2147,
+      "step": 59480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.058837079206292e-05,
+      "loss": 1.3928,
+      "step": 59490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.058508012767778e-05,
+      "loss": 1.2245,
+      "step": 59500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.058178946329264e-05,
+      "loss": 1.3863,
+      "step": 59510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.05784987989075e-05,
+      "loss": 1.1911,
+      "step": 59520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.057520813452236e-05,
+      "loss": 1.3517,
+      "step": 59530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0571917470137226e-05,
+      "loss": 1.3435,
+      "step": 59540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.056862680575208e-05,
+      "loss": 1.1906,
+      "step": 59550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.056533614136695e-05,
+      "loss": 1.3064,
+      "step": 59560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.05620454769818e-05,
+      "loss": 1.2491,
+      "step": 59570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0558754812596664e-05,
+      "loss": 1.275,
+      "step": 59580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0555464148211525e-05,
+      "loss": 1.3464,
+      "step": 59590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0552173483826386e-05,
+      "loss": 1.1668,
+      "step": 59600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.054888281944125e-05,
+      "loss": 1.2312,
+      "step": 59610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.054559215505611e-05,
+      "loss": 1.3092,
+      "step": 59620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0542301490670963e-05,
+      "loss": 1.4149,
+      "step": 59630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.053901082628583e-05,
+      "loss": 1.2974,
+      "step": 59640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0535720161900686e-05,
+      "loss": 1.2537,
+      "step": 59650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0532429497515554e-05,
+      "loss": 1.4577,
+      "step": 59660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052913883313041e-05,
+      "loss": 1.133,
+      "step": 59670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052584816874527e-05,
+      "loss": 1.2196,
+      "step": 59680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052255750436013e-05,
+      "loss": 1.3786,
+      "step": 59690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.051926683997499e-05,
+      "loss": 0.9998,
+      "step": 59700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0515976175589857e-05,
+      "loss": 1.1854,
+      "step": 59710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0512685511204715e-05,
+      "loss": 1.2286,
+      "step": 59720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0509394846819573e-05,
+      "loss": 1.1799,
+      "step": 59730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0506104182434437e-05,
+      "loss": 1.2346,
+      "step": 59740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0502813518049295e-05,
+      "loss": 1.45,
+      "step": 59750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.049952285366416e-05,
+      "loss": 1.3944,
+      "step": 59760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0496232189279018e-05,
+      "loss": 1.2727,
+      "step": 59770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0492941524893876e-05,
+      "loss": 1.2447,
+      "step": 59780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.048965086050874e-05,
+      "loss": 1.1365,
+      "step": 59790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0486360196123598e-05,
+      "loss": 1.249,
+      "step": 59800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0483069531738463e-05,
+      "loss": 1.2576,
+      "step": 59810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.047977886735332e-05,
+      "loss": 1.3587,
+      "step": 59820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.047648820296818e-05,
+      "loss": 1.2217,
+      "step": 59830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0473197538583043e-05,
+      "loss": 1.3994,
+      "step": 59840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.04699068741979e-05,
+      "loss": 1.3322,
+      "step": 59850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0466616209812766e-05,
+      "loss": 1.3928,
+      "step": 59860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0463325545427624e-05,
+      "loss": 1.311,
+      "step": 59870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.046003488104248e-05,
+      "loss": 1.2839,
+      "step": 59880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0456744216657346e-05,
+      "loss": 1.5097,
+      "step": 59890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0453453552272204e-05,
+      "loss": 1.4631,
+      "step": 59900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.045016288788707e-05,
+      "loss": 1.2531,
+      "step": 59910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0446872223501927e-05,
+      "loss": 1.177,
+      "step": 59920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0443581559116785e-05,
+      "loss": 1.1111,
+      "step": 59930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.044029089473165e-05,
+      "loss": 1.278,
+      "step": 59940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0437000230346507e-05,
+      "loss": 1.4002,
+      "step": 59950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0433709565961372e-05,
+      "loss": 1.2468,
+      "step": 59960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.043041890157623e-05,
+      "loss": 1.3582,
+      "step": 59970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0427128237191088e-05,
+      "loss": 1.3774,
+      "step": 59980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0423837572805952e-05,
+      "loss": 1.2249,
+      "step": 59990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.042054690842081e-05,
+      "loss": 1.1665,
+      "step": 60000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0417256244035675e-05,
+      "loss": 1.2241,
+      "step": 60010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0413965579650533e-05,
+      "loss": 1.2245,
+      "step": 60020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.041067491526539e-05,
+      "loss": 1.2668,
+      "step": 60030
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0407384250880255e-05,
+      "loss": 1.2934,
+      "step": 60040
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0404093586495113e-05,
+      "loss": 1.1794,
+      "step": 60050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0400802922109978e-05,
+      "loss": 1.1465,
+      "step": 60060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0397512257724836e-05,
+      "loss": 1.321,
+      "step": 60070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0394221593339693e-05,
+      "loss": 1.2911,
+      "step": 60080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0390930928954558e-05,
+      "loss": 1.3238,
+      "step": 60090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0387640264569416e-05,
+      "loss": 1.1783,
+      "step": 60100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.038434960018428e-05,
+      "loss": 1.1837,
+      "step": 60110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.038105893579914e-05,
+      "loss": 1.2469,
+      "step": 60120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0377768271413996e-05,
+      "loss": 1.2654,
+      "step": 60130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.037447760702886e-05,
+      "loss": 1.3262,
+      "step": 60140
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.037118694264372e-05,
+      "loss": 1.0581,
+      "step": 60150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0367896278258584e-05,
+      "loss": 1.3028,
+      "step": 60160
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.036460561387344e-05,
+      "loss": 1.2261,
+      "step": 60170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.03613149494883e-05,
+      "loss": 1.1808,
+      "step": 60180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0358024285103164e-05,
+      "loss": 1.5855,
+      "step": 60190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0354733620718022e-05,
+      "loss": 1.3645,
+      "step": 60200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0351442956332887e-05,
+      "loss": 1.1761,
+      "step": 60210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0348152291947744e-05,
+      "loss": 1.2141,
+      "step": 60220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0344861627562606e-05,
+      "loss": 1.2031,
+      "step": 60230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0341570963177467e-05,
+      "loss": 1.2872,
+      "step": 60240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0338280298792325e-05,
+      "loss": 1.3639,
+      "step": 60250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.033498963440719e-05,
+      "loss": 1.2168,
+      "step": 60260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0331698970022047e-05,
+      "loss": 1.2946,
+      "step": 60270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.032840830563691e-05,
+      "loss": 1.1089,
+      "step": 60280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.032511764125177e-05,
+      "loss": 1.2554,
+      "step": 60290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.032182697686663e-05,
+      "loss": 1.3367,
+      "step": 60300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0318536312481492e-05,
+      "loss": 1.2158,
+      "step": 60310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0315245648096354e-05,
+      "loss": 1.2171,
+      "step": 60320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.031195498371121e-05,
+      "loss": 1.214,
+      "step": 60330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0308664319326073e-05,
+      "loss": 1.332,
+      "step": 60340
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0305373654940934e-05,
+      "loss": 1.4573,
+      "step": 60350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0302082990555795e-05,
+      "loss": 1.3176,
+      "step": 60360
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0298792326170657e-05,
+      "loss": 1.4765,
+      "step": 60370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0295501661785515e-05,
+      "loss": 1.471,
+      "step": 60380
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.029221099740038e-05,
+      "loss": 1.3629,
+      "step": 60390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0288920333015237e-05,
+      "loss": 1.2539,
+      "step": 60400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.02856296686301e-05,
+      "loss": 1.1708,
+      "step": 60410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.028233900424496e-05,
+      "loss": 1.2468,
+      "step": 60420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0279048339859817e-05,
+      "loss": 1.403,
+      "step": 60430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0275757675474682e-05,
+      "loss": 1.2123,
+      "step": 60440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.027246701108954e-05,
+      "loss": 1.3425,
+      "step": 60450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0269176346704405e-05,
+      "loss": 1.3059,
+      "step": 60460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0265885682319263e-05,
+      "loss": 1.2993,
+      "step": 60470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.026259501793412e-05,
+      "loss": 1.3002,
+      "step": 60480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0259304353548985e-05,
+      "loss": 1.1455,
+      "step": 60490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0256013689163843e-05,
+      "loss": 1.2829,
+      "step": 60500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0252723024778708e-05,
+      "loss": 1.1865,
+      "step": 60510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0249432360393565e-05,
+      "loss": 1.1663,
+      "step": 60520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0246141696008423e-05,
+      "loss": 1.1204,
+      "step": 60530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0242851031623288e-05,
+      "loss": 1.0681,
+      "step": 60540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0239560367238146e-05,
+      "loss": 1.1971,
+      "step": 60550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.023626970285301e-05,
+      "loss": 1.2666,
+      "step": 60560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.023297903846787e-05,
+      "loss": 1.3757,
+      "step": 60570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0229688374082726e-05,
+      "loss": 1.2792,
+      "step": 60580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.022639770969759e-05,
+      "loss": 1.2243,
+      "step": 60590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.022310704531245e-05,
+      "loss": 1.2794,
+      "step": 60600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0219816380927313e-05,
+      "loss": 1.2473,
+      "step": 60610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.021652571654217e-05,
+      "loss": 1.2798,
+      "step": 60620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.021323505215703e-05,
+      "loss": 1.2416,
+      "step": 60630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0209944387771894e-05,
+      "loss": 1.287,
+      "step": 60640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0206653723386752e-05,
+      "loss": 1.3781,
+      "step": 60650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0203363059001616e-05,
+      "loss": 1.1554,
+      "step": 60660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0200072394616474e-05,
+      "loss": 1.1902,
+      "step": 60670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0196781730231332e-05,
+      "loss": 1.1818,
+      "step": 60680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0193491065846197e-05,
+      "loss": 1.2517,
+      "step": 60690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0190200401461055e-05,
+      "loss": 1.3293,
+      "step": 60700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.018690973707592e-05,
+      "loss": 1.2475,
+      "step": 60710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0183619072690777e-05,
+      "loss": 1.3816,
+      "step": 60720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0180328408305635e-05,
+      "loss": 1.2597,
+      "step": 60730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.01770377439205e-05,
+      "loss": 1.2828,
+      "step": 60740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0173747079535358e-05,
+      "loss": 1.3673,
+      "step": 60750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0170456415150222e-05,
+      "loss": 1.0291,
+      "step": 60760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.016716575076508e-05,
+      "loss": 1.2783,
+      "step": 60770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0163875086379938e-05,
+      "loss": 1.2859,
+      "step": 60780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0160584421994803e-05,
+      "loss": 1.1533,
+      "step": 60790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.015729375760966e-05,
+      "loss": 1.3991,
+      "step": 60800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0154003093224525e-05,
+      "loss": 1.2558,
+      "step": 60810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0150712428839383e-05,
+      "loss": 1.3629,
+      "step": 60820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.014742176445424e-05,
+      "loss": 1.287,
+      "step": 60830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0144131100069106e-05,
+      "loss": 1.2946,
+      "step": 60840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0140840435683964e-05,
+      "loss": 1.3165,
+      "step": 60850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0137549771298828e-05,
+      "loss": 1.3601,
+      "step": 60860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0134259106913686e-05,
+      "loss": 1.325,
+      "step": 60870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0130968442528544e-05,
+      "loss": 1.4943,
+      "step": 60880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.012767777814341e-05,
+      "loss": 1.2957,
+      "step": 60890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0124387113758267e-05,
+      "loss": 1.1542,
+      "step": 60900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.012109644937313e-05,
+      "loss": 1.4496,
+      "step": 60910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.011780578498799e-05,
+      "loss": 1.3244,
+      "step": 60920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0114515120602847e-05,
+      "loss": 1.2938,
+      "step": 60930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.011122445621771e-05,
+      "loss": 1.33,
+      "step": 60940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.010793379183257e-05,
+      "loss": 1.4104,
+      "step": 60950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0104643127447434e-05,
+      "loss": 1.2451,
+      "step": 60960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0101352463062292e-05,
+      "loss": 1.2151,
+      "step": 60970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0098061798677153e-05,
+      "loss": 1.2922,
+      "step": 60980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0094771134292015e-05,
+      "loss": 1.3086,
+      "step": 60990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0091480469906876e-05,
+      "loss": 1.2853,
+      "step": 61000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0088189805521737e-05,
+      "loss": 1.2996,
+      "step": 61010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0084899141136595e-05,
+      "loss": 1.134,
+      "step": 61020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0081608476751456e-05,
+      "loss": 1.1688,
+      "step": 61030
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0078317812366318e-05,
+      "loss": 1.3754,
+      "step": 61040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.007502714798118e-05,
+      "loss": 1.1721,
+      "step": 61050
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.007173648359604e-05,
+      "loss": 1.2675,
+      "step": 61060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.00684458192109e-05,
+      "loss": 1.253,
+      "step": 61070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.006515515482576e-05,
+      "loss": 1.1973,
+      "step": 61080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.006186449044062e-05,
+      "loss": 1.2871,
+      "step": 61090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0058573826055482e-05,
+      "loss": 1.1334,
+      "step": 61100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0055283161670343e-05,
+      "loss": 1.3181,
+      "step": 61110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0051992497285204e-05,
+      "loss": 1.2334,
+      "step": 61120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0048701832900062e-05,
+      "loss": 1.254,
+      "step": 61130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0045411168514927e-05,
+      "loss": 1.3059,
+      "step": 61140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0042120504129785e-05,
+      "loss": 1.1849,
+      "step": 61150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.003882983974465e-05,
+      "loss": 1.4645,
+      "step": 61160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0035539175359507e-05,
+      "loss": 1.1909,
+      "step": 61170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0032248510974365e-05,
+      "loss": 1.375,
+      "step": 61180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.002895784658923e-05,
+      "loss": 1.2018,
+      "step": 61190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0025667182204088e-05,
+      "loss": 1.1313,
+      "step": 61200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0022376517818952e-05,
+      "loss": 1.3309,
+      "step": 61210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.001908585343381e-05,
+      "loss": 1.2332,
+      "step": 61220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0015795189048668e-05,
+      "loss": 1.2095,
+      "step": 61230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0012504524663533e-05,
+      "loss": 1.3457,
+      "step": 61240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.000921386027839e-05,
+      "loss": 1.2013,
+      "step": 61250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0005923195893255e-05,
+      "loss": 1.2552,
+      "step": 61260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0002632531508113e-05,
+      "loss": 1.2502,
+      "step": 61270
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.999934186712297e-05,
+      "loss": 1.3056,
+      "step": 61280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9996051202737836e-05,
+      "loss": 1.3576,
+      "step": 61290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9992760538352694e-05,
+      "loss": 1.1784,
+      "step": 61300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9989469873967558e-05,
+      "loss": 1.2936,
+      "step": 61310
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9986179209582416e-05,
+      "loss": 1.1668,
+      "step": 61320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9982888545197274e-05,
+      "loss": 1.2491,
+      "step": 61330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.997959788081214e-05,
+      "loss": 1.2943,
+      "step": 61340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9976307216426996e-05,
+      "loss": 1.0553,
+      "step": 61350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.997301655204186e-05,
+      "loss": 1.2256,
+      "step": 61360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.996972588765672e-05,
+      "loss": 1.2769,
+      "step": 61370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9966435223271577e-05,
+      "loss": 1.3327,
+      "step": 61380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.996314455888644e-05,
+      "loss": 1.3281,
+      "step": 61390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.99598538945013e-05,
+      "loss": 1.2487,
+      "step": 61400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9956563230116164e-05,
+      "loss": 1.2167,
+      "step": 61410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9953272565731022e-05,
+      "loss": 1.3644,
+      "step": 61420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.994998190134588e-05,
+      "loss": 1.3373,
+      "step": 61430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9946691236960744e-05,
+      "loss": 1.1829,
+      "step": 61440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9943400572575602e-05,
+      "loss": 1.6414,
+      "step": 61450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9940109908190467e-05,
+      "loss": 1.2834,
+      "step": 61460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9936819243805325e-05,
+      "loss": 1.2312,
+      "step": 61470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9933528579420183e-05,
+      "loss": 1.1584,
+      "step": 61480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9930237915035047e-05,
+      "loss": 1.2111,
+      "step": 61490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9926947250649905e-05,
+      "loss": 1.2364,
+      "step": 61500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.992365658626477e-05,
+      "loss": 1.1992,
+      "step": 61510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9920365921879628e-05,
+      "loss": 1.3359,
+      "step": 61520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9917075257494486e-05,
+      "loss": 1.3617,
+      "step": 61530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.991378459310935e-05,
+      "loss": 1.2406,
+      "step": 61540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9910493928724208e-05,
+      "loss": 1.3189,
+      "step": 61550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9907203264339073e-05,
+      "loss": 1.2325,
+      "step": 61560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.990391259995393e-05,
+      "loss": 1.1586,
+      "step": 61570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9900621935568795e-05,
+      "loss": 1.3387,
+      "step": 61580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9897331271183653e-05,
+      "loss": 1.2821,
+      "step": 61590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.989404060679851e-05,
+      "loss": 1.167,
+      "step": 61600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9890749942413376e-05,
+      "loss": 1.2437,
+      "step": 61610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9887459278028234e-05,
+      "loss": 1.2507,
+      "step": 61620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.98841686136431e-05,
+      "loss": 1.2567,
+      "step": 61630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9880877949257956e-05,
+      "loss": 1.2489,
+      "step": 61640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9877587284872814e-05,
+      "loss": 1.2549,
+      "step": 61650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.987429662048768e-05,
+      "loss": 1.3369,
+      "step": 61660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9871005956102537e-05,
+      "loss": 1.2841,
+      "step": 61670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.98677152917174e-05,
+      "loss": 1.2497,
+      "step": 61680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.986442462733226e-05,
+      "loss": 1.2546,
+      "step": 61690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9861133962947117e-05,
+      "loss": 1.4057,
+      "step": 61700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9857843298561982e-05,
+      "loss": 1.1996,
+      "step": 61710
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.985455263417684e-05,
+      "loss": 1.2921,
+      "step": 61720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9851261969791704e-05,
+      "loss": 1.3777,
+      "step": 61730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9847971305406562e-05,
+      "loss": 1.2602,
+      "step": 61740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9844680641021423e-05,
+      "loss": 1.2977,
+      "step": 61750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9841389976636285e-05,
+      "loss": 1.2772,
+      "step": 61760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9838099312251146e-05,
+      "loss": 1.3542,
+      "step": 61770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9834808647866007e-05,
+      "loss": 1.2948,
+      "step": 61780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9831517983480865e-05,
+      "loss": 1.28,
+      "step": 61790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9828227319095726e-05,
+      "loss": 1.4061,
+      "step": 61800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9824936654710588e-05,
+      "loss": 1.2756,
+      "step": 61810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.982164599032545e-05,
+      "loss": 1.1989,
+      "step": 61820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.981835532594031e-05,
+      "loss": 1.3395,
+      "step": 61830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.981506466155517e-05,
+      "loss": 1.2998,
+      "step": 61840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.981177399717003e-05,
+      "loss": 1.15,
+      "step": 61850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.980848333278489e-05,
+      "loss": 1.1962,
+      "step": 61860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9805192668399752e-05,
+      "loss": 1.2897,
+      "step": 61870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9801902004014613e-05,
+      "loss": 1.2579,
+      "step": 61880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9798611339629474e-05,
+      "loss": 1.2292,
+      "step": 61890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9795320675244332e-05,
+      "loss": 1.1169,
+      "step": 61900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9792030010859197e-05,
+      "loss": 1.0585,
+      "step": 61910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9788739346474055e-05,
+      "loss": 1.2778,
+      "step": 61920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.978544868208892e-05,
+      "loss": 1.1792,
+      "step": 61930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9782158017703777e-05,
+      "loss": 1.2032,
+      "step": 61940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9778867353318635e-05,
+      "loss": 1.3177,
+      "step": 61950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.97755766889335e-05,
+      "loss": 1.318,
+      "step": 61960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9772286024548358e-05,
+      "loss": 1.2236,
+      "step": 61970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9768995360163222e-05,
+      "loss": 1.1645,
+      "step": 61980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.976570469577808e-05,
+      "loss": 1.1987,
+      "step": 61990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9762414031392938e-05,
+      "loss": 1.073,
+      "step": 62000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9759123367007803e-05,
+      "loss": 1.4379,
+      "step": 62010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.975583270262266e-05,
+      "loss": 1.323,
+      "step": 62020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9752542038237525e-05,
+      "loss": 1.3468,
+      "step": 62030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9749251373852383e-05,
+      "loss": 1.4548,
+      "step": 62040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.974596070946724e-05,
+      "loss": 1.5525,
+      "step": 62050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9742670045082106e-05,
+      "loss": 1.2469,
+      "step": 62060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9739379380696964e-05,
+      "loss": 1.2701,
+      "step": 62070
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.973608871631183e-05,
+      "loss": 1.0737,
+      "step": 62080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9732798051926686e-05,
+      "loss": 1.2395,
+      "step": 62090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9729507387541544e-05,
+      "loss": 1.2689,
+      "step": 62100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.972621672315641e-05,
+      "loss": 1.1926,
+      "step": 62110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9722926058771267e-05,
+      "loss": 1.3414,
+      "step": 62120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.971963539438613e-05,
+      "loss": 1.2459,
+      "step": 62130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.971634473000099e-05,
+      "loss": 1.3305,
+      "step": 62140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9713054065615847e-05,
+      "loss": 1.2728,
+      "step": 62150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9709763401230712e-05,
+      "loss": 1.1862,
+      "step": 62160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.970647273684557e-05,
+      "loss": 1.2464,
+      "step": 62170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9703182072460434e-05,
+      "loss": 1.2537,
+      "step": 62180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9699891408075292e-05,
+      "loss": 1.2547,
+      "step": 62190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.969660074369015e-05,
+      "loss": 1.3284,
+      "step": 62200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9693310079305015e-05,
+      "loss": 1.3539,
+      "step": 62210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9690019414919873e-05,
+      "loss": 1.2033,
+      "step": 62220
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9686728750534737e-05,
+      "loss": 1.1941,
+      "step": 62230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9683438086149595e-05,
+      "loss": 1.2194,
+      "step": 62240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9680147421764453e-05,
+      "loss": 1.435,
+      "step": 62250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9676856757379318e-05,
+      "loss": 1.3028,
+      "step": 62260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9673566092994175e-05,
+      "loss": 1.3341,
+      "step": 62270
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.967027542860904e-05,
+      "loss": 1.2703,
+      "step": 62280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9666984764223898e-05,
+      "loss": 1.204,
+      "step": 62290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9663694099838756e-05,
+      "loss": 1.2084,
+      "step": 62300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.966040343545362e-05,
+      "loss": 1.2337,
+      "step": 62310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.965711277106848e-05,
+      "loss": 1.1947,
+      "step": 62320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9653822106683343e-05,
+      "loss": 1.1526,
+      "step": 62330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.96505314422982e-05,
+      "loss": 1.231,
+      "step": 62340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.964724077791306e-05,
+      "loss": 1.2067,
+      "step": 62350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9643950113527923e-05,
+      "loss": 1.3176,
+      "step": 62360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.964065944914278e-05,
+      "loss": 1.3906,
+      "step": 62370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9637368784757646e-05,
+      "loss": 1.1566,
+      "step": 62380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9634078120372504e-05,
+      "loss": 1.2788,
+      "step": 62390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9630787455987362e-05,
+      "loss": 1.1768,
+      "step": 62400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9627496791602226e-05,
+      "loss": 1.3488,
+      "step": 62410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9624206127217084e-05,
+      "loss": 1.2995,
+      "step": 62420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.962091546283195e-05,
+      "loss": 1.4537,
+      "step": 62430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9617624798446807e-05,
+      "loss": 1.1102,
+      "step": 62440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9614334134061668e-05,
+      "loss": 1.279,
+      "step": 62450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.961104346967653e-05,
+      "loss": 1.494,
+      "step": 62460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9607752805291387e-05,
+      "loss": 1.53,
+      "step": 62470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9604462140906252e-05,
+      "loss": 1.2336,
+      "step": 62480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.960117147652111e-05,
+      "loss": 1.3716,
+      "step": 62490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.959788081213597e-05,
+      "loss": 1.2383,
+      "step": 62500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9594590147750832e-05,
+      "loss": 1.2505,
+      "step": 62510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9591299483365694e-05,
+      "loss": 1.3959,
+      "step": 62520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9588008818980555e-05,
+      "loss": 1.2424,
+      "step": 62530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9584718154595413e-05,
+      "loss": 1.2331,
+      "step": 62540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9581427490210274e-05,
+      "loss": 1.4074,
+      "step": 62550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9578136825825135e-05,
+      "loss": 1.1222,
+      "step": 62560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9574846161439997e-05,
+      "loss": 1.2413,
+      "step": 62570
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9571555497054858e-05,
+      "loss": 1.233,
+      "step": 62580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.956826483266972e-05,
+      "loss": 1.2639,
+      "step": 62590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9564974168284577e-05,
+      "loss": 1.3782,
+      "step": 62600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.956168350389944e-05,
+      "loss": 1.329,
+      "step": 62610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.95583928395143e-05,
+      "loss": 1.2976,
+      "step": 62620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.955510217512916e-05,
+      "loss": 1.2756,
+      "step": 62630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9551811510744022e-05,
+      "loss": 1.496,
+      "step": 62640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.954852084635888e-05,
+      "loss": 1.2466,
+      "step": 62650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9545230181973745e-05,
+      "loss": 1.2858,
+      "step": 62660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9541939517588602e-05,
+      "loss": 1.2435,
+      "step": 62670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9538648853203467e-05,
+      "loss": 1.2099,
+      "step": 62680
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9535358188818325e-05,
+      "loss": 1.2622,
+      "step": 62690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9532067524433183e-05,
+      "loss": 1.3617,
+      "step": 62700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9528776860048048e-05,
+      "loss": 1.3444,
+      "step": 62710
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9525486195662905e-05,
+      "loss": 1.2269,
+      "step": 62720
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.952219553127777e-05,
+      "loss": 1.2743,
+      "step": 62730
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9518904866892628e-05,
+      "loss": 1.2046,
+      "step": 62740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9515614202507486e-05,
+      "loss": 1.1953,
+      "step": 62750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.951232353812235e-05,
+      "loss": 1.3933,
+      "step": 62760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.950903287373721e-05,
+      "loss": 1.2177,
+      "step": 62770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9505742209352073e-05,
+      "loss": 1.2603,
+      "step": 62780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.950245154496693e-05,
+      "loss": 1.1831,
+      "step": 62790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.949916088058179e-05,
+      "loss": 1.3093,
+      "step": 62800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9495870216196653e-05,
+      "loss": 1.1708,
+      "step": 62810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.949257955181151e-05,
+      "loss": 1.3371,
+      "step": 62820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9489288887426376e-05,
+      "loss": 1.2393,
+      "step": 62830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9485998223041234e-05,
+      "loss": 1.2116,
+      "step": 62840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9482707558656092e-05,
+      "loss": 1.4129,
+      "step": 62850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9479416894270956e-05,
+      "loss": 1.2456,
+      "step": 62860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9476126229885814e-05,
+      "loss": 1.362,
+      "step": 62870
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.947283556550068e-05,
+      "loss": 1.2287,
+      "step": 62880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9469544901115537e-05,
+      "loss": 1.4688,
+      "step": 62890
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9466254236730395e-05,
+      "loss": 1.4526,
+      "step": 62900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.946296357234526e-05,
+      "loss": 1.2813,
+      "step": 62910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9459672907960117e-05,
+      "loss": 1.0699,
+      "step": 62920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9456382243574982e-05,
+      "loss": 1.2111,
+      "step": 62930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.945309157918984e-05,
+      "loss": 1.3976,
+      "step": 62940
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9449800914804698e-05,
+      "loss": 1.1615,
+      "step": 62950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9446510250419562e-05,
+      "loss": 1.275,
+      "step": 62960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.944321958603442e-05,
+      "loss": 1.2168,
+      "step": 62970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9439928921649285e-05,
+      "loss": 1.2391,
+      "step": 62980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9436638257264143e-05,
+      "loss": 1.1486,
+      "step": 62990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9433347592879e-05,
+      "loss": 1.2549,
+      "step": 63000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9430056928493865e-05,
+      "loss": 1.2349,
+      "step": 63010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9426766264108723e-05,
+      "loss": 1.2622,
+      "step": 63020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9423475599723588e-05,
+      "loss": 1.2402,
+      "step": 63030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9420184935338446e-05,
+      "loss": 1.2052,
+      "step": 63040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9416894270953304e-05,
+      "loss": 1.3047,
+      "step": 63050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9413603606568168e-05,
+      "loss": 1.0853,
+      "step": 63060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9410312942183026e-05,
+      "loss": 1.1241,
+      "step": 63070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.940702227779789e-05,
+      "loss": 1.2216,
+      "step": 63080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.940373161341275e-05,
+      "loss": 1.032,
+      "step": 63090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9400440949027606e-05,
+      "loss": 1.2645,
+      "step": 63100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.939715028464247e-05,
+      "loss": 1.2541,
+      "step": 63110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.939385962025733e-05,
+      "loss": 1.3537,
+      "step": 63120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9390568955872194e-05,
+      "loss": 1.3119,
+      "step": 63130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.938727829148705e-05,
+      "loss": 1.192,
+      "step": 63140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.938398762710191e-05,
+      "loss": 1.3734,
+      "step": 63150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9380696962716774e-05,
+      "loss": 1.3256,
+      "step": 63160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9377406298331632e-05,
+      "loss": 1.2615,
+      "step": 63170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9374115633946497e-05,
+      "loss": 1.4445,
+      "step": 63180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9370824969561354e-05,
+      "loss": 1.2908,
+      "step": 63190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9367534305176216e-05,
+      "loss": 1.2409,
+      "step": 63200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9364243640791077e-05,
+      "loss": 1.2004,
+      "step": 63210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9360952976405935e-05,
+      "loss": 1.2467,
+      "step": 63220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.93576623120208e-05,
+      "loss": 1.298,
+      "step": 63230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9354371647635657e-05,
+      "loss": 1.3016,
+      "step": 63240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.935108098325052e-05,
+      "loss": 1.2034,
+      "step": 63250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.934779031886538e-05,
+      "loss": 1.2254,
+      "step": 63260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.934449965448024e-05,
+      "loss": 1.2274,
+      "step": 63270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9341208990095102e-05,
+      "loss": 1.179,
+      "step": 63280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9337918325709964e-05,
+      "loss": 1.2404,
+      "step": 63290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.933462766132482e-05,
+      "loss": 1.2102,
+      "step": 63300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9331336996939683e-05,
+      "loss": 1.2989,
+      "step": 63310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9328046332554544e-05,
+      "loss": 1.296,
+      "step": 63320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9324755668169405e-05,
+      "loss": 1.7748,
+      "step": 63330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9321465003784267e-05,
+      "loss": 1.2351,
+      "step": 63340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9318174339399125e-05,
+      "loss": 1.1707,
+      "step": 63350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.931488367501399e-05,
+      "loss": 1.1239,
+      "step": 63360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9311593010628847e-05,
+      "loss": 1.1571,
+      "step": 63370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.930830234624371e-05,
+      "loss": 1.1421,
+      "step": 63380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.930501168185857e-05,
+      "loss": 1.367,
+      "step": 63390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9301721017473428e-05,
+      "loss": 1.2492,
+      "step": 63400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9298430353088292e-05,
+      "loss": 1.2536,
+      "step": 63410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.929513968870315e-05,
+      "loss": 1.2151,
+      "step": 63420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9291849024318015e-05,
+      "loss": 1.3323,
+      "step": 63430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9288558359932873e-05,
+      "loss": 1.3575,
+      "step": 63440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.928526769554773e-05,
+      "loss": 1.4814,
+      "step": 63450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9281977031162595e-05,
+      "loss": 1.2643,
+      "step": 63460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9278686366777453e-05,
+      "loss": 1.3346,
+      "step": 63470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9275395702392318e-05,
+      "loss": 1.1551,
+      "step": 63480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9272105038007176e-05,
+      "loss": 1.238,
+      "step": 63490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9268814373622033e-05,
+      "loss": 1.4065,
+      "step": 63500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9265523709236898e-05,
+      "loss": 1.3564,
+      "step": 63510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9262233044851756e-05,
+      "loss": 1.2695,
+      "step": 63520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.925894238046662e-05,
+      "loss": 1.2443,
+      "step": 63530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.925565171608148e-05,
+      "loss": 1.3531,
+      "step": 63540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9252361051696336e-05,
+      "loss": 1.3706,
+      "step": 63550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.92490703873112e-05,
+      "loss": 1.2913,
+      "step": 63560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.924577972292606e-05,
+      "loss": 1.3853,
+      "step": 63570
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9242489058540924e-05,
+      "loss": 1.1859,
+      "step": 63580
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.923919839415578e-05,
+      "loss": 1.2854,
+      "step": 63590
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.923590772977064e-05,
+      "loss": 1.2525,
+      "step": 63600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9232617065385504e-05,
+      "loss": 1.3184,
+      "step": 63610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9229326401000362e-05,
+      "loss": 1.4478,
+      "step": 63620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9226035736615227e-05,
+      "loss": 1.1542,
+      "step": 63630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9222745072230084e-05,
+      "loss": 1.4554,
+      "step": 63640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9219454407844942e-05,
+      "loss": 1.1974,
+      "step": 63650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9216163743459807e-05,
+      "loss": 1.2281,
+      "step": 63660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9212873079074665e-05,
+      "loss": 1.2013,
+      "step": 63670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.920958241468953e-05,
+      "loss": 1.3159,
+      "step": 63680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9206291750304387e-05,
+      "loss": 1.2508,
+      "step": 63690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9203001085919245e-05,
+      "loss": 1.2021,
+      "step": 63700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.919971042153411e-05,
+      "loss": 1.2382,
+      "step": 63710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9196419757148968e-05,
+      "loss": 1.4625,
+      "step": 63720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9193129092763832e-05,
+      "loss": 1.2833,
+      "step": 63730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.918983842837869e-05,
+      "loss": 1.2718,
+      "step": 63740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9186547763993548e-05,
+      "loss": 1.3099,
+      "step": 63750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9183257099608413e-05,
+      "loss": 1.2482,
+      "step": 63760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.917996643522327e-05,
+      "loss": 1.3549,
+      "step": 63770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9176675770838135e-05,
+      "loss": 1.2862,
+      "step": 63780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9173385106452993e-05,
+      "loss": 1.317,
+      "step": 63790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.917009444206785e-05,
+      "loss": 1.1999,
+      "step": 63800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9166803777682716e-05,
+      "loss": 1.2485,
+      "step": 63810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9163513113297574e-05,
+      "loss": 1.1948,
+      "step": 63820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.916022244891244e-05,
+      "loss": 1.302,
+      "step": 63830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9156931784527296e-05,
+      "loss": 1.3874,
+      "step": 63840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9153641120142154e-05,
+      "loss": 1.2284,
+      "step": 63850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.915035045575702e-05,
+      "loss": 1.2903,
+      "step": 63860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9147059791371877e-05,
+      "loss": 1.2666,
+      "step": 63870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.914376912698674e-05,
+      "loss": 1.4185,
+      "step": 63880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.91404784626016e-05,
+      "loss": 1.1911,
+      "step": 63890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9137187798216457e-05,
+      "loss": 1.1646,
+      "step": 63900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.913389713383132e-05,
+      "loss": 1.2463,
+      "step": 63910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.913060646944618e-05,
+      "loss": 1.2897,
+      "step": 63920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9127315805061044e-05,
+      "loss": 1.4064,
+      "step": 63930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9124025140675902e-05,
+      "loss": 1.2671,
+      "step": 63940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9120734476290763e-05,
+      "loss": 1.2018,
+      "step": 63950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9117443811905625e-05,
+      "loss": 1.2216,
+      "step": 63960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9114153147520486e-05,
+      "loss": 1.2596,
+      "step": 63970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9110862483135347e-05,
+      "loss": 1.4195,
+      "step": 63980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9107571818750205e-05,
+      "loss": 1.1869,
+      "step": 63990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9104281154365066e-05,
+      "loss": 1.1956,
+      "step": 64000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9100990489979928e-05,
+      "loss": 1.2084,
+      "step": 64010
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.909769982559479e-05,
+      "loss": 1.3821,
+      "step": 64020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.909440916120965e-05,
+      "loss": 1.1946,
+      "step": 64030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.909111849682451e-05,
+      "loss": 1.277,
+      "step": 64040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.908782783243937e-05,
+      "loss": 1.3602,
+      "step": 64050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.908453716805423e-05,
+      "loss": 1.3972,
+      "step": 64060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9081246503669092e-05,
+      "loss": 1.3551,
+      "step": 64070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9077955839283953e-05,
+      "loss": 1.2989,
+      "step": 64080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9074665174898814e-05,
+      "loss": 1.2842,
+      "step": 64090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9071374510513672e-05,
+      "loss": 1.2463,
+      "step": 64100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9068083846128537e-05,
+      "loss": 1.2354,
+      "step": 64110
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9064793181743395e-05,
+      "loss": 1.4263,
+      "step": 64120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.906150251735826e-05,
+      "loss": 1.2641,
+      "step": 64130
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9058211852973117e-05,
+      "loss": 1.2904,
+      "step": 64140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9054921188587975e-05,
+      "loss": 1.1441,
+      "step": 64150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.905163052420284e-05,
+      "loss": 1.2232,
+      "step": 64160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9048339859817698e-05,
+      "loss": 1.1864,
+      "step": 64170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9045049195432562e-05,
+      "loss": 1.2404,
+      "step": 64180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.904175853104742e-05,
+      "loss": 1.2913,
+      "step": 64190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9038467866662278e-05,
+      "loss": 1.2759,
+      "step": 64200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9035177202277143e-05,
+      "loss": 1.3035,
+      "step": 64210
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9031886537892e-05,
+      "loss": 1.2918,
+      "step": 64220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9028595873506865e-05,
+      "loss": 1.1831,
+      "step": 64230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9025305209121723e-05,
+      "loss": 1.2414,
+      "step": 64240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.902201454473658e-05,
+      "loss": 1.3776,
+      "step": 64250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9018723880351446e-05,
+      "loss": 1.2725,
+      "step": 64260
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9015433215966304e-05,
+      "loss": 1.3727,
+      "step": 64270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9012142551581168e-05,
+      "loss": 1.2695,
+      "step": 64280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9008851887196026e-05,
+      "loss": 1.3627,
+      "step": 64290
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9005561222810884e-05,
+      "loss": 1.2925,
+      "step": 64300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.900227055842575e-05,
+      "loss": 1.1875,
+      "step": 64310
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8998979894040607e-05,
+      "loss": 1.33,
+      "step": 64320
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.899568922965547e-05,
+      "loss": 1.1841,
+      "step": 64330
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.899239856527033e-05,
+      "loss": 1.2661,
+      "step": 64340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8989107900885187e-05,
+      "loss": 1.2552,
+      "step": 64350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.898581723650005e-05,
+      "loss": 1.3471,
+      "step": 64360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.898252657211491e-05,
+      "loss": 1.1373,
+      "step": 64370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8979235907729774e-05,
+      "loss": 1.2437,
+      "step": 64380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8975945243344632e-05,
+      "loss": 1.1177,
+      "step": 64390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.897265457895949e-05,
+      "loss": 1.15,
+      "step": 64400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8969363914574355e-05,
+      "loss": 1.4009,
+      "step": 64410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8966073250189212e-05,
+      "loss": 1.4203,
+      "step": 64420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8962782585804077e-05,
+      "loss": 1.2404,
+      "step": 64430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8959491921418935e-05,
+      "loss": 1.2557,
+      "step": 64440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.89562012570338e-05,
+      "loss": 1.2662,
+      "step": 64450
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8952910592648658e-05,
+      "loss": 1.2778,
+      "step": 64460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8949619928263515e-05,
+      "loss": 1.0688,
+      "step": 64470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.894632926387838e-05,
+      "loss": 1.4601,
+      "step": 64480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8943038599493238e-05,
+      "loss": 1.3774,
+      "step": 64490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8939747935108103e-05,
+      "loss": 1.1955,
+      "step": 64500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.893645727072296e-05,
+      "loss": 1.1758,
+      "step": 64510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.893316660633782e-05,
+      "loss": 1.141,
+      "step": 64520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8929875941952683e-05,
+      "loss": 1.1872,
+      "step": 64530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.892658527756754e-05,
+      "loss": 1.364,
+      "step": 64540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8923294613182406e-05,
+      "loss": 1.5043,
+      "step": 64550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8920003948797263e-05,
+      "loss": 1.1793,
+      "step": 64560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.891671328441212e-05,
+      "loss": 1.2955,
+      "step": 64570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8913422620026986e-05,
+      "loss": 1.386,
+      "step": 64580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8910131955641844e-05,
+      "loss": 1.2198,
+      "step": 64590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.890684129125671e-05,
+      "loss": 1.3444,
+      "step": 64600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8903550626871566e-05,
+      "loss": 1.3415,
+      "step": 64610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8900259962486424e-05,
+      "loss": 1.2083,
+      "step": 64620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.889696929810129e-05,
+      "loss": 1.4249,
+      "step": 64630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8893678633716147e-05,
+      "loss": 1.4229,
+      "step": 64640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.889038796933101e-05,
+      "loss": 1.2038,
+      "step": 64650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.888709730494587e-05,
+      "loss": 1.1386,
+      "step": 64660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8883806640560727e-05,
+      "loss": 1.3178,
+      "step": 64670
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8880515976175592e-05,
+      "loss": 1.2084,
+      "step": 64680
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.887722531179045e-05,
+      "loss": 1.2505,
+      "step": 64690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8873934647405314e-05,
+      "loss": 1.2507,
+      "step": 64700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8870643983020172e-05,
+      "loss": 1.3959,
+      "step": 64710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8867353318635034e-05,
+      "loss": 1.2221,
+      "step": 64720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8864062654249895e-05,
+      "loss": 1.315,
+      "step": 64730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8860771989864756e-05,
+      "loss": 1.3078,
+      "step": 64740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8857481325479617e-05,
+      "loss": 1.2356,
+      "step": 64750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8854190661094475e-05,
+      "loss": 1.2734,
+      "step": 64760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8850899996709336e-05,
+      "loss": 1.1275,
+      "step": 64770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8847609332324198e-05,
+      "loss": 1.3865,
+      "step": 64780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.884431866793906e-05,
+      "loss": 1.4617,
+      "step": 64790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.884102800355392e-05,
+      "loss": 1.2723,
+      "step": 64800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.883773733916878e-05,
+      "loss": 1.2654,
+      "step": 64810
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.883444667478364e-05,
+      "loss": 1.329,
+      "step": 64820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.88311560103985e-05,
+      "loss": 1.18,
+      "step": 64830
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8827865346013362e-05,
+      "loss": 1.188,
+      "step": 64840
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8824574681628223e-05,
+      "loss": 1.2011,
+      "step": 64850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8821284017243085e-05,
+      "loss": 1.2416,
+      "step": 64860
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8817993352857942e-05,
+      "loss": 1.2434,
+      "step": 64870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8814702688472807e-05,
+      "loss": 1.1429,
+      "step": 64880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8811412024087665e-05,
+      "loss": 1.2723,
+      "step": 64890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.880812135970253e-05,
+      "loss": 1.2843,
+      "step": 64900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8804830695317387e-05,
+      "loss": 1.2259,
+      "step": 64910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8801540030932245e-05,
+      "loss": 1.2461,
+      "step": 64920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.879824936654711e-05,
+      "loss": 1.4098,
+      "step": 64930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8794958702161968e-05,
+      "loss": 1.1975,
+      "step": 64940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8791668037776833e-05,
+      "loss": 1.2464,
+      "step": 64950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.878837737339169e-05,
+      "loss": 1.2115,
+      "step": 64960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8785086709006548e-05,
+      "loss": 1.2246,
+      "step": 64970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8781796044621413e-05,
+      "loss": 1.1136,
+      "step": 64980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.877850538023627e-05,
+      "loss": 1.1524,
+      "step": 64990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8775214715851135e-05,
+      "loss": 1.2651,
+      "step": 65000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8771924051465993e-05,
+      "loss": 1.225,
+      "step": 65010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.876863338708085e-05,
+      "loss": 1.2458,
+      "step": 65020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8765342722695716e-05,
+      "loss": 1.2391,
+      "step": 65030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8762052058310574e-05,
+      "loss": 1.3239,
+      "step": 65040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.875876139392544e-05,
+      "loss": 1.4145,
+      "step": 65050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8755470729540296e-05,
+      "loss": 1.1834,
+      "step": 65060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8752180065155154e-05,
+      "loss": 1.2805,
+      "step": 65070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.874888940077002e-05,
+      "loss": 1.3367,
+      "step": 65080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8745598736384877e-05,
+      "loss": 1.1516,
+      "step": 65090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.874230807199974e-05,
+      "loss": 1.2526,
+      "step": 65100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.87390174076146e-05,
+      "loss": 1.3391,
+      "step": 65110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8735726743229457e-05,
+      "loss": 1.2773,
+      "step": 65120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8732436078844322e-05,
+      "loss": 1.0957,
+      "step": 65130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.872914541445918e-05,
+      "loss": 1.1941,
+      "step": 65140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8725854750074044e-05,
+      "loss": 1.3248,
+      "step": 65150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8722564085688902e-05,
+      "loss": 1.3331,
+      "step": 65160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.871927342130376e-05,
+      "loss": 1.3704,
+      "step": 65170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8715982756918625e-05,
+      "loss": 1.3243,
+      "step": 65180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8712692092533483e-05,
+      "loss": 1.1743,
+      "step": 65190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8709401428148347e-05,
+      "loss": 1.1158,
+      "step": 65200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8706110763763205e-05,
+      "loss": 1.3667,
+      "step": 65210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8702820099378063e-05,
+      "loss": 1.2876,
+      "step": 65220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8699529434992928e-05,
+      "loss": 1.2795,
+      "step": 65230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8696238770607786e-05,
+      "loss": 1.3439,
+      "step": 65240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.869294810622265e-05,
+      "loss": 1.3082,
+      "step": 65250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8689657441837508e-05,
+      "loss": 1.3836,
+      "step": 65260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8686366777452366e-05,
+      "loss": 1.1041,
+      "step": 65270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.868307611306723e-05,
+      "loss": 1.4612,
+      "step": 65280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.867978544868209e-05,
+      "loss": 1.3502,
+      "step": 65290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8676494784296953e-05,
+      "loss": 1.3294,
+      "step": 65300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.867320411991181e-05,
+      "loss": 1.3258,
+      "step": 65310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.866991345552667e-05,
+      "loss": 1.2739,
+      "step": 65320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8666622791141534e-05,
+      "loss": 1.2292,
+      "step": 65330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.866333212675639e-05,
+      "loss": 1.4037,
+      "step": 65340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8660041462371256e-05,
+      "loss": 1.3379,
+      "step": 65350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8656750797986114e-05,
+      "loss": 1.1742,
+      "step": 65360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8653460133600972e-05,
+      "loss": 1.3202,
+      "step": 65370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8650169469215837e-05,
+      "loss": 1.3052,
+      "step": 65380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8646878804830694e-05,
+      "loss": 1.1729,
+      "step": 65390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.864358814044556e-05,
+      "loss": 1.239,
+      "step": 65400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8640297476060417e-05,
+      "loss": 1.2397,
+      "step": 65410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8637006811675278e-05,
+      "loss": 1.3244,
+      "step": 65420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.863371614729014e-05,
+      "loss": 1.3634,
+      "step": 65430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8630425482904997e-05,
+      "loss": 1.3209,
+      "step": 65440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8627134818519862e-05,
+      "loss": 1.5059,
+      "step": 65450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.862384415413472e-05,
+      "loss": 1.3649,
+      "step": 65460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.862055348974958e-05,
+      "loss": 1.3829,
+      "step": 65470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8617262825364442e-05,
+      "loss": 1.2372,
+      "step": 65480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8613972160979304e-05,
+      "loss": 1.2673,
+      "step": 65490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8610681496594165e-05,
+      "loss": 1.188,
+      "step": 65500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8607390832209023e-05,
+      "loss": 1.2318,
+      "step": 65510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8604100167823884e-05,
+      "loss": 1.3177,
+      "step": 65520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8600809503438745e-05,
+      "loss": 1.1738,
+      "step": 65530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8597518839053607e-05,
+      "loss": 1.2966,
+      "step": 65540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8594228174668468e-05,
+      "loss": 1.2783,
+      "step": 65550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.859093751028333e-05,
+      "loss": 1.2842,
+      "step": 65560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8587646845898187e-05,
+      "loss": 1.1933,
+      "step": 65570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8584356181513052e-05,
+      "loss": 1.3159,
+      "step": 65580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.858106551712791e-05,
+      "loss": 1.2964,
+      "step": 65590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.857777485274277e-05,
+      "loss": 1.0174,
+      "step": 65600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8574484188357632e-05,
+      "loss": 1.1883,
+      "step": 65610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.857119352397249e-05,
+      "loss": 1.2257,
+      "step": 65620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8567902859587355e-05,
+      "loss": 1.1976,
+      "step": 65630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8564612195202213e-05,
+      "loss": 1.1868,
+      "step": 65640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8561321530817077e-05,
+      "loss": 1.175,
+      "step": 65650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8558030866431935e-05,
+      "loss": 1.2881,
+      "step": 65660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8554740202046793e-05,
+      "loss": 1.3304,
+      "step": 65670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8551449537661658e-05,
+      "loss": 1.2409,
+      "step": 65680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8548158873276515e-05,
+      "loss": 1.2345,
+      "step": 65690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.854486820889138e-05,
+      "loss": 1.2705,
+      "step": 65700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8541577544506238e-05,
+      "loss": 1.4621,
+      "step": 65710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8538286880121096e-05,
+      "loss": 1.3144,
+      "step": 65720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.853499621573596e-05,
+      "loss": 1.1425,
+      "step": 65730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.853170555135082e-05,
+      "loss": 1.29,
+      "step": 65740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8528414886965683e-05,
+      "loss": 1.1272,
+      "step": 65750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.852512422258054e-05,
+      "loss": 1.2754,
+      "step": 65760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.85218335581954e-05,
+      "loss": 1.3792,
+      "step": 65770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8518542893810264e-05,
+      "loss": 1.2636,
+      "step": 65780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.851525222942512e-05,
+      "loss": 1.4372,
+      "step": 65790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8511961565039986e-05,
+      "loss": 1.1878,
+      "step": 65800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8508670900654844e-05,
+      "loss": 1.1992,
+      "step": 65810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8505380236269702e-05,
+      "loss": 1.303,
+      "step": 65820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8502089571884566e-05,
+      "loss": 1.1603,
+      "step": 65830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8498798907499424e-05,
+      "loss": 1.2419,
+      "step": 65840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.849550824311429e-05,
+      "loss": 1.388,
+      "step": 65850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8492217578729147e-05,
+      "loss": 1.3403,
+      "step": 65860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8488926914344005e-05,
+      "loss": 1.293,
+      "step": 65870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.848563624995887e-05,
+      "loss": 1.234,
+      "step": 65880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8482345585573727e-05,
+      "loss": 1.1899,
+      "step": 65890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8479054921188592e-05,
+      "loss": 1.4034,
+      "step": 65900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.847576425680345e-05,
+      "loss": 1.1596,
+      "step": 65910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8472473592418308e-05,
+      "loss": 1.3212,
+      "step": 65920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8469182928033172e-05,
+      "loss": 1.3216,
+      "step": 65930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.846589226364803e-05,
+      "loss": 1.3668,
+      "step": 65940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8462601599262895e-05,
+      "loss": 1.265,
+      "step": 65950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8459310934877753e-05,
+      "loss": 1.2528,
+      "step": 65960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.845602027049261e-05,
+      "loss": 1.2571,
+      "step": 65970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8452729606107475e-05,
+      "loss": 1.2437,
+      "step": 65980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8449438941722333e-05,
+      "loss": 1.3179,
+      "step": 65990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8446148277337198e-05,
+      "loss": 1.194,
+      "step": 66000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8442857612952056e-05,
+      "loss": 1.2229,
+      "step": 66010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8439566948566914e-05,
+      "loss": 1.1801,
+      "step": 66020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8436276284181778e-05,
+      "loss": 1.2399,
+      "step": 66030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8432985619796636e-05,
+      "loss": 1.1992,
+      "step": 66040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.84296949554115e-05,
+      "loss": 1.2057,
+      "step": 66050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.842640429102636e-05,
+      "loss": 1.3109,
+      "step": 66060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8423113626641217e-05,
+      "loss": 1.305,
+      "step": 66070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.841982296225608e-05,
+      "loss": 1.1655,
+      "step": 66080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.841653229787094e-05,
+      "loss": 1.2315,
+      "step": 66090
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8413241633485804e-05,
+      "loss": 1.0172,
+      "step": 66100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.840995096910066e-05,
+      "loss": 1.4012,
+      "step": 66110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.840666030471552e-05,
+      "loss": 1.1475,
+      "step": 66120
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8403369640330384e-05,
+      "loss": 1.0985,
+      "step": 66130
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8400078975945242e-05,
+      "loss": 1.4572,
+      "step": 66140
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8396788311560107e-05,
+      "loss": 1.4522,
+      "step": 66150
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8393497647174965e-05,
+      "loss": 1.3018,
+      "step": 66160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8390206982789826e-05,
+      "loss": 1.0336,
+      "step": 66170
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8386916318404687e-05,
+      "loss": 1.2066,
+      "step": 66180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8383625654019545e-05,
+      "loss": 1.2352,
+      "step": 66190
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.838033498963441e-05,
+      "loss": 1.3336,
+      "step": 66200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8377044325249268e-05,
+      "loss": 1.2166,
+      "step": 66210
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.837375366086413e-05,
+      "loss": 1.2381,
+      "step": 66220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.837046299647899e-05,
+      "loss": 1.3134,
+      "step": 66230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.836717233209385e-05,
+      "loss": 1.2393,
+      "step": 66240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8363881667708713e-05,
+      "loss": 1.2672,
+      "step": 66250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8360591003323574e-05,
+      "loss": 1.2854,
+      "step": 66260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8357300338938432e-05,
+      "loss": 1.427,
+      "step": 66270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8354009674553293e-05,
+      "loss": 1.3284,
+      "step": 66280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8350719010168154e-05,
+      "loss": 1.2104,
+      "step": 66290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8347428345783016e-05,
+      "loss": 1.2084,
+      "step": 66300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8344137681397877e-05,
+      "loss": 1.4865,
+      "step": 66310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8340847017012735e-05,
+      "loss": 1.2081,
+      "step": 66320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.83375563526276e-05,
+      "loss": 1.4077,
+      "step": 66330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8334265688242457e-05,
+      "loss": 1.3114,
+      "step": 66340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.833097502385732e-05,
+      "loss": 1.0811,
+      "step": 66350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.832768435947218e-05,
+      "loss": 1.3845,
+      "step": 66360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8324393695087038e-05,
+      "loss": 1.3246,
+      "step": 66370
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8321103030701902e-05,
+      "loss": 1.2497,
+      "step": 66380
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.831781236631676e-05,
+      "loss": 1.2241,
+      "step": 66390
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8314521701931625e-05,
+      "loss": 1.2878,
+      "step": 66400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8311231037546483e-05,
+      "loss": 1.3017,
+      "step": 66410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.830794037316134e-05,
+      "loss": 1.2997,
+      "step": 66420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8304649708776205e-05,
+      "loss": 1.3804,
+      "step": 66430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8301359044391063e-05,
+      "loss": 1.3864,
+      "step": 66440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8298068380005928e-05,
+      "loss": 1.3268,
+      "step": 66450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8294777715620786e-05,
+      "loss": 1.1547,
+      "step": 66460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8291487051235644e-05,
+      "loss": 1.2185,
+      "step": 66470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8288196386850508e-05,
+      "loss": 1.1584,
+      "step": 66480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8284905722465366e-05,
+      "loss": 1.375,
+      "step": 66490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.828161505808023e-05,
+      "loss": 1.3404,
+      "step": 66500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.827832439369509e-05,
+      "loss": 1.1816,
+      "step": 66510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8275033729309946e-05,
+      "loss": 1.3199,
+      "step": 66520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.827174306492481e-05,
+      "loss": 1.2959,
+      "step": 66530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.826845240053967e-05,
+      "loss": 1.297,
+      "step": 66540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8265161736154534e-05,
+      "loss": 1.2864,
+      "step": 66550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.826187107176939e-05,
+      "loss": 1.2046,
+      "step": 66560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.825858040738425e-05,
+      "loss": 1.3759,
+      "step": 66570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8255289742999114e-05,
+      "loss": 1.2971,
+      "step": 66580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8251999078613972e-05,
+      "loss": 1.3223,
+      "step": 66590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8248708414228837e-05,
+      "loss": 1.242,
+      "step": 66600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8245417749843694e-05,
+      "loss": 1.3647,
+      "step": 66610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8242127085458552e-05,
+      "loss": 1.2642,
+      "step": 66620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8238836421073417e-05,
+      "loss": 1.2182,
+      "step": 66630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8235545756688275e-05,
+      "loss": 1.2064,
+      "step": 66640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.823225509230314e-05,
+      "loss": 1.2189,
+      "step": 66650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8228964427917997e-05,
+      "loss": 1.2408,
+      "step": 66660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8225673763532855e-05,
+      "loss": 1.2979,
+      "step": 66670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.822238309914772e-05,
+      "loss": 1.2619,
+      "step": 66680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8219092434762578e-05,
+      "loss": 1.3414,
+      "step": 66690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8215801770377443e-05,
+      "loss": 1.284,
+      "step": 66700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.82125111059923e-05,
+      "loss": 1.3763,
+      "step": 66710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8209220441607158e-05,
+      "loss": 1.1919,
+      "step": 66720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8205929777222023e-05,
+      "loss": 1.1316,
+      "step": 66730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.820263911283688e-05,
+      "loss": 1.3508,
+      "step": 66740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8199348448451745e-05,
+      "loss": 1.2007,
+      "step": 66750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8196057784066603e-05,
+      "loss": 1.2449,
+      "step": 66760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.819276711968146e-05,
+      "loss": 1.2581,
+      "step": 66770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8189476455296326e-05,
+      "loss": 1.1909,
+      "step": 66780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8186185790911184e-05,
+      "loss": 1.2829,
+      "step": 66790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.818289512652605e-05,
+      "loss": 1.3378,
+      "step": 66800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8179604462140906e-05,
+      "loss": 1.2802,
+      "step": 66810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8176313797755764e-05,
+      "loss": 1.3534,
+      "step": 66820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.817302313337063e-05,
+      "loss": 1.2107,
+      "step": 66830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8169732468985487e-05,
+      "loss": 1.1356,
+      "step": 66840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.816644180460035e-05,
+      "loss": 1.2834,
+      "step": 66850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.816315114021521e-05,
+      "loss": 1.2228,
+      "step": 66860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8159860475830067e-05,
+      "loss": 1.2229,
+      "step": 66870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8156569811444932e-05,
+      "loss": 1.4293,
+      "step": 66880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.815327914705979e-05,
+      "loss": 1.2631,
+      "step": 66890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8149988482674654e-05,
+      "loss": 1.3322,
+      "step": 66900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8146697818289512e-05,
+      "loss": 1.0907,
+      "step": 66910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8143407153904373e-05,
+      "loss": 1.3117,
+      "step": 66920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8140116489519235e-05,
+      "loss": 1.2801,
+      "step": 66930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8136825825134096e-05,
+      "loss": 1.3327,
+      "step": 66940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8133535160748957e-05,
+      "loss": 1.1937,
+      "step": 66950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8130244496363815e-05,
+      "loss": 1.2707,
+      "step": 66960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8126953831978676e-05,
+      "loss": 1.3832,
+      "step": 66970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8123663167593538e-05,
+      "loss": 1.3393,
+      "step": 66980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.81203725032084e-05,
+      "loss": 1.2218,
+      "step": 66990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.811708183882326e-05,
+      "loss": 1.1998,
+      "step": 67000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.811379117443812e-05,
+      "loss": 1.2844,
+      "step": 67010
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.811050051005298e-05,
+      "loss": 1.2338,
+      "step": 67020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.810720984566784e-05,
+      "loss": 1.2806,
+      "step": 67030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8103919181282702e-05,
+      "loss": 1.2461,
+      "step": 67040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8100628516897563e-05,
+      "loss": 1.3782,
+      "step": 67050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8097337852512424e-05,
+      "loss": 1.1156,
+      "step": 67060
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8094047188127282e-05,
+      "loss": 1.4721,
+      "step": 67070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8090756523742147e-05,
+      "loss": 1.2725,
+      "step": 67080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8087465859357005e-05,
+      "loss": 1.2797,
+      "step": 67090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.808417519497187e-05,
+      "loss": 1.2504,
+      "step": 67100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8080884530586727e-05,
+      "loss": 1.3153,
+      "step": 67110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8077593866201585e-05,
+      "loss": 1.4799,
+      "step": 67120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.807430320181645e-05,
+      "loss": 1.2366,
+      "step": 67130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8071012537431308e-05,
+      "loss": 1.2011,
+      "step": 67140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8067721873046172e-05,
+      "loss": 1.203,
+      "step": 67150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.806443120866103e-05,
+      "loss": 1.1549,
+      "step": 67160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8061140544275888e-05,
+      "loss": 1.2175,
+      "step": 67170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8057849879890753e-05,
+      "loss": 1.3017,
+      "step": 67180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.805455921550561e-05,
+      "loss": 1.3274,
+      "step": 67190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8051268551120475e-05,
+      "loss": 1.2415,
+      "step": 67200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8047977886735333e-05,
+      "loss": 1.276,
+      "step": 67210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.804468722235019e-05,
+      "loss": 1.2867,
+      "step": 67220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8041396557965056e-05,
+      "loss": 1.1057,
+      "step": 67230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8038105893579914e-05,
+      "loss": 1.255,
+      "step": 67240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.803481522919478e-05,
+      "loss": 1.1691,
+      "step": 67250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8031524564809636e-05,
+      "loss": 1.1981,
+      "step": 67260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.80282339004245e-05,
+      "loss": 1.2391,
+      "step": 67270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.802494323603936e-05,
+      "loss": 1.2582,
+      "step": 67280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8021652571654217e-05,
+      "loss": 1.2435,
+      "step": 67290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.801836190726908e-05,
+      "loss": 1.127,
+      "step": 67300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.801507124288394e-05,
+      "loss": 1.277,
+      "step": 67310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8011780578498804e-05,
+      "loss": 1.2556,
+      "step": 67320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8008489914113662e-05,
+      "loss": 1.1863,
+      "step": 67330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.800519924972852e-05,
+      "loss": 1.2359,
+      "step": 67340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8001908585343384e-05,
+      "loss": 1.4301,
+      "step": 67350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7998617920958242e-05,
+      "loss": 1.2417,
+      "step": 67360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7995327256573107e-05,
+      "loss": 1.4345,
+      "step": 67370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7992036592187965e-05,
+      "loss": 1.3368,
+      "step": 67380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7988745927802823e-05,
+      "loss": 1.2801,
+      "step": 67390
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7985455263417687e-05,
+      "loss": 1.3537,
+      "step": 67400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7982164599032545e-05,
+      "loss": 1.3165,
+      "step": 67410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.797887393464741e-05,
+      "loss": 1.3376,
+      "step": 67420
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7975583270262268e-05,
+      "loss": 1.1686,
+      "step": 67430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7972292605877125e-05,
+      "loss": 1.1646,
+      "step": 67440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.796900194149199e-05,
+      "loss": 1.3906,
+      "step": 67450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7965711277106848e-05,
+      "loss": 1.6261,
+      "step": 67460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7962420612721713e-05,
+      "loss": 1.2988,
+      "step": 67470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.795912994833657e-05,
+      "loss": 1.1829,
+      "step": 67480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.795583928395143e-05,
+      "loss": 1.3147,
+      "step": 67490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7952548619566293e-05,
+      "loss": 1.2561,
+      "step": 67500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.794925795518115e-05,
+      "loss": 1.3687,
+      "step": 67510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7945967290796016e-05,
+      "loss": 1.2599,
+      "step": 67520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7942676626410874e-05,
+      "loss": 1.2984,
+      "step": 67530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.793938596202573e-05,
+      "loss": 1.2147,
+      "step": 67540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7936095297640596e-05,
+      "loss": 1.1396,
+      "step": 67550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7932804633255454e-05,
+      "loss": 1.4524,
+      "step": 67560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.792951396887032e-05,
+      "loss": 1.2344,
+      "step": 67570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7926223304485176e-05,
+      "loss": 1.3581,
+      "step": 67580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7922932640100034e-05,
+      "loss": 1.3131,
+      "step": 67590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.79196419757149e-05,
+      "loss": 1.3983,
+      "step": 67600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7916351311329757e-05,
+      "loss": 1.2287,
+      "step": 67610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.791306064694462e-05,
+      "loss": 1.1358,
+      "step": 67620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.790976998255948e-05,
+      "loss": 1.2801,
+      "step": 67630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7906479318174337e-05,
+      "loss": 1.1172,
+      "step": 67640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7903188653789202e-05,
+      "loss": 1.2934,
+      "step": 67650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.789989798940406e-05,
+      "loss": 1.2535,
+      "step": 67660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7896607325018924e-05,
+      "loss": 1.2216,
+      "step": 67670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7893316660633782e-05,
+      "loss": 1.4892,
+      "step": 67680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7890025996248644e-05,
+      "loss": 1.2667,
+      "step": 67690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7886735331863505e-05,
+      "loss": 1.3887,
+      "step": 67700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7883444667478366e-05,
+      "loss": 1.2138,
+      "step": 67710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7880154003093227e-05,
+      "loss": 1.2828,
+      "step": 67720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7876863338708085e-05,
+      "loss": 1.2849,
+      "step": 67730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7873572674322947e-05,
+      "loss": 1.3431,
+      "step": 67740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7870282009937808e-05,
+      "loss": 1.335,
+      "step": 67750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.786699134555267e-05,
+      "loss": 1.3147,
+      "step": 67760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.786370068116753e-05,
+      "loss": 1.1487,
+      "step": 67770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.786041001678239e-05,
+      "loss": 1.1244,
+      "step": 67780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.785711935239725e-05,
+      "loss": 1.3609,
+      "step": 67790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.785382868801211e-05,
+      "loss": 1.2018,
+      "step": 67800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7850538023626972e-05,
+      "loss": 1.1975,
+      "step": 67810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7847247359241833e-05,
+      "loss": 1.3025,
+      "step": 67820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7843956694856695e-05,
+      "loss": 1.2555,
+      "step": 67830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7840666030471552e-05,
+      "loss": 1.2102,
+      "step": 67840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7837375366086417e-05,
+      "loss": 1.2875,
+      "step": 67850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7834084701701275e-05,
+      "loss": 1.3559,
+      "step": 67860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.783079403731614e-05,
+      "loss": 1.293,
+      "step": 67870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7827503372930998e-05,
+      "loss": 1.3149,
+      "step": 67880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7824212708545855e-05,
+      "loss": 1.2815,
+      "step": 67890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.782092204416072e-05,
+      "loss": 1.3045,
+      "step": 67900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7817631379775578e-05,
+      "loss": 1.0677,
+      "step": 67910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7814340715390443e-05,
+      "loss": 1.2231,
+      "step": 67920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.78110500510053e-05,
+      "loss": 1.4161,
+      "step": 67930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.780775938662016e-05,
+      "loss": 1.2961,
+      "step": 67940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7804468722235023e-05,
+      "loss": 1.2683,
+      "step": 67950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.780117805784988e-05,
+      "loss": 1.5166,
+      "step": 67960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7797887393464746e-05,
+      "loss": 1.3094,
+      "step": 67970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7794596729079603e-05,
+      "loss": 1.3345,
+      "step": 67980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.779130606469446e-05,
+      "loss": 1.196,
+      "step": 67990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7788015400309326e-05,
+      "loss": 1.2987,
+      "step": 68000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7784724735924184e-05,
+      "loss": 1.1614,
+      "step": 68010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.778143407153905e-05,
+      "loss": 1.2234,
+      "step": 68020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7778143407153906e-05,
+      "loss": 1.2049,
+      "step": 68030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7774852742768764e-05,
+      "loss": 1.2268,
+      "step": 68040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.777156207838363e-05,
+      "loss": 1.4959,
+      "step": 68050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7768271413998487e-05,
+      "loss": 1.3188,
+      "step": 68060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.776498074961335e-05,
+      "loss": 1.3118,
+      "step": 68070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.776169008522821e-05,
+      "loss": 1.374,
+      "step": 68080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7758399420843067e-05,
+      "loss": 1.3428,
+      "step": 68090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7755108756457932e-05,
+      "loss": 1.2928,
+      "step": 68100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.775181809207279e-05,
+      "loss": 1.3852,
+      "step": 68110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7748527427687654e-05,
+      "loss": 1.48,
+      "step": 68120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7745236763302512e-05,
+      "loss": 1.2162,
+      "step": 68130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.774194609891737e-05,
+      "loss": 1.1652,
+      "step": 68140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7738655434532235e-05,
+      "loss": 1.2577,
+      "step": 68150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7735364770147093e-05,
+      "loss": 1.3125,
+      "step": 68160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7732074105761957e-05,
+      "loss": 1.3653,
+      "step": 68170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7728783441376815e-05,
+      "loss": 1.0925,
+      "step": 68180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7725492776991673e-05,
+      "loss": 1.3039,
+      "step": 68190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7722202112606538e-05,
+      "loss": 1.2331,
+      "step": 68200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7718911448221396e-05,
+      "loss": 1.1708,
+      "step": 68210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.771562078383626e-05,
+      "loss": 1.2093,
+      "step": 68220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7712330119451118e-05,
+      "loss": 1.1086,
+      "step": 68230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7709039455065976e-05,
+      "loss": 1.2197,
+      "step": 68240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.770574879068084e-05,
+      "loss": 1.2694,
+      "step": 68250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.77024581262957e-05,
+      "loss": 1.1846,
+      "step": 68260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7699167461910563e-05,
+      "loss": 1.2284,
+      "step": 68270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.769587679752542e-05,
+      "loss": 1.1605,
+      "step": 68280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.769258613314028e-05,
+      "loss": 1.1739,
+      "step": 68290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7689295468755144e-05,
+      "loss": 1.3088,
+      "step": 68300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.768600480437e-05,
+      "loss": 1.2448,
+      "step": 68310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7682714139984866e-05,
+      "loss": 1.1279,
+      "step": 68320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7679423475599724e-05,
+      "loss": 1.1229,
+      "step": 68330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7676132811214582e-05,
+      "loss": 1.1291,
+      "step": 68340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7672842146829447e-05,
+      "loss": 1.4357,
+      "step": 68350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7669551482444304e-05,
+      "loss": 1.1542,
+      "step": 68360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.766626081805917e-05,
+      "loss": 1.1835,
+      "step": 68370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7662970153674027e-05,
+      "loss": 1.2095,
+      "step": 68380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7659679489288888e-05,
+      "loss": 1.1757,
+      "step": 68390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.765638882490375e-05,
+      "loss": 1.2588,
+      "step": 68400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7653098160518607e-05,
+      "loss": 1.6506,
+      "step": 68410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7649807496133472e-05,
+      "loss": 1.2072,
+      "step": 68420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.764651683174833e-05,
+      "loss": 1.1169,
+      "step": 68430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.764322616736319e-05,
+      "loss": 1.4721,
+      "step": 68440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7639935502978053e-05,
+      "loss": 1.3896,
+      "step": 68450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7636644838592914e-05,
+      "loss": 1.3738,
+      "step": 68460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7633354174207775e-05,
+      "loss": 1.3037,
+      "step": 68470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7630063509822633e-05,
+      "loss": 1.2871,
+      "step": 68480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7626772845437494e-05,
+      "loss": 1.302,
+      "step": 68490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7623482181052355e-05,
+      "loss": 1.2509,
+      "step": 68500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7620191516667217e-05,
+      "loss": 1.1528,
+      "step": 68510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7616900852282078e-05,
+      "loss": 1.3056,
+      "step": 68520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.761361018789694e-05,
+      "loss": 1.7011,
+      "step": 68530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7610319523511797e-05,
+      "loss": 1.31,
+      "step": 68540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7607028859126662e-05,
+      "loss": 1.4535,
+      "step": 68550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.760373819474152e-05,
+      "loss": 1.3432,
+      "step": 68560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.760044753035638e-05,
+      "loss": 1.1579,
+      "step": 68570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7597156865971242e-05,
+      "loss": 1.324,
+      "step": 68580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.75938662015861e-05,
+      "loss": 1.3143,
+      "step": 68590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7590575537200965e-05,
+      "loss": 1.0735,
+      "step": 68600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7587284872815823e-05,
+      "loss": 1.3547,
+      "step": 68610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7583994208430687e-05,
+      "loss": 1.2504,
+      "step": 68620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7580703544045545e-05,
+      "loss": 1.3434,
+      "step": 68630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7577412879660403e-05,
+      "loss": 1.2781,
+      "step": 68640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7574122215275268e-05,
+      "loss": 1.0957,
+      "step": 68650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7570831550890126e-05,
+      "loss": 1.2694,
+      "step": 68660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.756754088650499e-05,
+      "loss": 1.2425,
+      "step": 68670
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7564250222119848e-05,
+      "loss": 1.2393,
+      "step": 68680
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7560959557734706e-05,
+      "loss": 1.1722,
+      "step": 68690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.755766889334957e-05,
+      "loss": 1.491,
+      "step": 68700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.755437822896443e-05,
+      "loss": 1.1293,
+      "step": 68710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7551087564579293e-05,
+      "loss": 1.3637,
+      "step": 68720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.754779690019415e-05,
+      "loss": 1.2879,
+      "step": 68730
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.754450623580901e-05,
+      "loss": 1.1421,
+      "step": 68740
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7541215571423874e-05,
+      "loss": 1.2652,
+      "step": 68750
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.753792490703873e-05,
+      "loss": 1.1632,
+      "step": 68760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7534634242653596e-05,
+      "loss": 1.3865,
+      "step": 68770
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7531343578268454e-05,
+      "loss": 1.3669,
+      "step": 68780
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7528052913883312e-05,
+      "loss": 1.4115,
+      "step": 68790
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7524762249498177e-05,
+      "loss": 1.2815,
+      "step": 68800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7521471585113034e-05,
+      "loss": 1.2626,
+      "step": 68810
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.75181809207279e-05,
+      "loss": 1.1931,
+      "step": 68820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7514890256342757e-05,
+      "loss": 1.2436,
+      "step": 68830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7511599591957615e-05,
+      "loss": 1.2916,
+      "step": 68840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.750830892757248e-05,
+      "loss": 1.4657,
+      "step": 68850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7505018263187337e-05,
+      "loss": 1.2922,
+      "step": 68860
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7501727598802202e-05,
+      "loss": 1.2784,
+      "step": 68870
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.749843693441706e-05,
+      "loss": 1.3447,
+      "step": 68880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7495146270031918e-05,
+      "loss": 1.249,
+      "step": 68890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7491855605646782e-05,
+      "loss": 1.24,
+      "step": 68900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.748856494126164e-05,
+      "loss": 1.2514,
+      "step": 68910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7485274276876505e-05,
+      "loss": 1.2195,
+      "step": 68920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7481983612491363e-05,
+      "loss": 1.3301,
+      "step": 68930
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.747869294810622e-05,
+      "loss": 1.2777,
+      "step": 68940
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7475402283721085e-05,
+      "loss": 1.5927,
+      "step": 68950
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7472111619335943e-05,
+      "loss": 1.2746,
+      "step": 68960
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7468820954950808e-05,
+      "loss": 1.1876,
+      "step": 68970
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7465530290565666e-05,
+      "loss": 1.1721,
+      "step": 68980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7462239626180524e-05,
+      "loss": 1.2042,
+      "step": 68990
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.745894896179539e-05,
+      "loss": 1.3189,
+      "step": 69000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7455658297410246e-05,
+      "loss": 1.4018,
+      "step": 69010
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.745236763302511e-05,
+      "loss": 1.4108,
+      "step": 69020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.744907696863997e-05,
+      "loss": 1.337,
+      "step": 69030
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7445786304254827e-05,
+      "loss": 1.1445,
+      "step": 69040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.744249563986969e-05,
+      "loss": 1.2827,
+      "step": 69050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.743920497548455e-05,
+      "loss": 1.2692,
+      "step": 69060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7435914311099414e-05,
+      "loss": 1.3121,
+      "step": 69070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7432623646714272e-05,
+      "loss": 1.2211,
+      "step": 69080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.742933298232913e-05,
+      "loss": 1.0857,
+      "step": 69090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7426042317943994e-05,
+      "loss": 1.3432,
+      "step": 69100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7422751653558852e-05,
+      "loss": 1.2778,
+      "step": 69110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7419460989173717e-05,
+      "loss": 1.2572,
+      "step": 69120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7416170324788575e-05,
+      "loss": 1.1844,
+      "step": 69130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7412879660403436e-05,
+      "loss": 1.3307,
+      "step": 69140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7409588996018297e-05,
+      "loss": 1.1505,
+      "step": 69150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7406298331633155e-05,
+      "loss": 1.244,
+      "step": 69160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.740300766724802e-05,
+      "loss": 1.3217,
+      "step": 69170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7399717002862878e-05,
+      "loss": 1.3628,
+      "step": 69180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.739642633847774e-05,
+      "loss": 1.2099,
+      "step": 69190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.73931356740926e-05,
+      "loss": 1.1423,
+      "step": 69200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.738984500970746e-05,
+      "loss": 1.3638,
+      "step": 69210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7386554345322323e-05,
+      "loss": 1.2779,
+      "step": 69220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7383263680937184e-05,
+      "loss": 1.3202,
+      "step": 69230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7379973016552042e-05,
+      "loss": 1.3,
+      "step": 69240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7376682352166903e-05,
+      "loss": 1.2113,
+      "step": 69250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7373391687781764e-05,
+      "loss": 1.3485,
+      "step": 69260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7370101023396626e-05,
+      "loss": 1.1764,
+      "step": 69270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7366810359011487e-05,
+      "loss": 1.3124,
+      "step": 69280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7363519694626345e-05,
+      "loss": 1.272,
+      "step": 69290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.736022903024121e-05,
+      "loss": 1.1531,
+      "step": 69300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7356938365856067e-05,
+      "loss": 1.3102,
+      "step": 69310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.735364770147093e-05,
+      "loss": 1.3622,
+      "step": 69320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.735035703708579e-05,
+      "loss": 1.1885,
+      "step": 69330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7347066372700648e-05,
+      "loss": 1.364,
+      "step": 69340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7343775708315512e-05,
+      "loss": 1.3576,
+      "step": 69350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.734048504393037e-05,
+      "loss": 1.4638,
+      "step": 69360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7337194379545235e-05,
+      "loss": 1.3217,
+      "step": 69370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7333903715160093e-05,
+      "loss": 1.2173,
+      "step": 69380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.733061305077495e-05,
+      "loss": 1.3323,
+      "step": 69390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7327322386389815e-05,
+      "loss": 1.3599,
+      "step": 69400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7324031722004673e-05,
+      "loss": 1.2496,
+      "step": 69410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7320741057619538e-05,
+      "loss": 1.3134,
+      "step": 69420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7317450393234396e-05,
+      "loss": 1.2612,
+      "step": 69430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7314159728849254e-05,
+      "loss": 1.3881,
+      "step": 69440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7310869064464118e-05,
+      "loss": 1.2847,
+      "step": 69450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7307578400078976e-05,
+      "loss": 1.2528,
+      "step": 69460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.730428773569384e-05,
+      "loss": 1.1982,
+      "step": 69470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.73009970713087e-05,
+      "loss": 1.2345,
+      "step": 69480
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7297706406923557e-05,
+      "loss": 1.0764,
+      "step": 69490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.729441574253842e-05,
+      "loss": 1.2152,
+      "step": 69500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.729112507815328e-05,
+      "loss": 1.2342,
+      "step": 69510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7287834413768144e-05,
+      "loss": 1.2422,
+      "step": 69520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7284543749383e-05,
+      "loss": 1.3163,
+      "step": 69530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.728125308499786e-05,
+      "loss": 1.3718,
+      "step": 69540
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7277962420612724e-05,
+      "loss": 1.2141,
+      "step": 69550
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7274671756227582e-05,
+      "loss": 1.2679,
+      "step": 69560
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7271381091842447e-05,
+      "loss": 1.1994,
+      "step": 69570
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7268090427457305e-05,
+      "loss": 1.2883,
+      "step": 69580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7264799763072162e-05,
+      "loss": 1.0966,
+      "step": 69590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7261509098687027e-05,
+      "loss": 1.2134,
+      "step": 69600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7258218434301885e-05,
+      "loss": 1.1539,
+      "step": 69610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.725492776991675e-05,
+      "loss": 1.2869,
+      "step": 69620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7251637105531608e-05,
+      "loss": 1.1407,
+      "step": 69630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7248346441146465e-05,
+      "loss": 1.2589,
+      "step": 69640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.724505577676133e-05,
+      "loss": 1.3705,
+      "step": 69650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7241765112376188e-05,
+      "loss": 1.1632,
+      "step": 69660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7238474447991053e-05,
+      "loss": 1.306,
+      "step": 69670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.723518378360591e-05,
+      "loss": 1.3054,
+      "step": 69680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.723189311922077e-05,
+      "loss": 1.2669,
+      "step": 69690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7228602454835633e-05,
+      "loss": 1.4908,
+      "step": 69700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.722531179045049e-05,
+      "loss": 1.251,
+      "step": 69710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7222021126065356e-05,
+      "loss": 1.2813,
+      "step": 69720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7218730461680213e-05,
+      "loss": 1.2901,
+      "step": 69730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.721543979729507e-05,
+      "loss": 1.2285,
+      "step": 69740
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7212149132909936e-05,
+      "loss": 1.2241,
+      "step": 69750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7208858468524794e-05,
+      "loss": 1.423,
+      "step": 69760
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.720556780413966e-05,
+      "loss": 1.2737,
+      "step": 69770
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7202277139754516e-05,
+      "loss": 1.1606,
+      "step": 69780
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7198986475369374e-05,
+      "loss": 1.3127,
+      "step": 69790
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.719569581098424e-05,
+      "loss": 1.2642,
+      "step": 69800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7192405146599097e-05,
+      "loss": 1.161,
+      "step": 69810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.718911448221396e-05,
+      "loss": 1.2375,
+      "step": 69820
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.718582381782882e-05,
+      "loss": 1.4086,
+      "step": 69830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7182533153443677e-05,
+      "loss": 1.3443,
+      "step": 69840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7179242489058542e-05,
+      "loss": 1.3674,
+      "step": 69850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.71759518246734e-05,
+      "loss": 1.3564,
+      "step": 69860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7172661160288264e-05,
+      "loss": 1.3701,
+      "step": 69870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7169370495903122e-05,
+      "loss": 1.2128,
+      "step": 69880
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7166079831517984e-05,
+      "loss": 1.3287,
+      "step": 69890
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7162789167132845e-05,
+      "loss": 1.2657,
+      "step": 69900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7159498502747706e-05,
+      "loss": 1.5814,
+      "step": 69910
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7156207838362567e-05,
+      "loss": 1.3938,
+      "step": 69920
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7152917173977425e-05,
+      "loss": 1.3145,
+      "step": 69930
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7149626509592286e-05,
+      "loss": 1.292,
+      "step": 69940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7146335845207148e-05,
+      "loss": 1.3419,
+      "step": 69950
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.714304518082201e-05,
+      "loss": 1.3076,
+      "step": 69960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.713975451643687e-05,
+      "loss": 1.154,
+      "step": 69970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.713646385205173e-05,
+      "loss": 1.2066,
+      "step": 69980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.713317318766659e-05,
+      "loss": 1.283,
+      "step": 69990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7129882523281454e-05,
+      "loss": 1.3137,
+      "step": 70000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7126591858896312e-05,
+      "loss": 1.3055,
+      "step": 70010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7123301194511173e-05,
+      "loss": 1.4139,
+      "step": 70020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7120010530126035e-05,
+      "loss": 1.3055,
+      "step": 70030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7116719865740892e-05,
+      "loss": 1.2175,
+      "step": 70040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7113429201355757e-05,
+      "loss": 1.2963,
+      "step": 70050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7110138536970615e-05,
+      "loss": 1.3921,
+      "step": 70060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.710684787258548e-05,
+      "loss": 1.0528,
+      "step": 70070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7103557208200337e-05,
+      "loss": 1.1488,
+      "step": 70080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.71002665438152e-05,
+      "loss": 1.1916,
+      "step": 70090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.709697587943006e-05,
+      "loss": 1.1604,
+      "step": 70100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7093685215044918e-05,
+      "loss": 1.3252,
+      "step": 70110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7090394550659783e-05,
+      "loss": 1.2739,
+      "step": 70120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.708710388627464e-05,
+      "loss": 1.2627,
+      "step": 70130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7083813221889505e-05,
+      "loss": 1.154,
+      "step": 70140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7080522557504363e-05,
+      "loss": 1.2096,
+      "step": 70150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.707723189311922e-05,
+      "loss": 1.3445,
+      "step": 70160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7073941228734085e-05,
+      "loss": 1.2773,
+      "step": 70170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7070650564348943e-05,
+      "loss": 1.4242,
+      "step": 70180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7067359899963808e-05,
+      "loss": 1.1139,
+      "step": 70190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7064069235578666e-05,
+      "loss": 1.2098,
+      "step": 70200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7060778571193524e-05,
+      "loss": 1.2539,
+      "step": 70210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.705748790680839e-05,
+      "loss": 1.0327,
+      "step": 70220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7054197242423246e-05,
+      "loss": 1.357,
+      "step": 70230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.705090657803811e-05,
+      "loss": 1.39,
+      "step": 70240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.704761591365297e-05,
+      "loss": 1.0576,
+      "step": 70250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7044325249267827e-05,
+      "loss": 1.3008,
+      "step": 70260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.704103458488269e-05,
+      "loss": 1.4373,
+      "step": 70270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.703774392049755e-05,
+      "loss": 1.4462,
+      "step": 70280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7034453256112414e-05,
+      "loss": 1.1894,
+      "step": 70290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7031162591727272e-05,
+      "loss": 1.3388,
+      "step": 70300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.702787192734213e-05,
+      "loss": 1.3457,
+      "step": 70310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7024581262956994e-05,
+      "loss": 1.2534,
+      "step": 70320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7021290598571852e-05,
+      "loss": 1.2052,
+      "step": 70330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7017999934186717e-05,
+      "loss": 1.3971,
+      "step": 70340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7014709269801575e-05,
+      "loss": 1.1955,
+      "step": 70350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7011418605416433e-05,
+      "loss": 1.3235,
+      "step": 70360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7008127941031297e-05,
+      "loss": 1.118,
+      "step": 70370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7004837276646155e-05,
+      "loss": 1.2739,
+      "step": 70380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.700154661226102e-05,
+      "loss": 1.1949,
+      "step": 70390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6998255947875878e-05,
+      "loss": 1.2593,
+      "step": 70400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6994965283490736e-05,
+      "loss": 1.3725,
+      "step": 70410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.69916746191056e-05,
+      "loss": 1.3161,
+      "step": 70420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6988383954720458e-05,
+      "loss": 1.2765,
+      "step": 70430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6985093290335323e-05,
+      "loss": 1.3018,
+      "step": 70440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.698180262595018e-05,
+      "loss": 1.3239,
+      "step": 70450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.697851196156504e-05,
+      "loss": 1.3093,
+      "step": 70460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6975221297179903e-05,
+      "loss": 1.2218,
+      "step": 70470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.697193063279476e-05,
+      "loss": 1.2747,
+      "step": 70480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6968639968409626e-05,
+      "loss": 1.198,
+      "step": 70490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6965349304024484e-05,
+      "loss": 1.2522,
+      "step": 70500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.696205863963934e-05,
+      "loss": 1.1927,
+      "step": 70510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6958767975254206e-05,
+      "loss": 1.2877,
+      "step": 70520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6955477310869064e-05,
+      "loss": 1.1962,
+      "step": 70530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.695218664648393e-05,
+      "loss": 1.2245,
+      "step": 70540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6948895982098787e-05,
+      "loss": 1.3817,
+      "step": 70550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6945605317713644e-05,
+      "loss": 1.1301,
+      "step": 70560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.694231465332851e-05,
+      "loss": 1.3051,
+      "step": 70570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6939023988943367e-05,
+      "loss": 1.4861,
+      "step": 70580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.693573332455823e-05,
+      "loss": 1.3773,
+      "step": 70590
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.693244266017309e-05,
+      "loss": 1.1878,
+      "step": 70600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6929151995787947e-05,
+      "loss": 1.1534,
+      "step": 70610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6925861331402812e-05,
+      "loss": 1.372,
+      "step": 70620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.692257066701767e-05,
+      "loss": 1.2283,
+      "step": 70630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6919280002632535e-05,
+      "loss": 1.2637,
+      "step": 70640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6915989338247392e-05,
+      "loss": 1.2426,
+      "step": 70650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6912698673862254e-05,
+      "loss": 1.3854,
+      "step": 70660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6909408009477115e-05,
+      "loss": 1.2986,
+      "step": 70670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6906117345091976e-05,
+      "loss": 1.2344,
+      "step": 70680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6902826680706838e-05,
+      "loss": 1.1564,
+      "step": 70690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6899536016321695e-05,
+      "loss": 1.3071,
+      "step": 70700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6896245351936557e-05,
+      "loss": 1.1452,
+      "step": 70710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6892954687551418e-05,
+      "loss": 1.4978,
+      "step": 70720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.688966402316628e-05,
+      "loss": 1.3823,
+      "step": 70730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.688637335878114e-05,
+      "loss": 1.3391,
+      "step": 70740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6883082694396002e-05,
+      "loss": 1.2767,
+      "step": 70750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.687979203001086e-05,
+      "loss": 1.1113,
+      "step": 70760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.687650136562572e-05,
+      "loss": 1.3647,
+      "step": 70770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6873210701240582e-05,
+      "loss": 1.3916,
+      "step": 70780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6869920036855443e-05,
+      "loss": 1.3199,
+      "step": 70790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6866629372470305e-05,
+      "loss": 1.3164,
+      "step": 70800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6863338708085163e-05,
+      "loss": 1.1119,
+      "step": 70810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6860048043700027e-05,
+      "loss": 1.1126,
+      "step": 70820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6856757379314885e-05,
+      "loss": 1.2559,
+      "step": 70830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.685346671492975e-05,
+      "loss": 1.1963,
+      "step": 70840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6850176050544608e-05,
+      "loss": 1.379,
+      "step": 70850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6846885386159466e-05,
+      "loss": 1.4568,
+      "step": 70860
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.684359472177433e-05,
+      "loss": 1.2734,
+      "step": 70870
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6840304057389188e-05,
+      "loss": 1.4835,
+      "step": 70880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6837013393004053e-05,
+      "loss": 1.175,
+      "step": 70890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.683372272861891e-05,
+      "loss": 1.2624,
+      "step": 70900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.683043206423377e-05,
+      "loss": 1.2602,
+      "step": 70910
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6827141399848633e-05,
+      "loss": 1.3973,
+      "step": 70920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.682385073546349e-05,
+      "loss": 1.3131,
+      "step": 70930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6820560071078356e-05,
+      "loss": 1.0518,
+      "step": 70940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6817269406693214e-05,
+      "loss": 1.2229,
+      "step": 70950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.681397874230807e-05,
+      "loss": 1.1416,
+      "step": 70960
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6810688077922936e-05,
+      "loss": 1.073,
+      "step": 70970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6807397413537794e-05,
+      "loss": 1.1564,
+      "step": 70980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.680410674915266e-05,
+      "loss": 1.1797,
+      "step": 70990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6800816084767516e-05,
+      "loss": 1.2501,
+      "step": 71000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6797525420382374e-05,
+      "loss": 1.3583,
+      "step": 71010
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.679423475599724e-05,
+      "loss": 1.3375,
+      "step": 71020
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6790944091612097e-05,
+      "loss": 1.3319,
+      "step": 71030
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.678765342722696e-05,
+      "loss": 1.2266,
+      "step": 71040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.678436276284182e-05,
+      "loss": 1.2807,
+      "step": 71050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6781072098456677e-05,
+      "loss": 1.2103,
+      "step": 71060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6777781434071542e-05,
+      "loss": 1.2619,
+      "step": 71070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.67744907696864e-05,
+      "loss": 1.4024,
+      "step": 71080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6771200105301264e-05,
+      "loss": 1.178,
+      "step": 71090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6767909440916122e-05,
+      "loss": 1.3452,
+      "step": 71100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.676461877653098e-05,
+      "loss": 1.2985,
+      "step": 71110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6761328112145845e-05,
+      "loss": 1.1471,
+      "step": 71120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6758037447760703e-05,
+      "loss": 1.2748,
+      "step": 71130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6754746783375567e-05,
+      "loss": 1.2113,
+      "step": 71140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6751456118990425e-05,
+      "loss": 1.5246,
+      "step": 71150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6748165454605283e-05,
+      "loss": 1.2164,
+      "step": 71160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6744874790220148e-05,
+      "loss": 1.2172,
+      "step": 71170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6741584125835006e-05,
+      "loss": 1.4466,
+      "step": 71180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.673829346144987e-05,
+      "loss": 1.2927,
+      "step": 71190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6735002797064728e-05,
+      "loss": 1.2374,
+      "step": 71200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6731712132679586e-05,
+      "loss": 1.296,
+      "step": 71210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.672842146829445e-05,
+      "loss": 1.2561,
+      "step": 71220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.672513080390931e-05,
+      "loss": 1.4224,
+      "step": 71230
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6721840139524173e-05,
+      "loss": 1.3173,
+      "step": 71240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.671854947513903e-05,
+      "loss": 1.2467,
+      "step": 71250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.671525881075389e-05,
+      "loss": 1.3612,
+      "step": 71260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6711968146368754e-05,
+      "loss": 1.1348,
+      "step": 71270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.670867748198361e-05,
+      "loss": 1.3368,
+      "step": 71280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6705386817598476e-05,
+      "loss": 1.3281,
+      "step": 71290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6702096153213334e-05,
+      "loss": 1.2012,
+      "step": 71300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6698805488828192e-05,
+      "loss": 1.2077,
+      "step": 71310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6695514824443057e-05,
+      "loss": 1.4033,
+      "step": 71320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6692224160057915e-05,
+      "loss": 1.1968,
+      "step": 71330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.668893349567278e-05,
+      "loss": 1.3609,
+      "step": 71340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6685642831287637e-05,
+      "loss": 1.2566,
+      "step": 71350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.66823521669025e-05,
+      "loss": 1.2625,
+      "step": 71360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.667906150251736e-05,
+      "loss": 1.3462,
+      "step": 71370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6675770838132218e-05,
+      "loss": 1.2623,
+      "step": 71380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6672480173747082e-05,
+      "loss": 1.2039,
+      "step": 71390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.666918950936194e-05,
+      "loss": 1.2844,
+      "step": 71400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.66658988449768e-05,
+      "loss": 1.3452,
+      "step": 71410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6662608180591663e-05,
+      "loss": 1.3912,
+      "step": 71420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6659317516206524e-05,
+      "loss": 1.1056,
+      "step": 71430
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6656026851821385e-05,
+      "loss": 1.2014,
+      "step": 71440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6652736187436243e-05,
+      "loss": 1.2307,
+      "step": 71450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6649445523051104e-05,
+      "loss": 1.0938,
+      "step": 71460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6646154858665966e-05,
+      "loss": 1.2244,
+      "step": 71470
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6642864194280827e-05,
+      "loss": 1.2659,
+      "step": 71480
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6639573529895688e-05,
+      "loss": 1.2275,
+      "step": 71490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.663628286551055e-05,
+      "loss": 1.5002,
+      "step": 71500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6632992201125407e-05,
+      "loss": 1.2358,
+      "step": 71510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6629701536740272e-05,
+      "loss": 1.321,
+      "step": 71520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.662641087235513e-05,
+      "loss": 1.2466,
+      "step": 71530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.662312020796999e-05,
+      "loss": 1.312,
+      "step": 71540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6619829543584852e-05,
+      "loss": 1.3093,
+      "step": 71550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.661653887919971e-05,
+      "loss": 1.3793,
+      "step": 71560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6613248214814575e-05,
+      "loss": 1.3938,
+      "step": 71570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6609957550429433e-05,
+      "loss": 0.999,
+      "step": 71580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6606666886044297e-05,
+      "loss": 1.3274,
+      "step": 71590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6603376221659155e-05,
+      "loss": 1.3749,
+      "step": 71600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6600085557274013e-05,
+      "loss": 1.249,
+      "step": 71610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6596794892888878e-05,
+      "loss": 1.205,
+      "step": 71620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6593504228503736e-05,
+      "loss": 1.1908,
+      "step": 71630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.65902135641186e-05,
+      "loss": 1.1478,
+      "step": 71640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6586922899733458e-05,
+      "loss": 1.3696,
+      "step": 71650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6583632235348316e-05,
+      "loss": 1.2418,
+      "step": 71660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.658034157096318e-05,
+      "loss": 1.2173,
+      "step": 71670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.657705090657804e-05,
+      "loss": 1.3038,
+      "step": 71680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6573760242192903e-05,
+      "loss": 1.2858,
+      "step": 71690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.657046957780776e-05,
+      "loss": 1.3009,
+      "step": 71700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.656717891342262e-05,
+      "loss": 1.4369,
+      "step": 71710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6563888249037484e-05,
+      "loss": 1.116,
+      "step": 71720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.656059758465234e-05,
+      "loss": 1.0885,
+      "step": 71730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6557306920267206e-05,
+      "loss": 1.2029,
+      "step": 71740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6554016255882064e-05,
+      "loss": 1.328,
+      "step": 71750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6550725591496922e-05,
+      "loss": 1.2974,
+      "step": 71760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6547434927111787e-05,
+      "loss": 1.2591,
+      "step": 71770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6544144262726645e-05,
+      "loss": 1.2734,
+      "step": 71780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.654085359834151e-05,
+      "loss": 1.2277,
+      "step": 71790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6537562933956367e-05,
+      "loss": 1.2095,
+      "step": 71800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6534272269571225e-05,
+      "loss": 1.2345,
+      "step": 71810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.653098160518609e-05,
+      "loss": 1.2105,
+      "step": 71820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6527690940800947e-05,
+      "loss": 1.1915,
+      "step": 71830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6524400276415812e-05,
+      "loss": 1.4467,
+      "step": 71840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.652110961203067e-05,
+      "loss": 1.3197,
+      "step": 71850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6517818947645528e-05,
+      "loss": 1.3889,
+      "step": 71860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6514528283260393e-05,
+      "loss": 1.2632,
+      "step": 71870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.651123761887525e-05,
+      "loss": 1.4764,
+      "step": 71880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6507946954490115e-05,
+      "loss": 1.3242,
+      "step": 71890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6504656290104973e-05,
+      "loss": 1.2584,
+      "step": 71900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.650136562571983e-05,
+      "loss": 1.1679,
+      "step": 71910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6498074961334695e-05,
+      "loss": 1.233,
+      "step": 71920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6494784296949553e-05,
+      "loss": 1.2236,
+      "step": 71930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6491493632564418e-05,
+      "loss": 1.2963,
+      "step": 71940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6488202968179276e-05,
+      "loss": 1.158,
+      "step": 71950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6484912303794134e-05,
+      "loss": 1.4337,
+      "step": 71960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6481621639409e-05,
+      "loss": 1.1504,
+      "step": 71970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6478330975023856e-05,
+      "loss": 1.2095,
+      "step": 71980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.647504031063872e-05,
+      "loss": 1.2739,
+      "step": 71990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.647174964625358e-05,
+      "loss": 1.3649,
+      "step": 72000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6468458981868437e-05,
+      "loss": 1.2905,
+      "step": 72010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.64651683174833e-05,
+      "loss": 1.1975,
+      "step": 72020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.646187765309816e-05,
+      "loss": 1.4249,
+      "step": 72030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6458586988713024e-05,
+      "loss": 1.2173,
+      "step": 72040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6455296324327882e-05,
+      "loss": 1.3974,
+      "step": 72050
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.645200565994274e-05,
+      "loss": 1.3413,
+      "step": 72060
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6448714995557604e-05,
+      "loss": 1.2347,
+      "step": 72070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6445424331172462e-05,
+      "loss": 1.2397,
+      "step": 72080
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6442133666787327e-05,
+      "loss": 1.2492,
+      "step": 72090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6438843002402185e-05,
+      "loss": 1.4158,
+      "step": 72100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6435552338017046e-05,
+      "loss": 1.4926,
+      "step": 72110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6432261673631907e-05,
+      "loss": 1.3488,
+      "step": 72120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6428971009246765e-05,
+      "loss": 1.2879,
+      "step": 72130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.642568034486163e-05,
+      "loss": 1.2322,
+      "step": 72140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6422389680476488e-05,
+      "loss": 1.1364,
+      "step": 72150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.641909901609135e-05,
+      "loss": 1.3049,
+      "step": 72160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.641580835170621e-05,
+      "loss": 1.3018,
+      "step": 72170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.641251768732107e-05,
+      "loss": 1.2166,
+      "step": 72180
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6409227022935933e-05,
+      "loss": 1.2578,
+      "step": 72190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6405936358550794e-05,
+      "loss": 1.3445,
+      "step": 72200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6402645694165652e-05,
+      "loss": 1.2245,
+      "step": 72210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6399355029780513e-05,
+      "loss": 1.3538,
+      "step": 72220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6396064365395374e-05,
+      "loss": 1.2564,
+      "step": 72230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6392773701010236e-05,
+      "loss": 1.1903,
+      "step": 72240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6389483036625097e-05,
+      "loss": 1.2196,
+      "step": 72250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6386192372239955e-05,
+      "loss": 1.334,
+      "step": 72260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.638290170785482e-05,
+      "loss": 1.0856,
+      "step": 72270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6379611043469677e-05,
+      "loss": 1.3401,
+      "step": 72280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.637632037908454e-05,
+      "loss": 1.1926,
+      "step": 72290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.63730297146994e-05,
+      "loss": 1.4281,
+      "step": 72300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6369739050314258e-05,
+      "loss": 1.2211,
+      "step": 72310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6366448385929122e-05,
+      "loss": 1.4408,
+      "step": 72320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.636315772154398e-05,
+      "loss": 1.2426,
+      "step": 72330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6359867057158845e-05,
+      "loss": 1.2428,
+      "step": 72340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6356576392773703e-05,
+      "loss": 1.2635,
+      "step": 72350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.635328572838856e-05,
+      "loss": 1.2662,
+      "step": 72360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6349995064003425e-05,
+      "loss": 1.3811,
+      "step": 72370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6346704399618283e-05,
+      "loss": 1.2462,
+      "step": 72380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6343413735233148e-05,
+      "loss": 1.2811,
+      "step": 72390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6340123070848006e-05,
+      "loss": 1.1522,
+      "step": 72400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6336832406462864e-05,
+      "loss": 1.1568,
+      "step": 72410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.633354174207773e-05,
+      "loss": 1.1258,
+      "step": 72420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6330251077692586e-05,
+      "loss": 1.2709,
+      "step": 72430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.632696041330745e-05,
+      "loss": 1.2442,
+      "step": 72440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.632366974892231e-05,
+      "loss": 1.2021,
+      "step": 72450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6320379084537167e-05,
+      "loss": 1.2888,
+      "step": 72460
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.631708842015203e-05,
+      "loss": 1.3019,
+      "step": 72470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.631379775576689e-05,
+      "loss": 1.377,
+      "step": 72480
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6310507091381754e-05,
+      "loss": 1.4309,
+      "step": 72490
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6307216426996612e-05,
+      "loss": 1.2271,
+      "step": 72500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630392576261147e-05,
+      "loss": 1.3695,
+      "step": 72510
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6300635098226334e-05,
+      "loss": 1.379,
+      "step": 72520
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6297344433841192e-05,
+      "loss": 1.3607,
+      "step": 72530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6294053769456057e-05,
+      "loss": 1.1918,
+      "step": 72540
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6290763105070915e-05,
+      "loss": 1.1752,
+      "step": 72550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6287472440685773e-05,
+      "loss": 1.0889,
+      "step": 72560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6284181776300637e-05,
+      "loss": 1.1525,
+      "step": 72570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6280891111915495e-05,
+      "loss": 1.2516,
+      "step": 72580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.627760044753036e-05,
+      "loss": 1.2766,
+      "step": 72590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6274309783145218e-05,
+      "loss": 1.1779,
+      "step": 72600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6271019118760076e-05,
+      "loss": 1.4325,
+      "step": 72610
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.626772845437494e-05,
+      "loss": 1.2772,
+      "step": 72620
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6264437789989798e-05,
+      "loss": 1.2713,
+      "step": 72630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6261147125604663e-05,
+      "loss": 1.148,
+      "step": 72640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.625785646121952e-05,
+      "loss": 1.2599,
+      "step": 72650
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.625456579683438e-05,
+      "loss": 1.1615,
+      "step": 72660
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6251275132449243e-05,
+      "loss": 1.2224,
+      "step": 72670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.62479844680641e-05,
+      "loss": 1.2404,
+      "step": 72680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6244693803678966e-05,
+      "loss": 1.6224,
+      "step": 72690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6241403139293824e-05,
+      "loss": 1.3779,
+      "step": 72700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.623811247490868e-05,
+      "loss": 1.3498,
+      "step": 72710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6234821810523546e-05,
+      "loss": 1.3564,
+      "step": 72720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6231531146138404e-05,
+      "loss": 1.17,
+      "step": 72730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.622824048175327e-05,
+      "loss": 1.3014,
+      "step": 72740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6224949817368126e-05,
+      "loss": 1.3219,
+      "step": 72750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6221659152982984e-05,
+      "loss": 1.2324,
+      "step": 72760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.621836848859785e-05,
+      "loss": 1.173,
+      "step": 72770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6215077824212707e-05,
+      "loss": 1.14,
+      "step": 72780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.621178715982757e-05,
+      "loss": 1.2407,
+      "step": 72790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.620849649544243e-05,
+      "loss": 1.2068,
+      "step": 72800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6205205831057287e-05,
+      "loss": 1.3888,
+      "step": 72810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6201915166672152e-05,
+      "loss": 1.2171,
+      "step": 72820
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.619862450228701e-05,
+      "loss": 1.6396,
+      "step": 72830
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6195333837901874e-05,
+      "loss": 1.2629,
+      "step": 72840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6192043173516732e-05,
+      "loss": 1.302,
+      "step": 72850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6188752509131594e-05,
+      "loss": 1.2569,
+      "step": 72860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6185461844746455e-05,
+      "loss": 1.3847,
+      "step": 72870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6182171180361316e-05,
+      "loss": 1.1928,
+      "step": 72880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6178880515976177e-05,
+      "loss": 1.2418,
+      "step": 72890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6175589851591035e-05,
+      "loss": 1.2042,
+      "step": 72900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6172299187205897e-05,
+      "loss": 1.2098,
+      "step": 72910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6169008522820758e-05,
+      "loss": 1.1497,
+      "step": 72920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.616571785843562e-05,
+      "loss": 1.2079,
+      "step": 72930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.616242719405048e-05,
+      "loss": 1.235,
+      "step": 72940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.615913652966534e-05,
+      "loss": 1.1552,
+      "step": 72950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6155845865280203e-05,
+      "loss": 0.9922,
+      "step": 72960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6152555200895064e-05,
+      "loss": 1.1175,
+      "step": 72970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6149264536509922e-05,
+      "loss": 1.2794,
+      "step": 72980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6145973872124783e-05,
+      "loss": 1.3372,
+      "step": 72990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6142683207739645e-05,
+      "loss": 1.2127,
+      "step": 73000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6139392543354506e-05,
+      "loss": 1.3837,
+      "step": 73010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6136101878969367e-05,
+      "loss": 1.208,
+      "step": 73020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6132811214584225e-05,
+      "loss": 1.3694,
+      "step": 73030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.612952055019909e-05,
+      "loss": 1.2874,
+      "step": 73040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6126229885813948e-05,
+      "loss": 1.248,
+      "step": 73050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.612293922142881e-05,
+      "loss": 1.3306,
+      "step": 73060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.611964855704367e-05,
+      "loss": 1.2244,
+      "step": 73070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6116357892658528e-05,
+      "loss": 1.34,
+      "step": 73080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6113067228273393e-05,
+      "loss": 1.0372,
+      "step": 73090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.610977656388825e-05,
+      "loss": 1.2122,
+      "step": 73100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6106485899503115e-05,
+      "loss": 1.2575,
+      "step": 73110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6103195235117973e-05,
+      "loss": 1.2828,
+      "step": 73120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.609990457073283e-05,
+      "loss": 1.16,
+      "step": 73130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6096613906347696e-05,
+      "loss": 1.3624,
+      "step": 73140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6093323241962553e-05,
+      "loss": 1.2992,
+      "step": 73150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6090032577577418e-05,
+      "loss": 1.1957,
+      "step": 73160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6086741913192276e-05,
+      "loss": 1.3316,
+      "step": 73170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6083451248807134e-05,
+      "loss": 1.219,
+      "step": 73180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6080160584422e-05,
+      "loss": 1.1365,
+      "step": 73190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6076869920036856e-05,
+      "loss": 1.2232,
+      "step": 73200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.607357925565172e-05,
+      "loss": 1.2514,
+      "step": 73210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.607028859126658e-05,
+      "loss": 1.3777,
+      "step": 73220
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6066997926881437e-05,
+      "loss": 1.2146,
+      "step": 73230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.60637072624963e-05,
+      "loss": 1.3273,
+      "step": 73240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.606041659811116e-05,
+      "loss": 1.125,
+      "step": 73250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6057125933726024e-05,
+      "loss": 1.3142,
+      "step": 73260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6053835269340882e-05,
+      "loss": 1.257,
+      "step": 73270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.605054460495574e-05,
+      "loss": 1.2551,
+      "step": 73280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6047253940570604e-05,
+      "loss": 1.2726,
+      "step": 73290
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6043963276185462e-05,
+      "loss": 1.3237,
+      "step": 73300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6040672611800327e-05,
+      "loss": 1.3562,
+      "step": 73310
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6037381947415185e-05,
+      "loss": 1.1453,
+      "step": 73320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6034091283030043e-05,
+      "loss": 1.4094,
+      "step": 73330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6030800618644907e-05,
+      "loss": 1.2264,
+      "step": 73340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6027509954259765e-05,
+      "loss": 1.281,
+      "step": 73350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.602421928987463e-05,
+      "loss": 1.3155,
+      "step": 73360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6020928625489488e-05,
+      "loss": 1.2532,
+      "step": 73370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6017637961104346e-05,
+      "loss": 1.1722,
+      "step": 73380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.601434729671921e-05,
+      "loss": 1.2553,
+      "step": 73390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6011056632334068e-05,
+      "loss": 1.2595,
+      "step": 73400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6007765967948933e-05,
+      "loss": 1.2852,
+      "step": 73410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.600447530356379e-05,
+      "loss": 1.355,
+      "step": 73420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.600118463917865e-05,
+      "loss": 1.1275,
+      "step": 73430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5997893974793513e-05,
+      "loss": 1.2381,
+      "step": 73440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.599460331040837e-05,
+      "loss": 1.2654,
+      "step": 73450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5991312646023236e-05,
+      "loss": 1.2042,
+      "step": 73460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5988021981638094e-05,
+      "loss": 1.1899,
+      "step": 73470
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.598473131725295e-05,
+      "loss": 1.1604,
+      "step": 73480
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5981440652867816e-05,
+      "loss": 1.357,
+      "step": 73490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5978149988482674e-05,
+      "loss": 1.3182,
+      "step": 73500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.597485932409754e-05,
+      "loss": 1.2914,
+      "step": 73510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5971568659712397e-05,
+      "loss": 1.1723,
+      "step": 73520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5968277995327255e-05,
+      "loss": 1.4813,
+      "step": 73530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.596498733094212e-05,
+      "loss": 1.2904,
+      "step": 73540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5961696666556977e-05,
+      "loss": 0.9449,
+      "step": 73550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5958406002171842e-05,
+      "loss": 1.4733,
+      "step": 73560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.59551153377867e-05,
+      "loss": 1.4666,
+      "step": 73570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5951824673401557e-05,
+      "loss": 1.2448,
+      "step": 73580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5948534009016422e-05,
+      "loss": 1.4654,
+      "step": 73590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.594524334463128e-05,
+      "loss": 1.2003,
+      "step": 73600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5941952680246145e-05,
+      "loss": 1.0784,
+      "step": 73610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5938662015861003e-05,
+      "loss": 1.462,
+      "step": 73620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5935371351475864e-05,
+      "loss": 1.4884,
+      "step": 73630
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5932080687090725e-05,
+      "loss": 1.428,
+      "step": 73640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5928790022705586e-05,
+      "loss": 1.4096,
+      "step": 73650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5925499358320448e-05,
+      "loss": 1.2915,
+      "step": 73660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5922208693935305e-05,
+      "loss": 1.2427,
+      "step": 73670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5918918029550167e-05,
+      "loss": 1.1912,
+      "step": 73680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5915627365165028e-05,
+      "loss": 1.186,
+      "step": 73690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.591233670077989e-05,
+      "loss": 1.2099,
+      "step": 73700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.590904603639475e-05,
+      "loss": 1.2233,
+      "step": 73710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5905755372009612e-05,
+      "loss": 1.2191,
+      "step": 73720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.590246470762447e-05,
+      "loss": 1.2662,
+      "step": 73730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.589917404323933e-05,
+      "loss": 1.304,
+      "step": 73740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5895883378854192e-05,
+      "loss": 1.3577,
+      "step": 73750
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5892592714469054e-05,
+      "loss": 1.3821,
+      "step": 73760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5889302050083915e-05,
+      "loss": 1.1932,
+      "step": 73770
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5886011385698773e-05,
+      "loss": 1.2378,
+      "step": 73780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5882720721313637e-05,
+      "loss": 1.3668,
+      "step": 73790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5879430056928495e-05,
+      "loss": 1.1666,
+      "step": 73800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.587613939254336e-05,
+      "loss": 1.2695,
+      "step": 73810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5872848728158218e-05,
+      "loss": 1.2688,
+      "step": 73820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5869558063773076e-05,
+      "loss": 1.298,
+      "step": 73830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.586626739938794e-05,
+      "loss": 1.2356,
+      "step": 73840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5862976735002798e-05,
+      "loss": 1.333,
+      "step": 73850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5859686070617663e-05,
+      "loss": 1.1594,
+      "step": 73860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.585639540623252e-05,
+      "loss": 1.3272,
+      "step": 73870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.585310474184738e-05,
+      "loss": 1.4041,
+      "step": 73880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5849814077462243e-05,
+      "loss": 1.2672,
+      "step": 73890
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.58465234130771e-05,
+      "loss": 1.1725,
+      "step": 73900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5843232748691966e-05,
+      "loss": 1.1756,
+      "step": 73910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5839942084306824e-05,
+      "loss": 1.3063,
+      "step": 73920
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.583665141992168e-05,
+      "loss": 1.3723,
+      "step": 73930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5833360755536546e-05,
+      "loss": 1.1739,
+      "step": 73940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5830070091151404e-05,
+      "loss": 1.3305,
+      "step": 73950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.582677942676627e-05,
+      "loss": 1.2446,
+      "step": 73960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5823488762381127e-05,
+      "loss": 1.376,
+      "step": 73970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5820198097995984e-05,
+      "loss": 1.4352,
+      "step": 73980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.581690743361085e-05,
+      "loss": 1.4243,
+      "step": 73990
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5813616769225707e-05,
+      "loss": 1.3737,
+      "step": 74000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.581032610484057e-05,
+      "loss": 1.3535,
+      "step": 74010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.580703544045543e-05,
+      "loss": 1.095,
+      "step": 74020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5803744776070287e-05,
+      "loss": 1.469,
+      "step": 74030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5800454111685152e-05,
+      "loss": 1.2266,
+      "step": 74040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.579716344730001e-05,
+      "loss": 1.2752,
+      "step": 74050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5793872782914875e-05,
+      "loss": 1.2495,
+      "step": 74060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5790582118529732e-05,
+      "loss": 1.2055,
+      "step": 74070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.578729145414459e-05,
+      "loss": 1.1602,
+      "step": 74080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5784000789759455e-05,
+      "loss": 1.0463,
+      "step": 74090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5780710125374313e-05,
+      "loss": 1.436,
+      "step": 74100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5777419460989178e-05,
+      "loss": 1.1819,
+      "step": 74110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5774128796604035e-05,
+      "loss": 1.2684,
+      "step": 74120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5770838132218893e-05,
+      "loss": 1.1898,
+      "step": 74130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5767547467833758e-05,
+      "loss": 1.2316,
+      "step": 74140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5764256803448616e-05,
+      "loss": 1.2505,
+      "step": 74150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.576096613906348e-05,
+      "loss": 1.2318,
+      "step": 74160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.575767547467834e-05,
+      "loss": 1.3205,
+      "step": 74170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5754384810293196e-05,
+      "loss": 1.4093,
+      "step": 74180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.575109414590806e-05,
+      "loss": 1.2648,
+      "step": 74190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.574780348152292e-05,
+      "loss": 1.1499,
+      "step": 74200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5744512817137783e-05,
+      "loss": 1.4033,
+      "step": 74210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.574122215275264e-05,
+      "loss": 1.2214,
+      "step": 74220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.57379314883675e-05,
+      "loss": 1.2147,
+      "step": 74230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5734640823982364e-05,
+      "loss": 1.1816,
+      "step": 74240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5731350159597222e-05,
+      "loss": 1.2357,
+      "step": 74250
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5728059495212086e-05,
+      "loss": 1.3097,
+      "step": 74260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5724768830826944e-05,
+      "loss": 1.2137,
+      "step": 74270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5721478166441802e-05,
+      "loss": 1.1319,
+      "step": 74280
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5718187502056667e-05,
+      "loss": 1.3136,
+      "step": 74290
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5714896837671525e-05,
+      "loss": 1.3822,
+      "step": 74300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.571160617328639e-05,
+      "loss": 1.2347,
+      "step": 74310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5708315508901247e-05,
+      "loss": 1.2067,
+      "step": 74320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.570502484451611e-05,
+      "loss": 1.2646,
+      "step": 74330
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.570173418013097e-05,
+      "loss": 1.323,
+      "step": 74340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5698443515745828e-05,
+      "loss": 1.2101,
+      "step": 74350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5695152851360692e-05,
+      "loss": 1.5937,
+      "step": 74360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.569186218697555e-05,
+      "loss": 1.1867,
+      "step": 74370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.568857152259041e-05,
+      "loss": 1.1677,
+      "step": 74380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5685280858205273e-05,
+      "loss": 1.3624,
+      "step": 74390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5681990193820134e-05,
+      "loss": 1.5022,
+      "step": 74400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5678699529434995e-05,
+      "loss": 1.1849,
+      "step": 74410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5675408865049853e-05,
+      "loss": 1.3135,
+      "step": 74420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5672118200664714e-05,
+      "loss": 1.1656,
+      "step": 74430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5668827536279576e-05,
+      "loss": 1.2294,
+      "step": 74440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5665536871894437e-05,
+      "loss": 1.326,
+      "step": 74450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5662246207509298e-05,
+      "loss": 1.3734,
+      "step": 74460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.565895554312416e-05,
+      "loss": 1.2105,
+      "step": 74470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5655664878739017e-05,
+      "loss": 1.1664,
+      "step": 74480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5652374214353882e-05,
+      "loss": 1.2939,
+      "step": 74490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.564908354996874e-05,
+      "loss": 1.4312,
+      "step": 74500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.56457928855836e-05,
+      "loss": 1.1513,
+      "step": 74510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5642502221198462e-05,
+      "loss": 1.2453,
+      "step": 74520
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.563921155681332e-05,
+      "loss": 1.2487,
+      "step": 74530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5635920892428185e-05,
+      "loss": 1.2899,
+      "step": 74540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5632630228043043e-05,
+      "loss": 1.1854,
+      "step": 74550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5629339563657907e-05,
+      "loss": 1.326,
+      "step": 74560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5626048899272765e-05,
+      "loss": 1.2376,
+      "step": 74570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5622758234887623e-05,
+      "loss": 1.2351,
+      "step": 74580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5619467570502488e-05,
+      "loss": 1.2505,
+      "step": 74590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5616176906117346e-05,
+      "loss": 1.2954,
+      "step": 74600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.561288624173221e-05,
+      "loss": 1.2065,
+      "step": 74610
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5609595577347068e-05,
+      "loss": 1.1489,
+      "step": 74620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5606304912961926e-05,
+      "loss": 1.3605,
+      "step": 74630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.560301424857679e-05,
+      "loss": 1.2393,
+      "step": 74640
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.559972358419165e-05,
+      "loss": 1.3192,
+      "step": 74650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5596432919806513e-05,
+      "loss": 1.3656,
+      "step": 74660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.559314225542137e-05,
+      "loss": 1.3057,
+      "step": 74670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.558985159103623e-05,
+      "loss": 1.2016,
+      "step": 74680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5586560926651094e-05,
+      "loss": 1.2712,
+      "step": 74690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.558327026226595e-05,
+      "loss": 1.3185,
+      "step": 74700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5579979597880816e-05,
+      "loss": 1.4328,
+      "step": 74710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5576688933495674e-05,
+      "loss": 1.19,
+      "step": 74720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5573398269110532e-05,
+      "loss": 1.2381,
+      "step": 74730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5570107604725397e-05,
+      "loss": 1.1982,
+      "step": 74740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5566816940340255e-05,
+      "loss": 1.2624,
+      "step": 74750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.556352627595512e-05,
+      "loss": 1.3062,
+      "step": 74760
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5560235611569977e-05,
+      "loss": 1.2661,
+      "step": 74770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5556944947184835e-05,
+      "loss": 1.2937,
+      "step": 74780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.55536542827997e-05,
+      "loss": 1.4314,
+      "step": 74790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5550363618414558e-05,
+      "loss": 1.1656,
+      "step": 74800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5547072954029422e-05,
+      "loss": 1.3399,
+      "step": 74810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.554378228964428e-05,
+      "loss": 1.5562,
+      "step": 74820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5540491625259138e-05,
+      "loss": 1.2992,
+      "step": 74830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5537200960874003e-05,
+      "loss": 1.1955,
+      "step": 74840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.553391029648886e-05,
+      "loss": 1.3605,
+      "step": 74850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5530619632103725e-05,
+      "loss": 1.1548,
+      "step": 74860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5527328967718583e-05,
+      "loss": 1.1505,
+      "step": 74870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.552403830333344e-05,
+      "loss": 1.2493,
+      "step": 74880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5520747638948306e-05,
+      "loss": 1.3941,
+      "step": 74890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5517456974563163e-05,
+      "loss": 1.1913,
+      "step": 74900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5514166310178028e-05,
+      "loss": 1.2391,
+      "step": 74910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5510875645792886e-05,
+      "loss": 1.6715,
+      "step": 74920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5507584981407744e-05,
+      "loss": 1.2779,
+      "step": 74930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.550429431702261e-05,
+      "loss": 1.1607,
+      "step": 74940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5501003652637466e-05,
+      "loss": 1.2982,
+      "step": 74950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.549771298825233e-05,
+      "loss": 1.1585,
+      "step": 74960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.549442232386719e-05,
+      "loss": 1.4293,
+      "step": 74970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5491131659482047e-05,
+      "loss": 1.2698,
+      "step": 74980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.548784099509691e-05,
+      "loss": 1.2675,
+      "step": 74990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.548455033071177e-05,
+      "loss": 1.4253,
+      "step": 75000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5481259666326634e-05,
+      "loss": 1.3601,
+      "step": 75010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5477969001941492e-05,
+      "loss": 1.2619,
+      "step": 75020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.547467833755635e-05,
+      "loss": 1.2903,
+      "step": 75030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5471387673171214e-05,
+      "loss": 1.2096,
+      "step": 75040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5468097008786072e-05,
+      "loss": 1.3083,
+      "step": 75050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5464806344400937e-05,
+      "loss": 1.2685,
+      "step": 75060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5461515680015795e-05,
+      "loss": 1.1956,
+      "step": 75070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5458225015630656e-05,
+      "loss": 1.223,
+      "step": 75080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5454934351245517e-05,
+      "loss": 1.2834,
+      "step": 75090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5451643686860375e-05,
+      "loss": 1.0978,
+      "step": 75100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.544835302247524e-05,
+      "loss": 1.3071,
+      "step": 75110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5445062358090098e-05,
+      "loss": 1.2846,
+      "step": 75120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.544177169370496e-05,
+      "loss": 1.1952,
+      "step": 75130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.543848102931982e-05,
+      "loss": 1.198,
+      "step": 75140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.543519036493468e-05,
+      "loss": 1.2409,
+      "step": 75150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5431899700549543e-05,
+      "loss": 1.1954,
+      "step": 75160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5428609036164404e-05,
+      "loss": 1.3003,
+      "step": 75170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5425318371779262e-05,
+      "loss": 1.3216,
+      "step": 75180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5422027707394123e-05,
+      "loss": 1.2068,
+      "step": 75190
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5418737043008985e-05,
+      "loss": 1.1949,
+      "step": 75200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5415446378623846e-05,
+      "loss": 1.306,
+      "step": 75210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5412155714238707e-05,
+      "loss": 1.2028,
+      "step": 75220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5408865049853565e-05,
+      "loss": 1.1531,
+      "step": 75230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.540557438546843e-05,
+      "loss": 1.126,
+      "step": 75240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5402283721083287e-05,
+      "loss": 1.1112,
+      "step": 75250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.539899305669815e-05,
+      "loss": 1.2474,
+      "step": 75260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.539570239231301e-05,
+      "loss": 1.4027,
+      "step": 75270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5392411727927868e-05,
+      "loss": 1.2614,
+      "step": 75280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5389121063542733e-05,
+      "loss": 1.2821,
+      "step": 75290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.538583039915759e-05,
+      "loss": 1.1071,
+      "step": 75300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5382539734772455e-05,
+      "loss": 1.461,
+      "step": 75310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5379249070387313e-05,
+      "loss": 1.1046,
+      "step": 75320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.537595840600217e-05,
+      "loss": 1.3317,
+      "step": 75330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5372667741617036e-05,
+      "loss": 1.2188,
+      "step": 75340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5369377077231893e-05,
+      "loss": 1.1898,
+      "step": 75350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5366086412846758e-05,
+      "loss": 1.3915,
+      "step": 75360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5362795748461616e-05,
+      "loss": 1.1056,
+      "step": 75370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5359505084076474e-05,
+      "loss": 1.3659,
+      "step": 75380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.535621441969134e-05,
+      "loss": 1.3123,
+      "step": 75390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5352923755306196e-05,
+      "loss": 1.3797,
+      "step": 75400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.534963309092106e-05,
+      "loss": 1.3135,
+      "step": 75410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.534634242653592e-05,
+      "loss": 1.2951,
+      "step": 75420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5343051762150777e-05,
+      "loss": 1.6734,
+      "step": 75430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.533976109776564e-05,
+      "loss": 1.3908,
+      "step": 75440
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.53364704333805e-05,
+      "loss": 1.2851,
+      "step": 75450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5333179768995364e-05,
+      "loss": 1.3334,
+      "step": 75460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5329889104610222e-05,
+      "loss": 1.3746,
+      "step": 75470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.532659844022508e-05,
+      "loss": 1.239,
+      "step": 75480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5323307775839944e-05,
+      "loss": 1.2173,
+      "step": 75490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5320017111454802e-05,
+      "loss": 1.2816,
+      "step": 75500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5316726447069667e-05,
+      "loss": 1.5993,
+      "step": 75510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5313435782684525e-05,
+      "loss": 1.2237,
+      "step": 75520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5310145118299383e-05,
+      "loss": 1.5309,
+      "step": 75530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5306854453914247e-05,
+      "loss": 1.2384,
+      "step": 75540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5303563789529105e-05,
+      "loss": 1.3142,
+      "step": 75550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.530027312514397e-05,
+      "loss": 1.2674,
+      "step": 75560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5296982460758828e-05,
+      "loss": 1.1539,
+      "step": 75570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5293691796373686e-05,
+      "loss": 1.1566,
+      "step": 75580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.529040113198855e-05,
+      "loss": 1.2508,
+      "step": 75590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5287110467603408e-05,
+      "loss": 1.1995,
+      "step": 75600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5283819803218273e-05,
+      "loss": 1.2231,
+      "step": 75610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.528052913883313e-05,
+      "loss": 1.1659,
+      "step": 75620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.527723847444799e-05,
+      "loss": 1.292,
+      "step": 75630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5273947810062853e-05,
+      "loss": 1.2105,
+      "step": 75640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.527065714567771e-05,
+      "loss": 1.1737,
+      "step": 75650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5267366481292576e-05,
+      "loss": 1.2562,
+      "step": 75660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5264075816907434e-05,
+      "loss": 1.1773,
+      "step": 75670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.526078515252229e-05,
+      "loss": 1.3184,
+      "step": 75680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5257494488137156e-05,
+      "loss": 1.3167,
+      "step": 75690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5254203823752014e-05,
+      "loss": 1.1912,
+      "step": 75700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.525091315936688e-05,
+      "loss": 1.4354,
+      "step": 75710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5247622494981737e-05,
+      "loss": 1.2418,
+      "step": 75720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5244331830596594e-05,
+      "loss": 1.2601,
+      "step": 75730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.524104116621146e-05,
+      "loss": 1.219,
+      "step": 75740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5237750501826317e-05,
+      "loss": 1.4137,
+      "step": 75750
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.523445983744118e-05,
+      "loss": 1.2874,
+      "step": 75760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.523116917305604e-05,
+      "loss": 1.0948,
+      "step": 75770
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5227878508670904e-05,
+      "loss": 1.3414,
+      "step": 75780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5224587844285762e-05,
+      "loss": 1.2002,
+      "step": 75790
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.522129717990062e-05,
+      "loss": 1.4125,
+      "step": 75800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5218006515515485e-05,
+      "loss": 1.222,
+      "step": 75810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5214715851130342e-05,
+      "loss": 1.2363,
+      "step": 75820
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5211425186745207e-05,
+      "loss": 1.3426,
+      "step": 75830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5208134522360065e-05,
+      "loss": 1.2221,
+      "step": 75840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5204843857974926e-05,
+      "loss": 1.1748,
+      "step": 75850
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5201553193589788e-05,
+      "loss": 1.2864,
+      "step": 75860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5198262529204645e-05,
+      "loss": 1.207,
+      "step": 75870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.519497186481951e-05,
+      "loss": 1.21,
+      "step": 75880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5191681200434368e-05,
+      "loss": 1.319,
+      "step": 75890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.518839053604923e-05,
+      "loss": 1.1981,
+      "step": 75900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.518509987166409e-05,
+      "loss": 1.2528,
+      "step": 75910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5181809207278952e-05,
+      "loss": 1.4223,
+      "step": 75920
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5178518542893813e-05,
+      "loss": 1.1612,
+      "step": 75930
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5175227878508674e-05,
+      "loss": 1.1254,
+      "step": 75940
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5171937214123532e-05,
+      "loss": 1.2691,
+      "step": 75950
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5168646549738393e-05,
+      "loss": 1.2227,
+      "step": 75960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5165355885353255e-05,
+      "loss": 1.244,
+      "step": 75970
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5162065220968116e-05,
+      "loss": 1.2617,
+      "step": 75980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5158774556582977e-05,
+      "loss": 1.1739,
+      "step": 75990
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5155483892197835e-05,
+      "loss": 1.3845,
+      "step": 76000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.51521932278127e-05,
+      "loss": 1.2815,
+      "step": 76010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5148902563427558e-05,
+      "loss": 1.2881,
+      "step": 76020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.514561189904242e-05,
+      "loss": 1.3547,
+      "step": 76030
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.514232123465728e-05,
+      "loss": 1.2765,
+      "step": 76040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5139030570272138e-05,
+      "loss": 1.3405,
+      "step": 76050
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5135739905887003e-05,
+      "loss": 1.3458,
+      "step": 76060
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.513244924150186e-05,
+      "loss": 1.3756,
+      "step": 76070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5129158577116725e-05,
+      "loss": 1.2273,
+      "step": 76080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5125867912731583e-05,
+      "loss": 1.3006,
+      "step": 76090
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.512257724834644e-05,
+      "loss": 1.2289,
+      "step": 76100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5119286583961306e-05,
+      "loss": 1.1361,
+      "step": 76110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5115995919576164e-05,
+      "loss": 1.3331,
+      "step": 76120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5112705255191028e-05,
+      "loss": 1.5182,
+      "step": 76130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5109414590805886e-05,
+      "loss": 1.3368,
+      "step": 76140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5106123926420744e-05,
+      "loss": 1.2505,
+      "step": 76150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.510283326203561e-05,
+      "loss": 1.3622,
+      "step": 76160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5099542597650466e-05,
+      "loss": 1.2234,
+      "step": 76170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.509625193326533e-05,
+      "loss": 1.2127,
+      "step": 76180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.509296126888019e-05,
+      "loss": 1.0289,
+      "step": 76190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5089670604495047e-05,
+      "loss": 1.3366,
+      "step": 76200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.508637994010991e-05,
+      "loss": 1.1503,
+      "step": 76210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.508308927572477e-05,
+      "loss": 1.643,
+      "step": 76220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5079798611339634e-05,
+      "loss": 1.1234,
+      "step": 76230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5076507946954492e-05,
+      "loss": 1.4009,
+      "step": 76240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.507321728256935e-05,
+      "loss": 1.328,
+      "step": 76250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5069926618184215e-05,
+      "loss": 1.2409,
+      "step": 76260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5066635953799072e-05,
+      "loss": 1.3472,
+      "step": 76270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5063345289413937e-05,
+      "loss": 1.386,
+      "step": 76280
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5060054625028795e-05,
+      "loss": 1.2409,
+      "step": 76290
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5056763960643653e-05,
+      "loss": 1.2022,
+      "step": 76300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5053473296258517e-05,
+      "loss": 1.2767,
+      "step": 76310
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5050182631873375e-05,
+      "loss": 1.2095,
+      "step": 76320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.504689196748824e-05,
+      "loss": 1.3941,
+      "step": 76330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5043601303103098e-05,
+      "loss": 1.2614,
+      "step": 76340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5040310638717956e-05,
+      "loss": 1.2771,
+      "step": 76350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.503701997433282e-05,
+      "loss": 1.2289,
+      "step": 76360
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5033729309947678e-05,
+      "loss": 1.4576,
+      "step": 76370
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5030438645562543e-05,
+      "loss": 1.2574,
+      "step": 76380
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.50271479811774e-05,
+      "loss": 1.1962,
+      "step": 76390
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.502385731679226e-05,
+      "loss": 1.2246,
+      "step": 76400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5020566652407123e-05,
+      "loss": 1.2586,
+      "step": 76410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.501727598802198e-05,
+      "loss": 1.383,
+      "step": 76420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5013985323636846e-05,
+      "loss": 1.1504,
+      "step": 76430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5010694659251704e-05,
+      "loss": 1.3241,
+      "step": 76440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.500740399486656e-05,
+      "loss": 1.3602,
+      "step": 76450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5004113330481426e-05,
+      "loss": 1.2135,
+      "step": 76460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5000822666096284e-05,
+      "loss": 1.226,
+      "step": 76470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4997532001711145e-05,
+      "loss": 1.1127,
+      "step": 76480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4994241337326007e-05,
+      "loss": 1.2361,
+      "step": 76490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4990950672940868e-05,
+      "loss": 1.2412,
+      "step": 76500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.498766000855573e-05,
+      "loss": 0.9849,
+      "step": 76510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4984369344170587e-05,
+      "loss": 1.0385,
+      "step": 76520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.498107867978545e-05,
+      "loss": 1.2313,
+      "step": 76530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.497778801540031e-05,
+      "loss": 1.4055,
+      "step": 76540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.497449735101517e-05,
+      "loss": 1.5855,
+      "step": 76550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4971206686630032e-05,
+      "loss": 1.2516,
+      "step": 76560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.496791602224489e-05,
+      "loss": 1.2192,
+      "step": 76570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.496462535785975e-05,
+      "loss": 1.2661,
+      "step": 76580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4961334693474613e-05,
+      "loss": 1.3832,
+      "step": 76590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4958044029089474e-05,
+      "loss": 1.084,
+      "step": 76600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4954753364704335e-05,
+      "loss": 1.6262,
+      "step": 76610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4951462700319196e-05,
+      "loss": 1.3968,
+      "step": 76620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4948172035934054e-05,
+      "loss": 1.2271,
+      "step": 76630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4944881371548916e-05,
+      "loss": 1.4918,
+      "step": 76640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4941590707163777e-05,
+      "loss": 1.4059,
+      "step": 76650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4938300042778638e-05,
+      "loss": 1.2658,
+      "step": 76660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.49350093783935e-05,
+      "loss": 1.2485,
+      "step": 76670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.493171871400836e-05,
+      "loss": 1.2694,
+      "step": 76680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4928428049623222e-05,
+      "loss": 1.3983,
+      "step": 76690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4925137385238083e-05,
+      "loss": 1.1279,
+      "step": 76700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.492184672085294e-05,
+      "loss": 1.24,
+      "step": 76710
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4918556056467802e-05,
+      "loss": 1.2502,
+      "step": 76720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4915265392082664e-05,
+      "loss": 1.2773,
+      "step": 76730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4911974727697525e-05,
+      "loss": 1.2889,
+      "step": 76740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4908684063312386e-05,
+      "loss": 1.3707,
+      "step": 76750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4905393398927247e-05,
+      "loss": 1.1777,
+      "step": 76760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4902102734542105e-05,
+      "loss": 1.3221,
+      "step": 76770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4898812070156967e-05,
+      "loss": 1.1944,
+      "step": 76780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4895521405771828e-05,
+      "loss": 1.1898,
+      "step": 76790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.489223074138669e-05,
+      "loss": 1.0773,
+      "step": 76800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.488894007700155e-05,
+      "loss": 1.2678,
+      "step": 76810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4885649412616408e-05,
+      "loss": 1.3656,
+      "step": 76820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.488235874823127e-05,
+      "loss": 1.313,
+      "step": 76830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.487906808384613e-05,
+      "loss": 1.1819,
+      "step": 76840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4875777419460992e-05,
+      "loss": 1.4622,
+      "step": 76850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4872486755075853e-05,
+      "loss": 1.3704,
+      "step": 76860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.486919609069071e-05,
+      "loss": 1.34,
+      "step": 76870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4865905426305572e-05,
+      "loss": 1.1626,
+      "step": 76880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4862614761920434e-05,
+      "loss": 1.2836,
+      "step": 76890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4859324097535295e-05,
+      "loss": 1.2251,
+      "step": 76900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4856033433150156e-05,
+      "loss": 1.2043,
+      "step": 76910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4852742768765014e-05,
+      "loss": 1.4293,
+      "step": 76920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4849452104379875e-05,
+      "loss": 1.194,
+      "step": 76930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4846161439994737e-05,
+      "loss": 1.1676,
+      "step": 76940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4842870775609598e-05,
+      "loss": 1.6936,
+      "step": 76950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.483958011122446e-05,
+      "loss": 1.3644,
+      "step": 76960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4836289446839317e-05,
+      "loss": 1.2192,
+      "step": 76970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.483299878245418e-05,
+      "loss": 1.3545,
+      "step": 76980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.482970811806904e-05,
+      "loss": 1.1118,
+      "step": 76990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.48264174536839e-05,
+      "loss": 1.3983,
+      "step": 77000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4823126789298762e-05,
+      "loss": 1.3372,
+      "step": 77010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.481983612491362e-05,
+      "loss": 1.2,
+      "step": 77020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.481654546052848e-05,
+      "loss": 1.167,
+      "step": 77030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4813254796143343e-05,
+      "loss": 1.2868,
+      "step": 77040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4809964131758204e-05,
+      "loss": 1.2078,
+      "step": 77050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4806673467373065e-05,
+      "loss": 1.3141,
+      "step": 77060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4803382802987923e-05,
+      "loss": 1.441,
+      "step": 77070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4800092138602784e-05,
+      "loss": 1.3323,
+      "step": 77080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4796801474217646e-05,
+      "loss": 1.3261,
+      "step": 77090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4793510809832507e-05,
+      "loss": 1.2671,
+      "step": 77100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4790220145447368e-05,
+      "loss": 1.2423,
+      "step": 77110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4786929481062226e-05,
+      "loss": 1.2501,
+      "step": 77120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4783638816677087e-05,
+      "loss": 1.32,
+      "step": 77130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.478034815229195e-05,
+      "loss": 1.1786,
+      "step": 77140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.477705748790681e-05,
+      "loss": 1.1902,
+      "step": 77150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.477376682352167e-05,
+      "loss": 1.2282,
+      "step": 77160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.477047615913653e-05,
+      "loss": 1.3349,
+      "step": 77170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.476718549475139e-05,
+      "loss": 1.3425,
+      "step": 77180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.476389483036625e-05,
+      "loss": 1.3328,
+      "step": 77190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4760604165981113e-05,
+      "loss": 1.2615,
+      "step": 77200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4757313501595974e-05,
+      "loss": 1.1956,
+      "step": 77210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4754022837210832e-05,
+      "loss": 1.3244,
+      "step": 77220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4750732172825693e-05,
+      "loss": 1.3673,
+      "step": 77230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4747441508440554e-05,
+      "loss": 1.1641,
+      "step": 77240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4744150844055416e-05,
+      "loss": 1.2668,
+      "step": 77250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4740860179670277e-05,
+      "loss": 1.4012,
+      "step": 77260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4737569515285135e-05,
+      "loss": 1.1393,
+      "step": 77270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4734278850899996e-05,
+      "loss": 1.2567,
+      "step": 77280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4730988186514857e-05,
+      "loss": 1.148,
+      "step": 77290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.472769752212972e-05,
+      "loss": 1.18,
+      "step": 77300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.472440685774458e-05,
+      "loss": 1.2015,
+      "step": 77310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4721116193359438e-05,
+      "loss": 1.1631,
+      "step": 77320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.47178255289743e-05,
+      "loss": 1.336,
+      "step": 77330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.471453486458916e-05,
+      "loss": 1.105,
+      "step": 77340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.471124420020402e-05,
+      "loss": 1.328,
+      "step": 77350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4707953535818883e-05,
+      "loss": 1.2932,
+      "step": 77360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4704662871433744e-05,
+      "loss": 1.3407,
+      "step": 77370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4701372207048605e-05,
+      "loss": 1.4071,
+      "step": 77380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4698081542663463e-05,
+      "loss": 1.4758,
+      "step": 77390
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4694790878278324e-05,
+      "loss": 1.4368,
+      "step": 77400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4691500213893186e-05,
+      "loss": 1.262,
+      "step": 77410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4688209549508047e-05,
+      "loss": 1.5797,
+      "step": 77420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4684918885122908e-05,
+      "loss": 1.2782,
+      "step": 77430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.468162822073777e-05,
+      "loss": 1.2399,
+      "step": 77440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.467833755635263e-05,
+      "loss": 1.3826,
+      "step": 77450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4675046891967492e-05,
+      "loss": 1.2401,
+      "step": 77460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.467175622758235e-05,
+      "loss": 1.3009,
+      "step": 77470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.466846556319721e-05,
+      "loss": 1.3391,
+      "step": 77480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4665174898812072e-05,
+      "loss": 1.2078,
+      "step": 77490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4661884234426934e-05,
+      "loss": 1.2981,
+      "step": 77500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4658593570041795e-05,
+      "loss": 1.258,
+      "step": 77510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4655302905656653e-05,
+      "loss": 1.1886,
+      "step": 77520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4652012241271514e-05,
+      "loss": 1.3101,
+      "step": 77530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4648721576886375e-05,
+      "loss": 1.2013,
+      "step": 77540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4645430912501237e-05,
+      "loss": 1.3462,
+      "step": 77550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4642140248116098e-05,
+      "loss": 1.2132,
+      "step": 77560
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4638849583730956e-05,
+      "loss": 1.2404,
+      "step": 77570
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4635558919345817e-05,
+      "loss": 1.2009,
+      "step": 77580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.463226825496068e-05,
+      "loss": 1.3704,
+      "step": 77590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.462897759057554e-05,
+      "loss": 1.1843,
+      "step": 77600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.46256869261904e-05,
+      "loss": 1.2748,
+      "step": 77610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.462239626180526e-05,
+      "loss": 1.1448,
+      "step": 77620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.461910559742012e-05,
+      "loss": 1.7449,
+      "step": 77630
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.461581493303498e-05,
+      "loss": 1.2478,
+      "step": 77640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4612524268649843e-05,
+      "loss": 1.2763,
+      "step": 77650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4609233604264704e-05,
+      "loss": 1.251,
+      "step": 77660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4605942939879562e-05,
+      "loss": 1.1718,
+      "step": 77670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4602652275494423e-05,
+      "loss": 1.3671,
+      "step": 77680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4599361611109284e-05,
+      "loss": 1.3123,
+      "step": 77690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4596070946724146e-05,
+      "loss": 1.1501,
+      "step": 77700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4592780282339007e-05,
+      "loss": 1.2375,
+      "step": 77710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4589489617953865e-05,
+      "loss": 1.1462,
+      "step": 77720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4586198953568726e-05,
+      "loss": 1.153,
+      "step": 77730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4582908289183587e-05,
+      "loss": 1.1197,
+      "step": 77740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.457961762479845e-05,
+      "loss": 1.1911,
+      "step": 77750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.457632696041331e-05,
+      "loss": 1.1567,
+      "step": 77760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4573036296028168e-05,
+      "loss": 1.3904,
+      "step": 77770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.456974563164303e-05,
+      "loss": 1.1856,
+      "step": 77780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.456645496725789e-05,
+      "loss": 1.403,
+      "step": 77790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.456316430287275e-05,
+      "loss": 1.3377,
+      "step": 77800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4559873638487613e-05,
+      "loss": 1.1315,
+      "step": 77810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.455658297410247e-05,
+      "loss": 1.1896,
+      "step": 77820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4553292309717332e-05,
+      "loss": 1.2286,
+      "step": 77830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4550001645332193e-05,
+      "loss": 1.336,
+      "step": 77840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4546710980947054e-05,
+      "loss": 1.2522,
+      "step": 77850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4543420316561916e-05,
+      "loss": 1.2544,
+      "step": 77860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4540129652176774e-05,
+      "loss": 1.2049,
+      "step": 77870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4536838987791635e-05,
+      "loss": 1.5102,
+      "step": 77880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4533548323406496e-05,
+      "loss": 1.4626,
+      "step": 77890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4530257659021357e-05,
+      "loss": 1.1932,
+      "step": 77900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.452696699463622e-05,
+      "loss": 1.2484,
+      "step": 77910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.452367633025108e-05,
+      "loss": 1.416,
+      "step": 77920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4520385665865938e-05,
+      "loss": 1.3461,
+      "step": 77930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.45170950014808e-05,
+      "loss": 1.2537,
+      "step": 77940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.451380433709566e-05,
+      "loss": 1.2196,
+      "step": 77950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.451051367271052e-05,
+      "loss": 1.285,
+      "step": 77960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4507223008325383e-05,
+      "loss": 1.2462,
+      "step": 77970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.450393234394024e-05,
+      "loss": 1.4681,
+      "step": 77980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4500641679555102e-05,
+      "loss": 1.3526,
+      "step": 77990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4497351015169963e-05,
+      "loss": 1.1997,
+      "step": 78000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4494060350784825e-05,
+      "loss": 1.2563,
+      "step": 78010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4490769686399686e-05,
+      "loss": 1.368,
+      "step": 78020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4487479022014544e-05,
+      "loss": 1.1868,
+      "step": 78030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4484188357629405e-05,
+      "loss": 1.1536,
+      "step": 78040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4480897693244266e-05,
+      "loss": 1.3832,
+      "step": 78050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4477607028859127e-05,
+      "loss": 1.2341,
+      "step": 78060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.447431636447399e-05,
+      "loss": 1.2761,
+      "step": 78070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4471025700088847e-05,
+      "loss": 1.1706,
+      "step": 78080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4467735035703708e-05,
+      "loss": 1.2601,
+      "step": 78090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.446444437131857e-05,
+      "loss": 1.3569,
+      "step": 78100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.446115370693343e-05,
+      "loss": 1.322,
+      "step": 78110
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.445786304254829e-05,
+      "loss": 1.3317,
+      "step": 78120
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4454572378163153e-05,
+      "loss": 1.1483,
+      "step": 78130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4451281713778014e-05,
+      "loss": 1.311,
+      "step": 78140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4447991049392872e-05,
+      "loss": 1.1814,
+      "step": 78150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4444700385007733e-05,
+      "loss": 1.2747,
+      "step": 78160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4441409720622595e-05,
+      "loss": 1.3197,
+      "step": 78170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4438119056237456e-05,
+      "loss": 1.3047,
+      "step": 78180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4434828391852317e-05,
+      "loss": 1.1718,
+      "step": 78190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.443153772746718e-05,
+      "loss": 1.3414,
+      "step": 78200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.442824706308204e-05,
+      "loss": 1.2461,
+      "step": 78210
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.44249563986969e-05,
+      "loss": 1.2365,
+      "step": 78220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.442166573431176e-05,
+      "loss": 1.0765,
+      "step": 78230
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.441837506992662e-05,
+      "loss": 1.1171,
+      "step": 78240
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.441508440554148e-05,
+      "loss": 1.2959,
+      "step": 78250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4411793741156343e-05,
+      "loss": 1.2906,
+      "step": 78260
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4408503076771204e-05,
+      "loss": 1.2532,
+      "step": 78270
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4405212412386062e-05,
+      "loss": 1.2121,
+      "step": 78280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4401921748000923e-05,
+      "loss": 1.2904,
+      "step": 78290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4398631083615784e-05,
+      "loss": 1.3462,
+      "step": 78300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4395340419230646e-05,
+      "loss": 1.285,
+      "step": 78310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4392049754845507e-05,
+      "loss": 1.2196,
+      "step": 78320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4388759090460365e-05,
+      "loss": 1.278,
+      "step": 78330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4385468426075226e-05,
+      "loss": 1.3399,
+      "step": 78340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4382177761690087e-05,
+      "loss": 1.2707,
+      "step": 78350
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.437888709730495e-05,
+      "loss": 1.1852,
+      "step": 78360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.437559643291981e-05,
+      "loss": 1.2594,
+      "step": 78370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4372305768534668e-05,
+      "loss": 1.3188,
+      "step": 78380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.436901510414953e-05,
+      "loss": 1.3368,
+      "step": 78390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.436572443976439e-05,
+      "loss": 1.1734,
+      "step": 78400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.436243377537925e-05,
+      "loss": 1.4126,
+      "step": 78410
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4359143110994113e-05,
+      "loss": 1.365,
+      "step": 78420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.435585244660897e-05,
+      "loss": 1.3466,
+      "step": 78430
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4352561782223832e-05,
+      "loss": 1.415,
+      "step": 78440
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4349271117838693e-05,
+      "loss": 1.2768,
+      "step": 78450
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4345980453453554e-05,
+      "loss": 1.2316,
+      "step": 78460
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4342689789068416e-05,
+      "loss": 1.2778,
+      "step": 78470
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4339399124683274e-05,
+      "loss": 1.262,
+      "step": 78480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4336108460298135e-05,
+      "loss": 1.2729,
+      "step": 78490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4332817795912996e-05,
+      "loss": 1.3722,
+      "step": 78500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4329527131527857e-05,
+      "loss": 1.3156,
+      "step": 78510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.432623646714272e-05,
+      "loss": 1.2445,
+      "step": 78520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4322945802757577e-05,
+      "loss": 1.066,
+      "step": 78530
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4319655138372438e-05,
+      "loss": 1.221,
+      "step": 78540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.43163644739873e-05,
+      "loss": 1.17,
+      "step": 78550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.431307380960216e-05,
+      "loss": 1.3622,
+      "step": 78560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.430978314521702e-05,
+      "loss": 1.2487,
+      "step": 78570
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.430649248083188e-05,
+      "loss": 1.2275,
+      "step": 78580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.430320181644674e-05,
+      "loss": 1.2411,
+      "step": 78590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4299911152061602e-05,
+      "loss": 1.2059,
+      "step": 78600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4296620487676463e-05,
+      "loss": 1.4271,
+      "step": 78610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4293329823291325e-05,
+      "loss": 1.469,
+      "step": 78620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4290039158906182e-05,
+      "loss": 1.2835,
+      "step": 78630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4286748494521044e-05,
+      "loss": 1.1503,
+      "step": 78640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4283457830135905e-05,
+      "loss": 1.3285,
+      "step": 78650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4280167165750766e-05,
+      "loss": 1.4427,
+      "step": 78660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4276876501365628e-05,
+      "loss": 1.2095,
+      "step": 78670
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4273585836980485e-05,
+      "loss": 1.3254,
+      "step": 78680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4270295172595347e-05,
+      "loss": 1.3713,
+      "step": 78690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4267004508210208e-05,
+      "loss": 1.3749,
+      "step": 78700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.426371384382507e-05,
+      "loss": 1.3567,
+      "step": 78710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.426042317943993e-05,
+      "loss": 1.186,
+      "step": 78720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.425713251505479e-05,
+      "loss": 1.3583,
+      "step": 78730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.425384185066965e-05,
+      "loss": 1.2293,
+      "step": 78740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.425055118628451e-05,
+      "loss": 1.3459,
+      "step": 78750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4247260521899372e-05,
+      "loss": 1.3203,
+      "step": 78760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4243969857514233e-05,
+      "loss": 1.2264,
+      "step": 78770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.424067919312909e-05,
+      "loss": 1.4215,
+      "step": 78780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4237388528743953e-05,
+      "loss": 1.1561,
+      "step": 78790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4234097864358814e-05,
+      "loss": 1.1379,
+      "step": 78800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4230807199973675e-05,
+      "loss": 1.1216,
+      "step": 78810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4227516535588536e-05,
+      "loss": 1.3016,
+      "step": 78820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4224225871203398e-05,
+      "loss": 1.2352,
+      "step": 78830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4220935206818255e-05,
+      "loss": 1.2385,
+      "step": 78840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4217644542433117e-05,
+      "loss": 1.4568,
+      "step": 78850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4214353878047978e-05,
+      "loss": 1.1967,
+      "step": 78860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.421106321366284e-05,
+      "loss": 1.3649,
+      "step": 78870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.42077725492777e-05,
+      "loss": 1.0977,
+      "step": 78880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4204481884892562e-05,
+      "loss": 1.3042,
+      "step": 78890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4201191220507423e-05,
+      "loss": 1.1093,
+      "step": 78900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4197900556122284e-05,
+      "loss": 1.4151,
+      "step": 78910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4194609891737142e-05,
+      "loss": 1.2238,
+      "step": 78920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4191319227352004e-05,
+      "loss": 1.1886,
+      "step": 78930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4188028562966865e-05,
+      "loss": 1.235,
+      "step": 78940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4184737898581726e-05,
+      "loss": 1.3,
+      "step": 78950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4181447234196587e-05,
+      "loss": 1.2128,
+      "step": 78960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.417815656981145e-05,
+      "loss": 1.3196,
+      "step": 78970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4174865905426306e-05,
+      "loss": 1.323,
+      "step": 78980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4171575241041168e-05,
+      "loss": 1.2926,
+      "step": 78990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.416828457665603e-05,
+      "loss": 1.179,
+      "step": 79000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.416499391227089e-05,
+      "loss": 1.1856,
+      "step": 79010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.416170324788575e-05,
+      "loss": 1.3449,
+      "step": 79020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.415841258350061e-05,
+      "loss": 1.4428,
+      "step": 79030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.415512191911547e-05,
+      "loss": 1.2781,
+      "step": 79040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4151831254730332e-05,
+      "loss": 1.1749,
+      "step": 79050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4148540590345193e-05,
+      "loss": 1.2046,
+      "step": 79060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4145249925960054e-05,
+      "loss": 1.3749,
+      "step": 79070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4141959261574912e-05,
+      "loss": 1.2443,
+      "step": 79080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4138668597189774e-05,
+      "loss": 1.1939,
+      "step": 79090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4135377932804635e-05,
+      "loss": 1.0937,
+      "step": 79100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4132087268419496e-05,
+      "loss": 1.4026,
+      "step": 79110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4128796604034357e-05,
+      "loss": 1.3605,
+      "step": 79120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4125505939649215e-05,
+      "loss": 1.4986,
+      "step": 79130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4122215275264077e-05,
+      "loss": 1.3113,
+      "step": 79140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4118924610878938e-05,
+      "loss": 1.2332,
+      "step": 79150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.41156339464938e-05,
+      "loss": 1.3638,
+      "step": 79160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.411234328210866e-05,
+      "loss": 1.2188,
+      "step": 79170
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4109052617723518e-05,
+      "loss": 1.4899,
+      "step": 79180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410576195333838e-05,
+      "loss": 1.2221,
+      "step": 79190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410247128895324e-05,
+      "loss": 1.3928,
+      "step": 79200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4099180624568102e-05,
+      "loss": 1.3394,
+      "step": 79210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4095889960182963e-05,
+      "loss": 1.253,
+      "step": 79220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.409259929579782e-05,
+      "loss": 1.3157,
+      "step": 79230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4089308631412682e-05,
+      "loss": 1.0606,
+      "step": 79240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4086017967027544e-05,
+      "loss": 1.2021,
+      "step": 79250
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4082727302642405e-05,
+      "loss": 1.3674,
+      "step": 79260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4079436638257266e-05,
+      "loss": 1.325,
+      "step": 79270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4076145973872124e-05,
+      "loss": 1.3635,
+      "step": 79280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4072855309486985e-05,
+      "loss": 1.3068,
+      "step": 79290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4069564645101847e-05,
+      "loss": 1.1527,
+      "step": 79300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4066273980716708e-05,
+      "loss": 1.2296,
+      "step": 79310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.406298331633157e-05,
+      "loss": 1.2565,
+      "step": 79320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.405969265194643e-05,
+      "loss": 1.1686,
+      "step": 79330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.405640198756129e-05,
+      "loss": 1.1858,
+      "step": 79340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.405311132317615e-05,
+      "loss": 1.3286,
+      "step": 79350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.404982065879101e-05,
+      "loss": 1.218,
+      "step": 79360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4046529994405872e-05,
+      "loss": 1.2961,
+      "step": 79370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4043239330020733e-05,
+      "loss": 1.2897,
+      "step": 79380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.403994866563559e-05,
+      "loss": 1.3884,
+      "step": 79390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4036658001250453e-05,
+      "loss": 1.4475,
+      "step": 79400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4033367336865314e-05,
+      "loss": 1.2455,
+      "step": 79410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4030076672480175e-05,
+      "loss": 1.2566,
+      "step": 79420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4026786008095036e-05,
+      "loss": 1.4259,
+      "step": 79430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4023495343709894e-05,
+      "loss": 1.2817,
+      "step": 79440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4020204679324756e-05,
+      "loss": 1.1578,
+      "step": 79450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4016914014939617e-05,
+      "loss": 1.2884,
+      "step": 79460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4013623350554478e-05,
+      "loss": 1.3041,
+      "step": 79470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.401033268616934e-05,
+      "loss": 1.4683,
+      "step": 79480
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4007042021784197e-05,
+      "loss": 1.243,
+      "step": 79490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.400375135739906e-05,
+      "loss": 1.2807,
+      "step": 79500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.400046069301392e-05,
+      "loss": 1.1162,
+      "step": 79510
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.399717002862878e-05,
+      "loss": 1.3714,
+      "step": 79520
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3993879364243642e-05,
+      "loss": 1.2695,
+      "step": 79530
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.39905886998585e-05,
+      "loss": 1.3564,
+      "step": 79540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.398729803547336e-05,
+      "loss": 1.1844,
+      "step": 79550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3984007371088223e-05,
+      "loss": 1.1798,
+      "step": 79560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3980716706703084e-05,
+      "loss": 1.1905,
+      "step": 79570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3977426042317945e-05,
+      "loss": 1.1376,
+      "step": 79580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3974135377932807e-05,
+      "loss": 1.2771,
+      "step": 79590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3970844713547664e-05,
+      "loss": 1.3661,
+      "step": 79600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3967554049162526e-05,
+      "loss": 1.2994,
+      "step": 79610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3964263384777387e-05,
+      "loss": 1.3105,
+      "step": 79620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3960972720392248e-05,
+      "loss": 1.276,
+      "step": 79630
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.395768205600711e-05,
+      "loss": 1.1645,
+      "step": 79640
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.395439139162197e-05,
+      "loss": 1.2111,
+      "step": 79650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3951100727236832e-05,
+      "loss": 1.2382,
+      "step": 79660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3947810062851693e-05,
+      "loss": 1.2525,
+      "step": 79670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.394451939846655e-05,
+      "loss": 1.2759,
+      "step": 79680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3941228734081412e-05,
+      "loss": 1.3695,
+      "step": 79690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3937938069696274e-05,
+      "loss": 1.2237,
+      "step": 79700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3934647405311135e-05,
+      "loss": 1.354,
+      "step": 79710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3931356740925996e-05,
+      "loss": 1.2821,
+      "step": 79720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3928066076540857e-05,
+      "loss": 1.3103,
+      "step": 79730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3924775412155715e-05,
+      "loss": 1.2242,
+      "step": 79740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3921484747770577e-05,
+      "loss": 1.3671,
+      "step": 79750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3918194083385438e-05,
+      "loss": 1.2834,
+      "step": 79760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.39149034190003e-05,
+      "loss": 1.1708,
+      "step": 79770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.391161275461516e-05,
+      "loss": 1.2098,
+      "step": 79780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.390832209023002e-05,
+      "loss": 1.3765,
+      "step": 79790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.390503142584488e-05,
+      "loss": 1.2407,
+      "step": 79800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.390174076145974e-05,
+      "loss": 1.1188,
+      "step": 79810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3898450097074602e-05,
+      "loss": 1.2379,
+      "step": 79820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3895159432689463e-05,
+      "loss": 1.2155,
+      "step": 79830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.389186876830432e-05,
+      "loss": 1.2446,
+      "step": 79840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3888578103919183e-05,
+      "loss": 1.1518,
+      "step": 79850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3885287439534044e-05,
+      "loss": 1.2215,
+      "step": 79860
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3881996775148905e-05,
+      "loss": 1.2828,
+      "step": 79870
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3878706110763766e-05,
+      "loss": 1.3469,
+      "step": 79880
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3875415446378624e-05,
+      "loss": 1.4182,
+      "step": 79890
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3872124781993485e-05,
+      "loss": 1.1451,
+      "step": 79900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3868834117608347e-05,
+      "loss": 1.1846,
+      "step": 79910
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3865543453223208e-05,
+      "loss": 1.5929,
+      "step": 79920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.386225278883807e-05,
+      "loss": 1.2735,
+      "step": 79930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3858962124452927e-05,
+      "loss": 1.2534,
+      "step": 79940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.385567146006779e-05,
+      "loss": 1.0985,
+      "step": 79950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.385238079568265e-05,
+      "loss": 1.3375,
+      "step": 79960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.384909013129751e-05,
+      "loss": 1.3015,
+      "step": 79970
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3845799466912372e-05,
+      "loss": 1.1702,
+      "step": 79980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.384250880252723e-05,
+      "loss": 1.292,
+      "step": 79990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.383921813814209e-05,
+      "loss": 1.0748,
+      "step": 80000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3835927473756953e-05,
+      "loss": 1.1698,
+      "step": 80010
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3832636809371814e-05,
+      "loss": 1.123,
+      "step": 80020
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3829346144986675e-05,
+      "loss": 1.2155,
+      "step": 80030
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3826055480601533e-05,
+      "loss": 1.3512,
+      "step": 80040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3822764816216394e-05,
+      "loss": 1.3898,
+      "step": 80050
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3819474151831256e-05,
+      "loss": 1.3678,
+      "step": 80060
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3816183487446117e-05,
+      "loss": 1.2722,
+      "step": 80070
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3812892823060978e-05,
+      "loss": 1.2196,
+      "step": 80080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3809602158675836e-05,
+      "loss": 1.2246,
+      "step": 80090
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3806311494290697e-05,
+      "loss": 1.225,
+      "step": 80100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.380302082990556e-05,
+      "loss": 1.37,
+      "step": 80110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.379973016552042e-05,
+      "loss": 1.1111,
+      "step": 80120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.379643950113528e-05,
+      "loss": 1.271,
+      "step": 80130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.379314883675014e-05,
+      "loss": 1.1801,
+      "step": 80140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3789858172365e-05,
+      "loss": 1.3039,
+      "step": 80150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.378656750797986e-05,
+      "loss": 1.1077,
+      "step": 80160
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3783276843594723e-05,
+      "loss": 1.2673,
+      "step": 80170
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3779986179209584e-05,
+      "loss": 1.1173,
+      "step": 80180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3776695514824442e-05,
+      "loss": 1.1502,
+      "step": 80190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3773404850439303e-05,
+      "loss": 1.3557,
+      "step": 80200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3770114186054164e-05,
+      "loss": 1.3254,
+      "step": 80210
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3766823521669026e-05,
+      "loss": 1.1208,
+      "step": 80220
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3763532857283887e-05,
+      "loss": 1.2605,
+      "step": 80230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3760242192898745e-05,
+      "loss": 1.3102,
+      "step": 80240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3756951528513606e-05,
+      "loss": 1.1411,
+      "step": 80250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3753660864128467e-05,
+      "loss": 1.3375,
+      "step": 80260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.375037019974333e-05,
+      "loss": 1.1708,
+      "step": 80270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.374707953535819e-05,
+      "loss": 1.3062,
+      "step": 80280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3743788870973048e-05,
+      "loss": 1.1835,
+      "step": 80290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.374049820658791e-05,
+      "loss": 1.1652,
+      "step": 80300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.373720754220277e-05,
+      "loss": 1.1929,
+      "step": 80310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.373391687781763e-05,
+      "loss": 1.2797,
+      "step": 80320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3730626213432493e-05,
+      "loss": 1.1459,
+      "step": 80330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3727335549047354e-05,
+      "loss": 1.22,
+      "step": 80340
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3724044884662215e-05,
+      "loss": 1.2403,
+      "step": 80350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3720754220277073e-05,
+      "loss": 1.3275,
+      "step": 80360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3717463555891935e-05,
+      "loss": 1.2146,
+      "step": 80370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3714172891506796e-05,
+      "loss": 1.1993,
+      "step": 80380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3710882227121657e-05,
+      "loss": 1.2458,
+      "step": 80390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.370759156273652e-05,
+      "loss": 1.2448,
+      "step": 80400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.370430089835138e-05,
+      "loss": 1.2423,
+      "step": 80410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.370101023396624e-05,
+      "loss": 1.2843,
+      "step": 80420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3697719569581102e-05,
+      "loss": 1.2806,
+      "step": 80430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.369442890519596e-05,
+      "loss": 1.2023,
+      "step": 80440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.369113824081082e-05,
+      "loss": 1.2316,
+      "step": 80450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3687847576425683e-05,
+      "loss": 1.1195,
+      "step": 80460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3684556912040544e-05,
+      "loss": 1.3362,
+      "step": 80470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3681266247655405e-05,
+      "loss": 1.1953,
+      "step": 80480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3677975583270263e-05,
+      "loss": 1.2563,
+      "step": 80490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3674684918885124e-05,
+      "loss": 1.3352,
+      "step": 80500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3671394254499986e-05,
+      "loss": 1.2536,
+      "step": 80510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3668103590114847e-05,
+      "loss": 1.319,
+      "step": 80520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3664812925729708e-05,
+      "loss": 1.3417,
+      "step": 80530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3661522261344566e-05,
+      "loss": 1.3251,
+      "step": 80540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3658231596959427e-05,
+      "loss": 1.2554,
+      "step": 80550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.365494093257429e-05,
+      "loss": 1.3978,
+      "step": 80560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.365165026818915e-05,
+      "loss": 1.2459,
+      "step": 80570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.364835960380401e-05,
+      "loss": 1.4007,
+      "step": 80580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.364506893941887e-05,
+      "loss": 1.3462,
+      "step": 80590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.364177827503373e-05,
+      "loss": 1.3178,
+      "step": 80600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.363848761064859e-05,
+      "loss": 1.4742,
+      "step": 80610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3635196946263453e-05,
+      "loss": 1.2076,
+      "step": 80620
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3631906281878314e-05,
+      "loss": 1.202,
+      "step": 80630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3628615617493172e-05,
+      "loss": 1.1506,
+      "step": 80640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3625324953108033e-05,
+      "loss": 1.278,
+      "step": 80650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3622034288722894e-05,
+      "loss": 1.3676,
+      "step": 80660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3618743624337756e-05,
+      "loss": 1.475,
+      "step": 80670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3615452959952617e-05,
+      "loss": 1.2013,
+      "step": 80680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3612162295567475e-05,
+      "loss": 1.2887,
+      "step": 80690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3608871631182336e-05,
+      "loss": 1.3318,
+      "step": 80700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3605580966797197e-05,
+      "loss": 1.1602,
+      "step": 80710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.360229030241206e-05,
+      "loss": 1.2616,
+      "step": 80720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.359899963802692e-05,
+      "loss": 1.2625,
+      "step": 80730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.359570897364178e-05,
+      "loss": 1.2505,
+      "step": 80740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.359241830925664e-05,
+      "loss": 1.541,
+      "step": 80750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.35891276448715e-05,
+      "loss": 1.2272,
+      "step": 80760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.358583698048636e-05,
+      "loss": 1.2975,
+      "step": 80770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3582546316101223e-05,
+      "loss": 1.2674,
+      "step": 80780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3579255651716084e-05,
+      "loss": 1.2029,
+      "step": 80790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3575964987330942e-05,
+      "loss": 1.2864,
+      "step": 80800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3572674322945803e-05,
+      "loss": 1.2955,
+      "step": 80810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3569383658560664e-05,
+      "loss": 1.1341,
+      "step": 80820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3566092994175526e-05,
+      "loss": 1.3017,
+      "step": 80830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3562802329790387e-05,
+      "loss": 1.3608,
+      "step": 80840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3559511665405245e-05,
+      "loss": 1.2879,
+      "step": 80850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3556221001020106e-05,
+      "loss": 1.119,
+      "step": 80860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3552930336634967e-05,
+      "loss": 1.089,
+      "step": 80870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.354963967224983e-05,
+      "loss": 1.2188,
+      "step": 80880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.354634900786469e-05,
+      "loss": 1.2614,
+      "step": 80890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3543058343479548e-05,
+      "loss": 1.2764,
+      "step": 80900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.353976767909441e-05,
+      "loss": 1.3043,
+      "step": 80910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.353647701470927e-05,
+      "loss": 1.2256,
+      "step": 80920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.353318635032413e-05,
+      "loss": 1.0708,
+      "step": 80930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3529895685938993e-05,
+      "loss": 1.2083,
+      "step": 80940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.352660502155385e-05,
+      "loss": 1.2519,
+      "step": 80950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3523314357168712e-05,
+      "loss": 1.2675,
+      "step": 80960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3520023692783573e-05,
+      "loss": 1.085,
+      "step": 80970
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3516733028398435e-05,
+      "loss": 1.2054,
+      "step": 80980
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3513442364013296e-05,
+      "loss": 1.3718,
+      "step": 80990
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3510151699628154e-05,
+      "loss": 1.2049,
+      "step": 81000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3506861035243015e-05,
+      "loss": 1.3225,
+      "step": 81010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3503570370857876e-05,
+      "loss": 1.2537,
+      "step": 81020
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3500279706472738e-05,
+      "loss": 1.2927,
+      "step": 81030
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.34969890420876e-05,
+      "loss": 1.2976,
+      "step": 81040
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3493698377702457e-05,
+      "loss": 1.4376,
+      "step": 81050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3490407713317318e-05,
+      "loss": 1.3042,
+      "step": 81060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.348711704893218e-05,
+      "loss": 1.3441,
+      "step": 81070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.348382638454704e-05,
+      "loss": 1.2295,
+      "step": 81080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3480535720161902e-05,
+      "loss": 1.3378,
+      "step": 81090
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3477245055776763e-05,
+      "loss": 1.31,
+      "step": 81100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3473954391391624e-05,
+      "loss": 1.2436,
+      "step": 81110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3470663727006482e-05,
+      "loss": 1.3525,
+      "step": 81120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3467373062621343e-05,
+      "loss": 1.2501,
+      "step": 81130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3464082398236205e-05,
+      "loss": 1.1879,
+      "step": 81140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3460791733851066e-05,
+      "loss": 1.2678,
+      "step": 81150
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3457501069465927e-05,
+      "loss": 1.5776,
+      "step": 81160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.345421040508079e-05,
+      "loss": 1.0832,
+      "step": 81170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.345091974069565e-05,
+      "loss": 1.2561,
+      "step": 81180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.344762907631051e-05,
+      "loss": 1.2672,
+      "step": 81190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.344433841192537e-05,
+      "loss": 1.132,
+      "step": 81200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.344104774754023e-05,
+      "loss": 1.2481,
+      "step": 81210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.343775708315509e-05,
+      "loss": 1.1935,
+      "step": 81220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3434466418769953e-05,
+      "loss": 1.2826,
+      "step": 81230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3431175754384814e-05,
+      "loss": 1.3466,
+      "step": 81240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3427885089999672e-05,
+      "loss": 1.1977,
+      "step": 81250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3424594425614533e-05,
+      "loss": 1.3392,
+      "step": 81260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3421303761229394e-05,
+      "loss": 1.3722,
+      "step": 81270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3418013096844256e-05,
+      "loss": 1.3061,
+      "step": 81280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3414722432459117e-05,
+      "loss": 1.3812,
+      "step": 81290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3411431768073975e-05,
+      "loss": 1.3705,
+      "step": 81300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3408141103688836e-05,
+      "loss": 1.2638,
+      "step": 81310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3404850439303697e-05,
+      "loss": 1.305,
+      "step": 81320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.340155977491856e-05,
+      "loss": 1.1776,
+      "step": 81330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.339826911053342e-05,
+      "loss": 1.2972,
+      "step": 81340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3394978446148278e-05,
+      "loss": 1.3078,
+      "step": 81350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.339168778176314e-05,
+      "loss": 1.2823,
+      "step": 81360
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3388397117378e-05,
+      "loss": 1.191,
+      "step": 81370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.338510645299286e-05,
+      "loss": 1.116,
+      "step": 81380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3381815788607723e-05,
+      "loss": 1.2051,
+      "step": 81390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.337852512422258e-05,
+      "loss": 1.2664,
+      "step": 81400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3375234459837442e-05,
+      "loss": 1.2201,
+      "step": 81410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3371943795452303e-05,
+      "loss": 1.2459,
+      "step": 81420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3368653131067165e-05,
+      "loss": 1.1195,
+      "step": 81430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3365362466682026e-05,
+      "loss": 1.2624,
+      "step": 81440
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3362071802296884e-05,
+      "loss": 1.4249,
+      "step": 81450
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3358781137911745e-05,
+      "loss": 1.313,
+      "step": 81460
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3355490473526606e-05,
+      "loss": 1.3329,
+      "step": 81470
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3352199809141467e-05,
+      "loss": 1.3313,
+      "step": 81480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.334890914475633e-05,
+      "loss": 1.3569,
+      "step": 81490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3345618480371187e-05,
+      "loss": 1.2206,
+      "step": 81500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3342327815986048e-05,
+      "loss": 1.2375,
+      "step": 81510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.333903715160091e-05,
+      "loss": 1.1733,
+      "step": 81520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.333574648721577e-05,
+      "loss": 1.2252,
+      "step": 81530
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.333245582283063e-05,
+      "loss": 1.2775,
+      "step": 81540
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.332916515844549e-05,
+      "loss": 1.229,
+      "step": 81550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.332587449406035e-05,
+      "loss": 1.1391,
+      "step": 81560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3322583829675212e-05,
+      "loss": 1.1264,
+      "step": 81570
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3319293165290073e-05,
+      "loss": 1.3002,
+      "step": 81580
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3316002500904935e-05,
+      "loss": 1.268,
+      "step": 81590
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3312711836519793e-05,
+      "loss": 1.3246,
+      "step": 81600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3309421172134654e-05,
+      "loss": 1.36,
+      "step": 81610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3306130507749515e-05,
+      "loss": 1.1172,
+      "step": 81620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3302839843364376e-05,
+      "loss": 1.3067,
+      "step": 81630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3299549178979238e-05,
+      "loss": 1.385,
+      "step": 81640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3296258514594095e-05,
+      "loss": 1.225,
+      "step": 81650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3292967850208957e-05,
+      "loss": 1.267,
+      "step": 81660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3289677185823818e-05,
+      "loss": 1.2439,
+      "step": 81670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.328638652143868e-05,
+      "loss": 1.2845,
+      "step": 81680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.328309585705354e-05,
+      "loss": 1.1993,
+      "step": 81690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.32798051926684e-05,
+      "loss": 1.3232,
+      "step": 81700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.327651452828326e-05,
+      "loss": 1.2376,
+      "step": 81710
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.327322386389812e-05,
+      "loss": 1.484,
+      "step": 81720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3269933199512982e-05,
+      "loss": 1.4299,
+      "step": 81730
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3266642535127843e-05,
+      "loss": 1.2914,
+      "step": 81740
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.32633518707427e-05,
+      "loss": 1.0279,
+      "step": 81750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3260061206357563e-05,
+      "loss": 1.1089,
+      "step": 81760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3256770541972424e-05,
+      "loss": 1.2201,
+      "step": 81770
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3253479877587285e-05,
+      "loss": 1.2305,
+      "step": 81780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3250189213202146e-05,
+      "loss": 1.2803,
+      "step": 81790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3246898548817008e-05,
+      "loss": 1.3583,
+      "step": 81800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3243607884431866e-05,
+      "loss": 1.3381,
+      "step": 81810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3240317220046727e-05,
+      "loss": 1.4796,
+      "step": 81820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3237026555661588e-05,
+      "loss": 1.119,
+      "step": 81830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.323373589127645e-05,
+      "loss": 1.3884,
+      "step": 81840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.323044522689131e-05,
+      "loss": 1.2098,
+      "step": 81850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3227154562506172e-05,
+      "loss": 1.2352,
+      "step": 81860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3223863898121033e-05,
+      "loss": 1.3232,
+      "step": 81870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3220573233735894e-05,
+      "loss": 1.16,
+      "step": 81880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3217282569350752e-05,
+      "loss": 1.2816,
+      "step": 81890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3213991904965614e-05,
+      "loss": 1.258,
+      "step": 81900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3210701240580475e-05,
+      "loss": 1.1903,
+      "step": 81910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3207410576195336e-05,
+      "loss": 1.3161,
+      "step": 81920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3204119911810197e-05,
+      "loss": 1.1842,
+      "step": 81930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.320082924742506e-05,
+      "loss": 1.3249,
+      "step": 81940
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3197538583039917e-05,
+      "loss": 1.1117,
+      "step": 81950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3194247918654778e-05,
+      "loss": 1.2833,
+      "step": 81960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.319095725426964e-05,
+      "loss": 1.2491,
+      "step": 81970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.31876665898845e-05,
+      "loss": 1.2489,
+      "step": 81980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.318437592549936e-05,
+      "loss": 1.2263,
+      "step": 81990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.318108526111422e-05,
+      "loss": 1.223,
+      "step": 82000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.317779459672908e-05,
+      "loss": 1.3771,
+      "step": 82010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3174503932343942e-05,
+      "loss": 1.242,
+      "step": 82020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3171213267958803e-05,
+      "loss": 1.3314,
+      "step": 82030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3167922603573665e-05,
+      "loss": 1.4277,
+      "step": 82040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3164631939188522e-05,
+      "loss": 1.135,
+      "step": 82050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3161341274803384e-05,
+      "loss": 1.3571,
+      "step": 82060
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3158050610418245e-05,
+      "loss": 1.1807,
+      "step": 82070
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3154759946033106e-05,
+      "loss": 1.24,
+      "step": 82080
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3151469281647968e-05,
+      "loss": 1.2197,
+      "step": 82090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3148178617262825e-05,
+      "loss": 1.4559,
+      "step": 82100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3144887952877687e-05,
+      "loss": 1.113,
+      "step": 82110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3141597288492548e-05,
+      "loss": 1.2737,
+      "step": 82120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.313830662410741e-05,
+      "loss": 1.3352,
+      "step": 82130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.313501595972227e-05,
+      "loss": 1.0809,
+      "step": 82140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.313172529533713e-05,
+      "loss": 1.1228,
+      "step": 82150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.312843463095199e-05,
+      "loss": 1.3575,
+      "step": 82160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.312514396656685e-05,
+      "loss": 1.0363,
+      "step": 82170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3121853302181712e-05,
+      "loss": 1.1926,
+      "step": 82180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3118562637796573e-05,
+      "loss": 1.241,
+      "step": 82190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3115271973411435e-05,
+      "loss": 1.2153,
+      "step": 82200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3111981309026293e-05,
+      "loss": 1.2248,
+      "step": 82210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3108690644641154e-05,
+      "loss": 1.2716,
+      "step": 82220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3105399980256015e-05,
+      "loss": 1.153,
+      "step": 82230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3102109315870876e-05,
+      "loss": 1.0927,
+      "step": 82240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3098818651485738e-05,
+      "loss": 1.334,
+      "step": 82250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3095527987100596e-05,
+      "loss": 1.2863,
+      "step": 82260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3092237322715457e-05,
+      "loss": 1.242,
+      "step": 82270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3088946658330318e-05,
+      "loss": 1.2887,
+      "step": 82280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.308565599394518e-05,
+      "loss": 1.1757,
+      "step": 82290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.308236532956004e-05,
+      "loss": 1.152,
+      "step": 82300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.30790746651749e-05,
+      "loss": 1.2392,
+      "step": 82310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.307578400078976e-05,
+      "loss": 1.4789,
+      "step": 82320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.307249333640462e-05,
+      "loss": 1.0358,
+      "step": 82330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3069202672019482e-05,
+      "loss": 1.2712,
+      "step": 82340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3065912007634344e-05,
+      "loss": 1.2705,
+      "step": 82350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.30626213432492e-05,
+      "loss": 1.1536,
+      "step": 82360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3059330678864063e-05,
+      "loss": 1.1816,
+      "step": 82370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3056040014478924e-05,
+      "loss": 1.3369,
+      "step": 82380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3052749350093785e-05,
+      "loss": 1.2048,
+      "step": 82390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3049458685708646e-05,
+      "loss": 1.3687,
+      "step": 82400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3046168021323504e-05,
+      "loss": 1.0775,
+      "step": 82410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3042877356938366e-05,
+      "loss": 1.2946,
+      "step": 82420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3039586692553227e-05,
+      "loss": 1.2025,
+      "step": 82430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3036296028168088e-05,
+      "loss": 1.2934,
+      "step": 82440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.303300536378295e-05,
+      "loss": 1.2743,
+      "step": 82450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3029714699397807e-05,
+      "loss": 1.2567,
+      "step": 82460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.302642403501267e-05,
+      "loss": 1.4165,
+      "step": 82470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.302313337062753e-05,
+      "loss": 1.2959,
+      "step": 82480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.301984270624239e-05,
+      "loss": 1.1815,
+      "step": 82490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3016552041857252e-05,
+      "loss": 1.1705,
+      "step": 82500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.301326137747211e-05,
+      "loss": 1.2996,
+      "step": 82510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.300997071308697e-05,
+      "loss": 1.3118,
+      "step": 82520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3006680048701833e-05,
+      "loss": 1.2295,
+      "step": 82530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3003389384316694e-05,
+      "loss": 1.2455,
+      "step": 82540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3000098719931555e-05,
+      "loss": 1.2725,
+      "step": 82550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2996808055546417e-05,
+      "loss": 1.2458,
+      "step": 82560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2993517391161274e-05,
+      "loss": 1.2168,
+      "step": 82570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2990226726776136e-05,
+      "loss": 1.1114,
+      "step": 82580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2986936062390997e-05,
+      "loss": 1.3257,
+      "step": 82590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2983645398005858e-05,
+      "loss": 1.2247,
+      "step": 82600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.298035473362072e-05,
+      "loss": 1.2599,
+      "step": 82610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.297706406923558e-05,
+      "loss": 1.2001,
+      "step": 82620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2973773404850442e-05,
+      "loss": 1.1984,
+      "step": 82630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2970482740465303e-05,
+      "loss": 1.2042,
+      "step": 82640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.296719207608016e-05,
+      "loss": 1.3706,
+      "step": 82650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2963901411695022e-05,
+      "loss": 1.0424,
+      "step": 82660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2960610747309884e-05,
+      "loss": 1.2164,
+      "step": 82670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2957320082924745e-05,
+      "loss": 1.1413,
+      "step": 82680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2954029418539606e-05,
+      "loss": 1.2044,
+      "step": 82690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2950738754154468e-05,
+      "loss": 1.2392,
+      "step": 82700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2947448089769325e-05,
+      "loss": 1.2238,
+      "step": 82710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2944157425384187e-05,
+      "loss": 1.2233,
+      "step": 82720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2940866760999048e-05,
+      "loss": 1.1771,
+      "step": 82730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.293757609661391e-05,
+      "loss": 1.2289,
+      "step": 82740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.293428543222877e-05,
+      "loss": 1.2929,
+      "step": 82750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.293099476784363e-05,
+      "loss": 1.3344,
+      "step": 82760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.292770410345849e-05,
+      "loss": 1.2985,
+      "step": 82770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.292441343907335e-05,
+      "loss": 1.187,
+      "step": 82780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2921122774688212e-05,
+      "loss": 1.2312,
+      "step": 82790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2917832110303073e-05,
+      "loss": 1.2859,
+      "step": 82800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.291454144591793e-05,
+      "loss": 1.2806,
+      "step": 82810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2911250781532793e-05,
+      "loss": 1.2765,
+      "step": 82820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2907960117147654e-05,
+      "loss": 1.198,
+      "step": 82830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2904669452762515e-05,
+      "loss": 1.1719,
+      "step": 82840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2901378788377376e-05,
+      "loss": 1.3274,
+      "step": 82850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2898088123992234e-05,
+      "loss": 1.1999,
+      "step": 82860
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2894797459607096e-05,
+      "loss": 1.3179,
+      "step": 82870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2891506795221957e-05,
+      "loss": 1.352,
+      "step": 82880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2888216130836818e-05,
+      "loss": 1.3381,
+      "step": 82890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.288492546645168e-05,
+      "loss": 1.2726,
+      "step": 82900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2881634802066537e-05,
+      "loss": 1.2899,
+      "step": 82910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.28783441376814e-05,
+      "loss": 1.1142,
+      "step": 82920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.287505347329626e-05,
+      "loss": 1.3013,
+      "step": 82930
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.287176280891112e-05,
+      "loss": 1.258,
+      "step": 82940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2868472144525982e-05,
+      "loss": 1.2597,
+      "step": 82950
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.286518148014084e-05,
+      "loss": 1.1952,
+      "step": 82960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.28618908157557e-05,
+      "loss": 1.124,
+      "step": 82970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2858600151370563e-05,
+      "loss": 1.1928,
+      "step": 82980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2855309486985424e-05,
+      "loss": 1.0998,
+      "step": 82990
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2852018822600285e-05,
+      "loss": 1.2078,
+      "step": 83000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2848728158215143e-05,
+      "loss": 1.1761,
+      "step": 83010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2845437493830004e-05,
+      "loss": 1.4503,
+      "step": 83020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2842146829444866e-05,
+      "loss": 1.2452,
+      "step": 83030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2838856165059727e-05,
+      "loss": 1.1218,
+      "step": 83040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2835565500674588e-05,
+      "loss": 1.2435,
+      "step": 83050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2832274836289446e-05,
+      "loss": 1.2837,
+      "step": 83060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2828984171904307e-05,
+      "loss": 1.1317,
+      "step": 83070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.282569350751917e-05,
+      "loss": 1.3383,
+      "step": 83080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.282240284313403e-05,
+      "loss": 1.092,
+      "step": 83090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.281911217874889e-05,
+      "loss": 1.2788,
+      "step": 83100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.281582151436375e-05,
+      "loss": 1.2511,
+      "step": 83110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.281253084997861e-05,
+      "loss": 1.3095,
+      "step": 83120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.280924018559347e-05,
+      "loss": 1.2072,
+      "step": 83130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2805949521208333e-05,
+      "loss": 1.2606,
+      "step": 83140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2802658856823194e-05,
+      "loss": 1.1032,
+      "step": 83150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2799368192438052e-05,
+      "loss": 1.315,
+      "step": 83160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2796077528052913e-05,
+      "loss": 1.2767,
+      "step": 83170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2792786863667775e-05,
+      "loss": 1.2824,
+      "step": 83180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2789496199282636e-05,
+      "loss": 1.1486,
+      "step": 83190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2786205534897497e-05,
+      "loss": 1.3164,
+      "step": 83200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2782914870512355e-05,
+      "loss": 1.2025,
+      "step": 83210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2779624206127216e-05,
+      "loss": 1.1092,
+      "step": 83220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2776333541742077e-05,
+      "loss": 1.3424,
+      "step": 83230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.277304287735694e-05,
+      "loss": 1.233,
+      "step": 83240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.27697522129718e-05,
+      "loss": 1.2749,
+      "step": 83250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2766461548586658e-05,
+      "loss": 1.3806,
+      "step": 83260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.276317088420152e-05,
+      "loss": 1.274,
+      "step": 83270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.275988021981638e-05,
+      "loss": 1.2728,
+      "step": 83280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.275658955543124e-05,
+      "loss": 1.2722,
+      "step": 83290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2753298891046103e-05,
+      "loss": 1.2784,
+      "step": 83300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2750008226660964e-05,
+      "loss": 1.2689,
+      "step": 83310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2746717562275825e-05,
+      "loss": 1.1312,
+      "step": 83320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2743426897890683e-05,
+      "loss": 1.2515,
+      "step": 83330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2740136233505545e-05,
+      "loss": 1.2255,
+      "step": 83340
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2736845569120406e-05,
+      "loss": 1.1732,
+      "step": 83350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2733554904735267e-05,
+      "loss": 1.1529,
+      "step": 83360
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.273026424035013e-05,
+      "loss": 1.3872,
+      "step": 83370
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.272697357596499e-05,
+      "loss": 1.1084,
+      "step": 83380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.272368291157985e-05,
+      "loss": 1.4339,
+      "step": 83390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2720392247194712e-05,
+      "loss": 1.2214,
+      "step": 83400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.271710158280957e-05,
+      "loss": 1.1808,
+      "step": 83410
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.271381091842443e-05,
+      "loss": 1.2735,
+      "step": 83420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2710520254039293e-05,
+      "loss": 1.3188,
+      "step": 83430
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2707229589654154e-05,
+      "loss": 1.0922,
+      "step": 83440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2703938925269015e-05,
+      "loss": 1.2918,
+      "step": 83450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2700648260883873e-05,
+      "loss": 1.3116,
+      "step": 83460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2697357596498734e-05,
+      "loss": 1.2587,
+      "step": 83470
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2694066932113596e-05,
+      "loss": 1.4689,
+      "step": 83480
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2690776267728457e-05,
+      "loss": 1.221,
+      "step": 83490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2687485603343318e-05,
+      "loss": 1.0537,
+      "step": 83500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2684194938958176e-05,
+      "loss": 1.3053,
+      "step": 83510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2680904274573037e-05,
+      "loss": 1.2886,
+      "step": 83520
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.26776136101879e-05,
+      "loss": 1.3781,
+      "step": 83530
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.267432294580276e-05,
+      "loss": 1.3766,
+      "step": 83540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.267103228141762e-05,
+      "loss": 1.2991,
+      "step": 83550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.266774161703248e-05,
+      "loss": 1.1896,
+      "step": 83560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.266445095264734e-05,
+      "loss": 1.2699,
+      "step": 83570
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.26611602882622e-05,
+      "loss": 1.2248,
+      "step": 83580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2657869623877063e-05,
+      "loss": 1.1389,
+      "step": 83590
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2654578959491924e-05,
+      "loss": 1.2476,
+      "step": 83600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2651288295106785e-05,
+      "loss": 1.2681,
+      "step": 83610
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2647997630721643e-05,
+      "loss": 1.1748,
+      "step": 83620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2644706966336504e-05,
+      "loss": 1.0464,
+      "step": 83630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2641416301951366e-05,
+      "loss": 1.2425,
+      "step": 83640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2638125637566227e-05,
+      "loss": 1.3872,
+      "step": 83650
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2634834973181088e-05,
+      "loss": 1.1438,
+      "step": 83660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2631544308795946e-05,
+      "loss": 1.1524,
+      "step": 83670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2628253644410807e-05,
+      "loss": 1.0884,
+      "step": 83680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.262496298002567e-05,
+      "loss": 1.1298,
+      "step": 83690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.262167231564053e-05,
+      "loss": 1.2475,
+      "step": 83700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.261838165125539e-05,
+      "loss": 1.3275,
+      "step": 83710
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.261509098687025e-05,
+      "loss": 1.2091,
+      "step": 83720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.261180032248511e-05,
+      "loss": 1.1639,
+      "step": 83730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.260850965809997e-05,
+      "loss": 1.3248,
+      "step": 83740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2605218993714833e-05,
+      "loss": 1.3587,
+      "step": 83750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2601928329329694e-05,
+      "loss": 1.2569,
+      "step": 83760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2598637664944552e-05,
+      "loss": 1.2625,
+      "step": 83770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2595347000559413e-05,
+      "loss": 1.4226,
+      "step": 83780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2592056336174275e-05,
+      "loss": 1.4994,
+      "step": 83790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2588765671789136e-05,
+      "loss": 1.0936,
+      "step": 83800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2585475007403997e-05,
+      "loss": 1.2012,
+      "step": 83810
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2582184343018855e-05,
+      "loss": 1.4433,
+      "step": 83820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2578893678633716e-05,
+      "loss": 1.2791,
+      "step": 83830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2575603014248578e-05,
+      "loss": 1.2504,
+      "step": 83840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.257231234986344e-05,
+      "loss": 1.2491,
+      "step": 83850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.25690216854783e-05,
+      "loss": 1.3307,
+      "step": 83860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2565731021093158e-05,
+      "loss": 1.1886,
+      "step": 83870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.256244035670802e-05,
+      "loss": 1.2625,
+      "step": 83880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.255914969232288e-05,
+      "loss": 1.2295,
+      "step": 83890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2555859027937742e-05,
+      "loss": 1.3129,
+      "step": 83900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2552568363552603e-05,
+      "loss": 1.2306,
+      "step": 83910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.254927769916746e-05,
+      "loss": 1.2385,
+      "step": 83920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2545987034782322e-05,
+      "loss": 1.1403,
+      "step": 83930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2542696370397183e-05,
+      "loss": 1.3273,
+      "step": 83940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2539405706012045e-05,
+      "loss": 1.3111,
+      "step": 83950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2536115041626906e-05,
+      "loss": 1.2849,
+      "step": 83960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2532824377241764e-05,
+      "loss": 1.2243,
+      "step": 83970
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2529533712856625e-05,
+      "loss": 1.2915,
+      "step": 83980
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2526243048471486e-05,
+      "loss": 1.347,
+      "step": 83990
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2522952384086348e-05,
+      "loss": 1.3335,
+      "step": 84000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.251966171970121e-05,
+      "loss": 1.2736,
+      "step": 84010
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2516371055316067e-05,
+      "loss": 1.3952,
+      "step": 84020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2513080390930928e-05,
+      "loss": 1.2246,
+      "step": 84030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.250978972654579e-05,
+      "loss": 1.2355,
+      "step": 84040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.250649906216065e-05,
+      "loss": 1.4462,
+      "step": 84050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2503208397775512e-05,
+      "loss": 1.1164,
+      "step": 84060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2499917733390373e-05,
+      "loss": 1.1345,
+      "step": 84070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2496627069005234e-05,
+      "loss": 1.0832,
+      "step": 84080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2493336404620092e-05,
+      "loss": 1.3865,
+      "step": 84090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2490045740234954e-05,
+      "loss": 1.3016,
+      "step": 84100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2486755075849815e-05,
+      "loss": 1.2349,
+      "step": 84110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2483464411464676e-05,
+      "loss": 1.2755,
+      "step": 84120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2480173747079537e-05,
+      "loss": 1.1927,
+      "step": 84130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.24768830826944e-05,
+      "loss": 1.3246,
+      "step": 84140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.247359241830926e-05,
+      "loss": 1.1129,
+      "step": 84150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.247030175392412e-05,
+      "loss": 1.2202,
+      "step": 84160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.246701108953898e-05,
+      "loss": 1.2492,
+      "step": 84170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.246372042515384e-05,
+      "loss": 1.2795,
+      "step": 84180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.24604297607687e-05,
+      "loss": 1.1916,
+      "step": 84190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2457139096383563e-05,
+      "loss": 1.3552,
+      "step": 84200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2453848431998424e-05,
+      "loss": 1.3679,
+      "step": 84210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2450557767613282e-05,
+      "loss": 1.2121,
+      "step": 84220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2447267103228143e-05,
+      "loss": 1.2362,
+      "step": 84230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2443976438843005e-05,
+      "loss": 1.201,
+      "step": 84240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2440685774457866e-05,
+      "loss": 1.2217,
+      "step": 84250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2437395110072727e-05,
+      "loss": 1.244,
+      "step": 84260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2434104445687585e-05,
+      "loss": 1.1908,
+      "step": 84270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2430813781302446e-05,
+      "loss": 1.3816,
+      "step": 84280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2427523116917307e-05,
+      "loss": 1.1135,
+      "step": 84290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.242423245253217e-05,
+      "loss": 1.2979,
+      "step": 84300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.242094178814703e-05,
+      "loss": 1.3148,
+      "step": 84310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2417651123761888e-05,
+      "loss": 1.4387,
+      "step": 84320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.241436045937675e-05,
+      "loss": 1.2237,
+      "step": 84330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.241106979499161e-05,
+      "loss": 1.3111,
+      "step": 84340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.240777913060647e-05,
+      "loss": 1.2713,
+      "step": 84350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2404488466221333e-05,
+      "loss": 1.1711,
+      "step": 84360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.240119780183619e-05,
+      "loss": 1.3259,
+      "step": 84370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2397907137451052e-05,
+      "loss": 1.1825,
+      "step": 84380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2394616473065913e-05,
+      "loss": 1.2895,
+      "step": 84390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2391325808680775e-05,
+      "loss": 1.3129,
+      "step": 84400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2388035144295636e-05,
+      "loss": 1.3248,
+      "step": 84410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2384744479910494e-05,
+      "loss": 1.3104,
+      "step": 84420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2381453815525355e-05,
+      "loss": 1.1481,
+      "step": 84430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2378163151140216e-05,
+      "loss": 1.2449,
+      "step": 84440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2374872486755078e-05,
+      "loss": 1.0946,
+      "step": 84450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.237158182236994e-05,
+      "loss": 1.2617,
+      "step": 84460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2368291157984797e-05,
+      "loss": 1.365,
+      "step": 84470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2365000493599658e-05,
+      "loss": 1.3897,
+      "step": 84480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.236170982921452e-05,
+      "loss": 1.3122,
+      "step": 84490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.235841916482938e-05,
+      "loss": 1.2719,
+      "step": 84500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2355128500444242e-05,
+      "loss": 1.2951,
+      "step": 84510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.23518378360591e-05,
+      "loss": 1.2715,
+      "step": 84520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.234854717167396e-05,
+      "loss": 1.1927,
+      "step": 84530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2345256507288822e-05,
+      "loss": 1.1019,
+      "step": 84540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2341965842903683e-05,
+      "loss": 1.2151,
+      "step": 84550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2338675178518545e-05,
+      "loss": 1.691,
+      "step": 84560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2335384514133403e-05,
+      "loss": 1.3373,
+      "step": 84570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2332093849748264e-05,
+      "loss": 1.3423,
+      "step": 84580
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2328803185363125e-05,
+      "loss": 1.2093,
+      "step": 84590
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2325512520977986e-05,
+      "loss": 1.3369,
+      "step": 84600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2322221856592848e-05,
+      "loss": 1.2091,
+      "step": 84610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2318931192207706e-05,
+      "loss": 1.2196,
+      "step": 84620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2315640527822567e-05,
+      "loss": 1.2398,
+      "step": 84630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2312349863437428e-05,
+      "loss": 1.2243,
+      "step": 84640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.230905919905229e-05,
+      "loss": 1.3896,
+      "step": 84650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.230576853466715e-05,
+      "loss": 1.2628,
+      "step": 84660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.230247787028201e-05,
+      "loss": 1.3473,
+      "step": 84670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.229918720589687e-05,
+      "loss": 1.3283,
+      "step": 84680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.229589654151173e-05,
+      "loss": 1.3162,
+      "step": 84690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2292605877126592e-05,
+      "loss": 1.3192,
+      "step": 84700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2289315212741454e-05,
+      "loss": 1.1796,
+      "step": 84710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.228602454835631e-05,
+      "loss": 1.287,
+      "step": 84720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2282733883971173e-05,
+      "loss": 1.1538,
+      "step": 84730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2279443219586034e-05,
+      "loss": 1.0795,
+      "step": 84740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2276152555200895e-05,
+      "loss": 1.2462,
+      "step": 84750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2272861890815757e-05,
+      "loss": 1.1855,
+      "step": 84760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2269571226430618e-05,
+      "loss": 1.4488,
+      "step": 84770
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2266280562045476e-05,
+      "loss": 1.1928,
+      "step": 84780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2262989897660337e-05,
+      "loss": 1.1108,
+      "step": 84790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2259699233275198e-05,
+      "loss": 1.2886,
+      "step": 84800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.225640856889006e-05,
+      "loss": 1.2694,
+      "step": 84810
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.225311790450492e-05,
+      "loss": 1.3961,
+      "step": 84820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2249827240119782e-05,
+      "loss": 1.2792,
+      "step": 84830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2246536575734643e-05,
+      "loss": 1.302,
+      "step": 84840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2243245911349505e-05,
+      "loss": 1.2606,
+      "step": 84850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2239955246964362e-05,
+      "loss": 1.2719,
+      "step": 84860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2236664582579224e-05,
+      "loss": 1.2738,
+      "step": 84870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2233373918194085e-05,
+      "loss": 1.1817,
+      "step": 84880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2230083253808946e-05,
+      "loss": 1.3694,
+      "step": 84890
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2226792589423807e-05,
+      "loss": 1.2458,
+      "step": 84900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.222350192503867e-05,
+      "loss": 1.1541,
+      "step": 84910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2220211260653527e-05,
+      "loss": 1.3217,
+      "step": 84920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2216920596268388e-05,
+      "loss": 1.2471,
+      "step": 84930
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.221362993188325e-05,
+      "loss": 1.3652,
+      "step": 84940
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.221033926749811e-05,
+      "loss": 1.3047,
+      "step": 84950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2207048603112972e-05,
+      "loss": 1.3038,
+      "step": 84960
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.220375793872783e-05,
+      "loss": 1.1579,
+      "step": 84970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.220046727434269e-05,
+      "loss": 1.3359,
+      "step": 84980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2197176609957552e-05,
+      "loss": 1.1597,
+      "step": 84990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2193885945572413e-05,
+      "loss": 1.1606,
+      "step": 85000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2190595281187275e-05,
+      "loss": 1.3335,
+      "step": 85010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2187304616802136e-05,
+      "loss": 1.2024,
+      "step": 85020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2184013952416994e-05,
+      "loss": 1.2437,
+      "step": 85030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2180723288031855e-05,
+      "loss": 1.2549,
+      "step": 85040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2177432623646716e-05,
+      "loss": 1.2033,
+      "step": 85050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2174141959261578e-05,
+      "loss": 1.3492,
+      "step": 85060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.217085129487644e-05,
+      "loss": 1.1395,
+      "step": 85070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2167560630491297e-05,
+      "loss": 1.2585,
+      "step": 85080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2164269966106158e-05,
+      "loss": 1.2414,
+      "step": 85090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.216097930172102e-05,
+      "loss": 1.217,
+      "step": 85100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.215768863733588e-05,
+      "loss": 1.2921,
+      "step": 85110
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2154397972950742e-05,
+      "loss": 1.2325,
+      "step": 85120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.21511073085656e-05,
+      "loss": 1.1533,
+      "step": 85130
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.214781664418046e-05,
+      "loss": 1.2669,
+      "step": 85140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2144525979795322e-05,
+      "loss": 1.1183,
+      "step": 85150
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2141235315410184e-05,
+      "loss": 1.2332,
+      "step": 85160
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2137944651025045e-05,
+      "loss": 1.3432,
+      "step": 85170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2134653986639903e-05,
+      "loss": 1.2467,
+      "step": 85180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2131363322254764e-05,
+      "loss": 1.3832,
+      "step": 85190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2128072657869625e-05,
+      "loss": 1.2137,
+      "step": 85200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2124781993484486e-05,
+      "loss": 1.3695,
+      "step": 85210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2121491329099348e-05,
+      "loss": 1.2458,
+      "step": 85220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2118200664714206e-05,
+      "loss": 1.2963,
+      "step": 85230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2114910000329067e-05,
+      "loss": 1.28,
+      "step": 85240
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2111619335943928e-05,
+      "loss": 1.2947,
+      "step": 85250
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.210832867155879e-05,
+      "loss": 1.1418,
+      "step": 85260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.210503800717365e-05,
+      "loss": 1.214,
+      "step": 85270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.210174734278851e-05,
+      "loss": 1.2717,
+      "step": 85280
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.209845667840337e-05,
+      "loss": 1.0818,
+      "step": 85290
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.209516601401823e-05,
+      "loss": 1.3735,
+      "step": 85300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2091875349633092e-05,
+      "loss": 1.1961,
+      "step": 85310
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2088584685247954e-05,
+      "loss": 1.4132,
+      "step": 85320
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.208529402086281e-05,
+      "loss": 1.1729,
+      "step": 85330
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2082003356477673e-05,
+      "loss": 1.4834,
+      "step": 85340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2078712692092534e-05,
+      "loss": 1.1356,
+      "step": 85350
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2075422027707395e-05,
+      "loss": 1.2778,
+      "step": 85360
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2072131363322257e-05,
+      "loss": 1.4594,
+      "step": 85370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2068840698937114e-05,
+      "loss": 1.3912,
+      "step": 85380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2065550034551976e-05,
+      "loss": 1.1873,
+      "step": 85390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2062259370166837e-05,
+      "loss": 1.3385,
+      "step": 85400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2058968705781698e-05,
+      "loss": 1.2321,
+      "step": 85410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.205567804139656e-05,
+      "loss": 1.2599,
+      "step": 85420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2052387377011417e-05,
+      "loss": 1.1444,
+      "step": 85430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.204909671262628e-05,
+      "loss": 1.3572,
+      "step": 85440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.204580604824114e-05,
+      "loss": 1.281,
+      "step": 85450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2042515383856e-05,
+      "loss": 1.4211,
+      "step": 85460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2039224719470862e-05,
+      "loss": 1.2778,
+      "step": 85470
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.203593405508572e-05,
+      "loss": 1.3065,
+      "step": 85480
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.203264339070058e-05,
+      "loss": 1.2094,
+      "step": 85490
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2029352726315443e-05,
+      "loss": 1.0634,
+      "step": 85500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2026062061930304e-05,
+      "loss": 1.1867,
+      "step": 85510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2022771397545165e-05,
+      "loss": 1.1222,
+      "step": 85520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2019480733160027e-05,
+      "loss": 1.2947,
+      "step": 85530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2016190068774885e-05,
+      "loss": 1.179,
+      "step": 85540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2012899404389746e-05,
+      "loss": 1.0823,
+      "step": 85550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2009608740004607e-05,
+      "loss": 1.2358,
+      "step": 85560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.200631807561947e-05,
+      "loss": 1.1659,
+      "step": 85570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.200302741123433e-05,
+      "loss": 1.3247,
+      "step": 85580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.199973674684919e-05,
+      "loss": 1.1785,
+      "step": 85590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1996446082464052e-05,
+      "loss": 1.3646,
+      "step": 85600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1993155418078913e-05,
+      "loss": 1.3226,
+      "step": 85610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.198986475369377e-05,
+      "loss": 1.2062,
+      "step": 85620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1986574089308633e-05,
+      "loss": 1.2429,
+      "step": 85630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1983283424923494e-05,
+      "loss": 1.2786,
+      "step": 85640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1979992760538355e-05,
+      "loss": 1.1844,
+      "step": 85650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1976702096153216e-05,
+      "loss": 1.3124,
+      "step": 85660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1973411431768078e-05,
+      "loss": 1.3313,
+      "step": 85670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1970120767382936e-05,
+      "loss": 1.3664,
+      "step": 85680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1966830102997797e-05,
+      "loss": 1.1722,
+      "step": 85690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1963539438612658e-05,
+      "loss": 1.1092,
+      "step": 85700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.196024877422752e-05,
+      "loss": 1.3009,
+      "step": 85710
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.195695810984238e-05,
+      "loss": 1.2391,
+      "step": 85720
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.195366744545724e-05,
+      "loss": 1.2984,
+      "step": 85730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.19503767810721e-05,
+      "loss": 1.2203,
+      "step": 85740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.194708611668696e-05,
+      "loss": 1.4094,
+      "step": 85750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1943795452301822e-05,
+      "loss": 1.2856,
+      "step": 85760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1940504787916684e-05,
+      "loss": 1.3508,
+      "step": 85770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.193721412353154e-05,
+      "loss": 1.1755,
+      "step": 85780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1933923459146403e-05,
+      "loss": 1.3189,
+      "step": 85790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1930632794761264e-05,
+      "loss": 1.2271,
+      "step": 85800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1927342130376125e-05,
+      "loss": 1.1897,
+      "step": 85810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1924051465990987e-05,
+      "loss": 1.1949,
+      "step": 85820
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1920760801605844e-05,
+      "loss": 1.1948,
+      "step": 85830
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1917470137220706e-05,
+      "loss": 1.1497,
+      "step": 85840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1914179472835567e-05,
+      "loss": 1.2187,
+      "step": 85850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1910888808450428e-05,
+      "loss": 1.2589,
+      "step": 85860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.190759814406529e-05,
+      "loss": 1.2251,
+      "step": 85870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1904307479680147e-05,
+      "loss": 1.3317,
+      "step": 85880
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.190101681529501e-05,
+      "loss": 1.198,
+      "step": 85890
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.189772615090987e-05,
+      "loss": 1.1572,
+      "step": 85900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.189443548652473e-05,
+      "loss": 1.4023,
+      "step": 85910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1891144822139592e-05,
+      "loss": 1.6361,
+      "step": 85920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.188785415775445e-05,
+      "loss": 1.2462,
+      "step": 85930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.188456349336931e-05,
+      "loss": 1.1339,
+      "step": 85940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1881272828984173e-05,
+      "loss": 1.0443,
+      "step": 85950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1877982164599034e-05,
+      "loss": 1.3079,
+      "step": 85960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1874691500213895e-05,
+      "loss": 1.2597,
+      "step": 85970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1871400835828753e-05,
+      "loss": 1.1005,
+      "step": 85980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1868110171443614e-05,
+      "loss": 1.2977,
+      "step": 85990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1864819507058476e-05,
+      "loss": 1.2361,
+      "step": 86000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1861528842673337e-05,
+      "loss": 1.3914,
+      "step": 86010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1858238178288198e-05,
+      "loss": 1.2902,
+      "step": 86020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1854947513903056e-05,
+      "loss": 1.2435,
+      "step": 86030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1851656849517917e-05,
+      "loss": 1.2731,
+      "step": 86040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.184836618513278e-05,
+      "loss": 1.371,
+      "step": 86050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.184507552074764e-05,
+      "loss": 1.3101,
+      "step": 86060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.18417848563625e-05,
+      "loss": 1.1883,
+      "step": 86070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.183849419197736e-05,
+      "loss": 1.4288,
+      "step": 86080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.183520352759222e-05,
+      "loss": 1.1432,
+      "step": 86090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.183191286320708e-05,
+      "loss": 1.172,
+      "step": 86100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1828622198821943e-05,
+      "loss": 1.1487,
+      "step": 86110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1825331534436804e-05,
+      "loss": 1.2675,
+      "step": 86120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1822040870051662e-05,
+      "loss": 1.3169,
+      "step": 86130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1818750205666523e-05,
+      "loss": 1.1806,
+      "step": 86140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1815459541281385e-05,
+      "loss": 1.2272,
+      "step": 86150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1812168876896246e-05,
+      "loss": 1.2122,
+      "step": 86160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1808878212511107e-05,
+      "loss": 1.243,
+      "step": 86170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1805587548125965e-05,
+      "loss": 1.2322,
+      "step": 86180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1802296883740826e-05,
+      "loss": 1.213,
+      "step": 86190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1799006219355688e-05,
+      "loss": 1.3136,
+      "step": 86200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.179571555497055e-05,
+      "loss": 1.2341,
+      "step": 86210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.179242489058541e-05,
+      "loss": 1.1769,
+      "step": 86220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1789134226200268e-05,
+      "loss": 1.2346,
+      "step": 86230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.178584356181513e-05,
+      "loss": 1.2721,
+      "step": 86240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.178255289742999e-05,
+      "loss": 1.3542,
+      "step": 86250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1779262233044852e-05,
+      "loss": 1.0848,
+      "step": 86260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1775971568659713e-05,
+      "loss": 1.2989,
+      "step": 86270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1772680904274574e-05,
+      "loss": 1.2751,
+      "step": 86280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1769390239889436e-05,
+      "loss": 1.3905,
+      "step": 86290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1766099575504293e-05,
+      "loss": 1.3401,
+      "step": 86300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1762808911119155e-05,
+      "loss": 1.0445,
+      "step": 86310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1759518246734016e-05,
+      "loss": 1.208,
+      "step": 86320
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1756227582348877e-05,
+      "loss": 1.1627,
+      "step": 86330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.175293691796374e-05,
+      "loss": 1.1974,
+      "step": 86340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.17496462535786e-05,
+      "loss": 1.159,
+      "step": 86350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.174635558919346e-05,
+      "loss": 1.3091,
+      "step": 86360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1743064924808322e-05,
+      "loss": 1.3843,
+      "step": 86370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.173977426042318e-05,
+      "loss": 1.177,
+      "step": 86380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.173648359603804e-05,
+      "loss": 1.2954,
+      "step": 86390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1733192931652903e-05,
+      "loss": 1.36,
+      "step": 86400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1729902267267764e-05,
+      "loss": 1.1604,
+      "step": 86410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1726611602882625e-05,
+      "loss": 1.0324,
+      "step": 86420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1723320938497487e-05,
+      "loss": 1.2278,
+      "step": 86430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1720030274112344e-05,
+      "loss": 1.2401,
+      "step": 86440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1716739609727206e-05,
+      "loss": 1.1915,
+      "step": 86450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1713448945342067e-05,
+      "loss": 1.1614,
+      "step": 86460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1710158280956928e-05,
+      "loss": 1.2704,
+      "step": 86470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.170686761657179e-05,
+      "loss": 1.4138,
+      "step": 86480
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1703576952186647e-05,
+      "loss": 1.2599,
+      "step": 86490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.170028628780151e-05,
+      "loss": 1.1352,
+      "step": 86500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.169699562341637e-05,
+      "loss": 1.2331,
+      "step": 86510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.169370495903123e-05,
+      "loss": 1.178,
+      "step": 86520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1690414294646092e-05,
+      "loss": 1.2886,
+      "step": 86530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.168712363026095e-05,
+      "loss": 1.3445,
+      "step": 86540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.168383296587581e-05,
+      "loss": 1.2284,
+      "step": 86550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1680542301490673e-05,
+      "loss": 1.419,
+      "step": 86560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1677251637105534e-05,
+      "loss": 1.1709,
+      "step": 86570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1673960972720395e-05,
+      "loss": 1.2071,
+      "step": 86580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1670670308335253e-05,
+      "loss": 1.2897,
+      "step": 86590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1667379643950115e-05,
+      "loss": 1.224,
+      "step": 86600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1664088979564976e-05,
+      "loss": 1.3189,
+      "step": 86610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1660798315179837e-05,
+      "loss": 1.1648,
+      "step": 86620
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.16575076507947e-05,
+      "loss": 1.3376,
+      "step": 86630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1654216986409556e-05,
+      "loss": 1.2133,
+      "step": 86640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1650926322024417e-05,
+      "loss": 1.3883,
+      "step": 86650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.164763565763928e-05,
+      "loss": 1.1983,
+      "step": 86660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.164434499325414e-05,
+      "loss": 1.1519,
+      "step": 86670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1641054328869e-05,
+      "loss": 1.3056,
+      "step": 86680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.163776366448386e-05,
+      "loss": 1.19,
+      "step": 86690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.163447300009872e-05,
+      "loss": 1.2527,
+      "step": 86700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1631182335713582e-05,
+      "loss": 1.0995,
+      "step": 86710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1627891671328443e-05,
+      "loss": 1.1268,
+      "step": 86720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1624601006943304e-05,
+      "loss": 1.2723,
+      "step": 86730
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1621310342558162e-05,
+      "loss": 1.2227,
+      "step": 86740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1618019678173023e-05,
+      "loss": 1.3731,
+      "step": 86750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1614729013787885e-05,
+      "loss": 1.2752,
+      "step": 86760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1611438349402746e-05,
+      "loss": 1.3193,
+      "step": 86770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1608147685017607e-05,
+      "loss": 1.3501,
+      "step": 86780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1604857020632465e-05,
+      "loss": 1.4031,
+      "step": 86790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1601566356247326e-05,
+      "loss": 1.2706,
+      "step": 86800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1598275691862188e-05,
+      "loss": 1.1622,
+      "step": 86810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.159498502747705e-05,
+      "loss": 1.3256,
+      "step": 86820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.159169436309191e-05,
+      "loss": 1.2468,
+      "step": 86830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1588403698706768e-05,
+      "loss": 1.2358,
+      "step": 86840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.158511303432163e-05,
+      "loss": 1.1905,
+      "step": 86850
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.158182236993649e-05,
+      "loss": 1.1242,
+      "step": 86860
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1578531705551352e-05,
+      "loss": 1.1983,
+      "step": 86870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1575241041166213e-05,
+      "loss": 1.244,
+      "step": 86880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.157195037678107e-05,
+      "loss": 1.2922,
+      "step": 86890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1568659712395932e-05,
+      "loss": 1.2269,
+      "step": 86900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1565369048010794e-05,
+      "loss": 1.2652,
+      "step": 86910
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1562078383625655e-05,
+      "loss": 1.2775,
+      "step": 86920
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1558787719240516e-05,
+      "loss": 1.122,
+      "step": 86930
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1555497054855374e-05,
+      "loss": 1.4138,
+      "step": 86940
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1552206390470235e-05,
+      "loss": 1.3403,
+      "step": 86950
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1548915726085096e-05,
+      "loss": 1.1967,
+      "step": 86960
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1545625061699958e-05,
+      "loss": 1.3351,
+      "step": 86970
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.154233439731482e-05,
+      "loss": 1.0631,
+      "step": 86980
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1539043732929677e-05,
+      "loss": 1.2856,
+      "step": 86990
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1535753068544538e-05,
+      "loss": 1.2008,
+      "step": 87000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.15324624041594e-05,
+      "loss": 1.2267,
+      "step": 87010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.152917173977426e-05,
+      "loss": 1.2659,
+      "step": 87020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1525881075389122e-05,
+      "loss": 1.3196,
+      "step": 87030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1522590411003983e-05,
+      "loss": 1.4157,
+      "step": 87040
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1519299746618844e-05,
+      "loss": 1.2928,
+      "step": 87050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1516009082233702e-05,
+      "loss": 1.2439,
+      "step": 87060
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1512718417848564e-05,
+      "loss": 1.2877,
+      "step": 87070
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1509427753463425e-05,
+      "loss": 1.3298,
+      "step": 87080
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1506137089078286e-05,
+      "loss": 1.2238,
+      "step": 87090
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1502846424693147e-05,
+      "loss": 1.2274,
+      "step": 87100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.149955576030801e-05,
+      "loss": 1.2626,
+      "step": 87110
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.149626509592287e-05,
+      "loss": 1.2234,
+      "step": 87120
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.149297443153773e-05,
+      "loss": 1.4888,
+      "step": 87130
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.148968376715259e-05,
+      "loss": 1.2666,
+      "step": 87140
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.148639310276745e-05,
+      "loss": 1.1016,
+      "step": 87150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.148310243838231e-05,
+      "loss": 1.2779,
+      "step": 87160
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1479811773997173e-05,
+      "loss": 1.3502,
+      "step": 87170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1476521109612034e-05,
+      "loss": 1.2647,
+      "step": 87180
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1473230445226892e-05,
+      "loss": 1.2593,
+      "step": 87190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1469939780841753e-05,
+      "loss": 1.2739,
+      "step": 87200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1466649116456615e-05,
+      "loss": 1.2219,
+      "step": 87210
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1463358452071476e-05,
+      "loss": 1.2198,
+      "step": 87220
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1460067787686337e-05,
+      "loss": 1.4116,
+      "step": 87230
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1456777123301195e-05,
+      "loss": 1.2591,
+      "step": 87240
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1453486458916056e-05,
+      "loss": 1.1253,
+      "step": 87250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1450195794530918e-05,
+      "loss": 1.251,
+      "step": 87260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.144690513014578e-05,
+      "loss": 1.2298,
+      "step": 87270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.144361446576064e-05,
+      "loss": 1.1257,
+      "step": 87280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1440323801375498e-05,
+      "loss": 1.2793,
+      "step": 87290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.143703313699036e-05,
+      "loss": 1.3272,
+      "step": 87300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.143374247260522e-05,
+      "loss": 1.2818,
+      "step": 87310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1430451808220082e-05,
+      "loss": 1.1261,
+      "step": 87320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1427161143834943e-05,
+      "loss": 1.2808,
+      "step": 87330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.14238704794498e-05,
+      "loss": 1.2326,
+      "step": 87340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1420579815064662e-05,
+      "loss": 1.2858,
+      "step": 87350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1417289150679523e-05,
+      "loss": 1.2729,
+      "step": 87360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1413998486294385e-05,
+      "loss": 1.3115,
+      "step": 87370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1410707821909246e-05,
+      "loss": 1.4775,
+      "step": 87380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1407417157524104e-05,
+      "loss": 1.262,
+      "step": 87390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1404126493138965e-05,
+      "loss": 1.1984,
+      "step": 87400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1400835828753826e-05,
+      "loss": 1.2555,
+      "step": 87410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1397545164368688e-05,
+      "loss": 1.2391,
+      "step": 87420
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.139425449998355e-05,
+      "loss": 1.2357,
+      "step": 87430
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1390963835598407e-05,
+      "loss": 1.1228,
+      "step": 87440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1387673171213268e-05,
+      "loss": 1.2203,
+      "step": 87450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.138438250682813e-05,
+      "loss": 1.0409,
+      "step": 87460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.138109184244299e-05,
+      "loss": 1.3468,
+      "step": 87470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1377801178057852e-05,
+      "loss": 1.2493,
+      "step": 87480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.137451051367271e-05,
+      "loss": 1.1366,
+      "step": 87490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.137121984928757e-05,
+      "loss": 1.2795,
+      "step": 87500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1367929184902432e-05,
+      "loss": 1.2259,
+      "step": 87510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1364638520517294e-05,
+      "loss": 1.1676,
+      "step": 87520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1361347856132155e-05,
+      "loss": 1.315,
+      "step": 87530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1358057191747013e-05,
+      "loss": 1.2776,
+      "step": 87540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1354766527361874e-05,
+      "loss": 1.2306,
+      "step": 87550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1351475862976735e-05,
+      "loss": 1.1166,
+      "step": 87560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1348185198591597e-05,
+      "loss": 1.3441,
+      "step": 87570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1344894534206458e-05,
+      "loss": 1.1564,
+      "step": 87580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1341603869821316e-05,
+      "loss": 1.199,
+      "step": 87590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1338313205436177e-05,
+      "loss": 1.0956,
+      "step": 87600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1335022541051038e-05,
+      "loss": 1.3935,
+      "step": 87610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.13317318766659e-05,
+      "loss": 1.3404,
+      "step": 87620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.132844121228076e-05,
+      "loss": 1.2437,
+      "step": 87630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.132515054789562e-05,
+      "loss": 1.2059,
+      "step": 87640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.132185988351048e-05,
+      "loss": 1.4406,
+      "step": 87650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.131856921912534e-05,
+      "loss": 1.2645,
+      "step": 87660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1315278554740202e-05,
+      "loss": 1.2376,
+      "step": 87670
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1311987890355064e-05,
+      "loss": 1.1551,
+      "step": 87680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.130869722596992e-05,
+      "loss": 1.2714,
+      "step": 87690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1305406561584783e-05,
+      "loss": 1.1914,
+      "step": 87700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1302115897199644e-05,
+      "loss": 1.2802,
+      "step": 87710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1298825232814505e-05,
+      "loss": 1.1379,
+      "step": 87720
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1295534568429367e-05,
+      "loss": 1.312,
+      "step": 87730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1292243904044228e-05,
+      "loss": 1.2021,
+      "step": 87740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1288953239659086e-05,
+      "loss": 1.1363,
+      "step": 87750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1285662575273947e-05,
+      "loss": 1.2758,
+      "step": 87760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1282371910888808e-05,
+      "loss": 1.2436,
+      "step": 87770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.127908124650367e-05,
+      "loss": 1.2096,
+      "step": 87780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.127579058211853e-05,
+      "loss": 1.1812,
+      "step": 87790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1272499917733392e-05,
+      "loss": 1.3211,
+      "step": 87800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1269209253348253e-05,
+      "loss": 1.3291,
+      "step": 87810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1265918588963115e-05,
+      "loss": 1.2291,
+      "step": 87820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1262627924577973e-05,
+      "loss": 1.2365,
+      "step": 87830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1259337260192834e-05,
+      "loss": 1.2076,
+      "step": 87840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1256046595807695e-05,
+      "loss": 1.1294,
+      "step": 87850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1252755931422556e-05,
+      "loss": 1.1903,
+      "step": 87860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1249465267037418e-05,
+      "loss": 1.2605,
+      "step": 87870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.124617460265228e-05,
+      "loss": 1.4082,
+      "step": 87880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.124288393826714e-05,
+      "loss": 1.4056,
+      "step": 87890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1239593273881998e-05,
+      "loss": 1.4094,
+      "step": 87900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.123630260949686e-05,
+      "loss": 1.0604,
+      "step": 87910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.123301194511172e-05,
+      "loss": 1.2276,
+      "step": 87920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1229721280726582e-05,
+      "loss": 1.3108,
+      "step": 87930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1226430616341443e-05,
+      "loss": 1.275,
+      "step": 87940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.12231399519563e-05,
+      "loss": 1.465,
+      "step": 87950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1219849287571162e-05,
+      "loss": 1.6637,
+      "step": 87960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1216558623186023e-05,
+      "loss": 1.2099,
+      "step": 87970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1213267958800885e-05,
+      "loss": 1.3846,
+      "step": 87980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1209977294415746e-05,
+      "loss": 1.3287,
+      "step": 87990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1206686630030604e-05,
+      "loss": 1.333,
+      "step": 88000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1203395965645465e-05,
+      "loss": 1.2378,
+      "step": 88010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1200105301260326e-05,
+      "loss": 1.2337,
+      "step": 88020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1196814636875188e-05,
+      "loss": 1.2734,
+      "step": 88030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.119352397249005e-05,
+      "loss": 1.169,
+      "step": 88040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1190233308104907e-05,
+      "loss": 1.1426,
+      "step": 88050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1186942643719768e-05,
+      "loss": 1.1693,
+      "step": 88060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.118365197933463e-05,
+      "loss": 1.2864,
+      "step": 88070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.118036131494949e-05,
+      "loss": 1.272,
+      "step": 88080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1177070650564352e-05,
+      "loss": 1.3067,
+      "step": 88090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.117377998617921e-05,
+      "loss": 1.2156,
+      "step": 88100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.117048932179407e-05,
+      "loss": 1.2353,
+      "step": 88110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1167198657408932e-05,
+      "loss": 1.3197,
+      "step": 88120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1163907993023794e-05,
+      "loss": 1.2085,
+      "step": 88130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1160617328638655e-05,
+      "loss": 1.3076,
+      "step": 88140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1157326664253513e-05,
+      "loss": 1.3205,
+      "step": 88150
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1154035999868374e-05,
+      "loss": 1.2131,
+      "step": 88160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1150745335483235e-05,
+      "loss": 1.2802,
+      "step": 88170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1147454671098097e-05,
+      "loss": 1.2458,
+      "step": 88180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1144164006712958e-05,
+      "loss": 1.1347,
+      "step": 88190
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1140873342327816e-05,
+      "loss": 1.3626,
+      "step": 88200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1137582677942677e-05,
+      "loss": 1.4773,
+      "step": 88210
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1134292013557538e-05,
+      "loss": 1.0883,
+      "step": 88220
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.11310013491724e-05,
+      "loss": 1.4021,
+      "step": 88230
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.112771068478726e-05,
+      "loss": 1.1431,
+      "step": 88240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.112442002040212e-05,
+      "loss": 1.2407,
+      "step": 88250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.112112935601698e-05,
+      "loss": 1.1265,
+      "step": 88260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.111783869163184e-05,
+      "loss": 1.3135,
+      "step": 88270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1114548027246702e-05,
+      "loss": 1.1304,
+      "step": 88280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1111257362861564e-05,
+      "loss": 1.2418,
+      "step": 88290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.110796669847642e-05,
+      "loss": 1.3083,
+      "step": 88300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1104676034091283e-05,
+      "loss": 1.3125,
+      "step": 88310
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1101385369706144e-05,
+      "loss": 1.2304,
+      "step": 88320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1098094705321005e-05,
+      "loss": 1.1966,
+      "step": 88330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1094804040935867e-05,
+      "loss": 1.1927,
+      "step": 88340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1091513376550725e-05,
+      "loss": 1.2652,
+      "step": 88350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1088222712165586e-05,
+      "loss": 1.1886,
+      "step": 88360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1084932047780447e-05,
+      "loss": 1.1374,
+      "step": 88370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.108164138339531e-05,
+      "loss": 1.1843,
+      "step": 88380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.107835071901017e-05,
+      "loss": 1.3405,
+      "step": 88390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1075060054625027e-05,
+      "loss": 1.2966,
+      "step": 88400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.107176939023989e-05,
+      "loss": 1.1931,
+      "step": 88410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.106847872585475e-05,
+      "loss": 1.2345,
+      "step": 88420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.106518806146961e-05,
+      "loss": 1.2951,
+      "step": 88430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1061897397084473e-05,
+      "loss": 1.2701,
+      "step": 88440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.105860673269933e-05,
+      "loss": 1.1062,
+      "step": 88450
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1055316068314192e-05,
+      "loss": 1.346,
+      "step": 88460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1052025403929053e-05,
+      "loss": 1.2164,
+      "step": 88470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1048734739543914e-05,
+      "loss": 1.2541,
+      "step": 88480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1045444075158776e-05,
+      "loss": 1.168,
+      "step": 88490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1042153410773637e-05,
+      "loss": 1.1534,
+      "step": 88500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1038862746388495e-05,
+      "loss": 1.1154,
+      "step": 88510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1035572082003356e-05,
+      "loss": 1.327,
+      "step": 88520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1032281417618217e-05,
+      "loss": 1.183,
+      "step": 88530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.102899075323308e-05,
+      "loss": 1.1622,
+      "step": 88540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.102570008884794e-05,
+      "loss": 1.2385,
+      "step": 88550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.10224094244628e-05,
+      "loss": 1.1816,
+      "step": 88560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1019118760077662e-05,
+      "loss": 1.2028,
+      "step": 88570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1015828095692524e-05,
+      "loss": 1.3215,
+      "step": 88580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.101253743130738e-05,
+      "loss": 1.2731,
+      "step": 88590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1009246766922243e-05,
+      "loss": 1.2811,
+      "step": 88600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1005956102537104e-05,
+      "loss": 1.3192,
+      "step": 88610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1002665438151965e-05,
+      "loss": 1.3218,
+      "step": 88620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0999374773766826e-05,
+      "loss": 1.1775,
+      "step": 88630
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0996084109381688e-05,
+      "loss": 1.3499,
+      "step": 88640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0992793444996546e-05,
+      "loss": 1.2992,
+      "step": 88650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0989502780611407e-05,
+      "loss": 1.3455,
+      "step": 88660
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0986212116226268e-05,
+      "loss": 1.0693,
+      "step": 88670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.098292145184113e-05,
+      "loss": 1.249,
+      "step": 88680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.097963078745599e-05,
+      "loss": 1.3415,
+      "step": 88690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.097634012307085e-05,
+      "loss": 1.2443,
+      "step": 88700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.097304945868571e-05,
+      "loss": 1.3516,
+      "step": 88710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.096975879430057e-05,
+      "loss": 1.1264,
+      "step": 88720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0966468129915432e-05,
+      "loss": 1.1971,
+      "step": 88730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0963177465530294e-05,
+      "loss": 1.0867,
+      "step": 88740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.095988680114515e-05,
+      "loss": 1.2427,
+      "step": 88750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0956596136760013e-05,
+      "loss": 1.2847,
+      "step": 88760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0953305472374874e-05,
+      "loss": 1.3145,
+      "step": 88770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0950014807989735e-05,
+      "loss": 1.2107,
+      "step": 88780
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0946724143604597e-05,
+      "loss": 1.2067,
+      "step": 88790
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0943433479219454e-05,
+      "loss": 1.2076,
+      "step": 88800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0940142814834316e-05,
+      "loss": 1.2791,
+      "step": 88810
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0936852150449177e-05,
+      "loss": 1.2266,
+      "step": 88820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0933561486064038e-05,
+      "loss": 1.3378,
+      "step": 88830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.09302708216789e-05,
+      "loss": 1.2739,
+      "step": 88840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0926980157293757e-05,
+      "loss": 1.2367,
+      "step": 88850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.092368949290862e-05,
+      "loss": 1.3458,
+      "step": 88860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.092039882852348e-05,
+      "loss": 1.287,
+      "step": 88870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.091710816413834e-05,
+      "loss": 1.2719,
+      "step": 88880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0913817499753202e-05,
+      "loss": 1.2618,
+      "step": 88890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.091052683536806e-05,
+      "loss": 1.2401,
+      "step": 88900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.090723617098292e-05,
+      "loss": 1.2556,
+      "step": 88910
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0903945506597783e-05,
+      "loss": 1.34,
+      "step": 88920
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0900654842212644e-05,
+      "loss": 1.335,
+      "step": 88930
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0897364177827505e-05,
+      "loss": 1.1468,
+      "step": 88940
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0894073513442363e-05,
+      "loss": 1.3254,
+      "step": 88950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0890782849057225e-05,
+      "loss": 1.4265,
+      "step": 88960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0887492184672086e-05,
+      "loss": 1.3163,
+      "step": 88970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0884201520286947e-05,
+      "loss": 1.3028,
+      "step": 88980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.088091085590181e-05,
+      "loss": 1.1275,
+      "step": 88990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0877620191516666e-05,
+      "loss": 1.4468,
+      "step": 89000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0874329527131528e-05,
+      "loss": 1.2888,
+      "step": 89010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.087103886274639e-05,
+      "loss": 1.3948,
+      "step": 89020
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.086774819836125e-05,
+      "loss": 1.3645,
+      "step": 89030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.086445753397611e-05,
+      "loss": 1.309,
+      "step": 89040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.086116686959097e-05,
+      "loss": 1.288,
+      "step": 89050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.085787620520583e-05,
+      "loss": 1.3539,
+      "step": 89060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0854585540820692e-05,
+      "loss": 1.2297,
+      "step": 89070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0851294876435553e-05,
+      "loss": 1.2729,
+      "step": 89080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0848004212050414e-05,
+      "loss": 1.2969,
+      "step": 89090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0844713547665272e-05,
+      "loss": 1.4066,
+      "step": 89100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0841422883280133e-05,
+      "loss": 1.2733,
+      "step": 89110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0838132218894995e-05,
+      "loss": 1.23,
+      "step": 89120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0834841554509856e-05,
+      "loss": 1.0991,
+      "step": 89130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0831550890124717e-05,
+      "loss": 1.2146,
+      "step": 89140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0828260225739575e-05,
+      "loss": 1.2816,
+      "step": 89150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0824969561354436e-05,
+      "loss": 1.2308,
+      "step": 89160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0821678896969298e-05,
+      "loss": 1.0914,
+      "step": 89170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.081838823258416e-05,
+      "loss": 1.2597,
+      "step": 89180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.081509756819902e-05,
+      "loss": 1.2756,
+      "step": 89190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0811806903813878e-05,
+      "loss": 1.172,
+      "step": 89200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.080851623942874e-05,
+      "loss": 1.0637,
+      "step": 89210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.08052255750436e-05,
+      "loss": 1.2871,
+      "step": 89220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0801934910658462e-05,
+      "loss": 1.1817,
+      "step": 89230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0798644246273323e-05,
+      "loss": 1.2969,
+      "step": 89240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0795353581888184e-05,
+      "loss": 1.2799,
+      "step": 89250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0792062917503046e-05,
+      "loss": 1.321,
+      "step": 89260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0788772253117904e-05,
+      "loss": 1.4372,
+      "step": 89270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0785481588732765e-05,
+      "loss": 1.2249,
+      "step": 89280
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0782190924347626e-05,
+      "loss": 1.117,
+      "step": 89290
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0778900259962487e-05,
+      "loss": 1.1178,
+      "step": 89300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.077560959557735e-05,
+      "loss": 1.3087,
+      "step": 89310
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.077231893119221e-05,
+      "loss": 1.1572,
+      "step": 89320
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.076902826680707e-05,
+      "loss": 1.116,
+      "step": 89330
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0765737602421932e-05,
+      "loss": 1.2255,
+      "step": 89340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.076244693803679e-05,
+      "loss": 1.1592,
+      "step": 89350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.075915627365165e-05,
+      "loss": 1.335,
+      "step": 89360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0755865609266513e-05,
+      "loss": 1.3226,
+      "step": 89370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0752574944881374e-05,
+      "loss": 1.3215,
+      "step": 89380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0749284280496235e-05,
+      "loss": 1.0823,
+      "step": 89390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0745993616111097e-05,
+      "loss": 1.2793,
+      "step": 89400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0742702951725955e-05,
+      "loss": 1.2073,
+      "step": 89410
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0739412287340816e-05,
+      "loss": 1.2895,
+      "step": 89420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0736121622955677e-05,
+      "loss": 1.2914,
+      "step": 89430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.073283095857054e-05,
+      "loss": 1.3168,
+      "step": 89440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.07295402941854e-05,
+      "loss": 1.0885,
+      "step": 89450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0726249629800257e-05,
+      "loss": 1.2072,
+      "step": 89460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.072295896541512e-05,
+      "loss": 1.2987,
+      "step": 89470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.071966830102998e-05,
+      "loss": 1.1233,
+      "step": 89480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.071637763664484e-05,
+      "loss": 1.3326,
+      "step": 89490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0713086972259703e-05,
+      "loss": 1.239,
+      "step": 89500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.070979630787456e-05,
+      "loss": 1.3496,
+      "step": 89510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.070650564348942e-05,
+      "loss": 1.3076,
+      "step": 89520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0703214979104283e-05,
+      "loss": 1.2308,
+      "step": 89530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0699924314719144e-05,
+      "loss": 1.3168,
+      "step": 89540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0696633650334005e-05,
+      "loss": 1.187,
+      "step": 89550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0693342985948863e-05,
+      "loss": 1.2616,
+      "step": 89560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0690052321563725e-05,
+      "loss": 1.158,
+      "step": 89570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0686761657178586e-05,
+      "loss": 1.2879,
+      "step": 89580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0683470992793447e-05,
+      "loss": 1.2179,
+      "step": 89590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.068018032840831e-05,
+      "loss": 1.2088,
+      "step": 89600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0676889664023166e-05,
+      "loss": 1.2142,
+      "step": 89610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0673598999638028e-05,
+      "loss": 1.3584,
+      "step": 89620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.067030833525289e-05,
+      "loss": 1.3868,
+      "step": 89630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.066701767086775e-05,
+      "loss": 1.199,
+      "step": 89640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.066372700648261e-05,
+      "loss": 1.4292,
+      "step": 89650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.066043634209747e-05,
+      "loss": 1.1854,
+      "step": 89660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.065714567771233e-05,
+      "loss": 1.1496,
+      "step": 89670
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0653855013327192e-05,
+      "loss": 1.2414,
+      "step": 89680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0650564348942053e-05,
+      "loss": 1.0629,
+      "step": 89690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0647273684556914e-05,
+      "loss": 1.2043,
+      "step": 89700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0643983020171772e-05,
+      "loss": 1.251,
+      "step": 89710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0640692355786633e-05,
+      "loss": 1.1539,
+      "step": 89720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0637401691401495e-05,
+      "loss": 1.2477,
+      "step": 89730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0634111027016356e-05,
+      "loss": 1.251,
+      "step": 89740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0630820362631217e-05,
+      "loss": 1.1727,
+      "step": 89750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0627529698246075e-05,
+      "loss": 1.3146,
+      "step": 89760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0624239033860936e-05,
+      "loss": 1.2124,
+      "step": 89770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0620948369475798e-05,
+      "loss": 1.176,
+      "step": 89780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.061765770509066e-05,
+      "loss": 1.2688,
+      "step": 89790
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.061436704070552e-05,
+      "loss": 1.0554,
+      "step": 89800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0611076376320378e-05,
+      "loss": 1.3701,
+      "step": 89810
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.060778571193524e-05,
+      "loss": 1.3795,
+      "step": 89820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.06044950475501e-05,
+      "loss": 1.2289,
+      "step": 89830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0601204383164962e-05,
+      "loss": 1.2997,
+      "step": 89840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0597913718779823e-05,
+      "loss": 1.3139,
+      "step": 89850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.059462305439468e-05,
+      "loss": 1.3449,
+      "step": 89860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0591332390009542e-05,
+      "loss": 1.3728,
+      "step": 89870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0588041725624404e-05,
+      "loss": 1.2278,
+      "step": 89880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0584751061239265e-05,
+      "loss": 1.1898,
+      "step": 89890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0581460396854126e-05,
+      "loss": 1.3155,
+      "step": 89900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0578169732468984e-05,
+      "loss": 1.2873,
+      "step": 89910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0574879068083845e-05,
+      "loss": 1.2544,
+      "step": 89920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0571588403698707e-05,
+      "loss": 1.1822,
+      "step": 89930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0568297739313568e-05,
+      "loss": 1.0774,
+      "step": 89940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.056500707492843e-05,
+      "loss": 1.3602,
+      "step": 89950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0561716410543287e-05,
+      "loss": 1.1062,
+      "step": 89960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0558425746158148e-05,
+      "loss": 1.2024,
+      "step": 89970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.055513508177301e-05,
+      "loss": 1.1848,
+      "step": 89980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.055184441738787e-05,
+      "loss": 1.2043,
+      "step": 89990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0548553753002732e-05,
+      "loss": 1.2915,
+      "step": 90000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0545263088617593e-05,
+      "loss": 1.0767,
+      "step": 90010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0541972424232455e-05,
+      "loss": 1.0937,
+      "step": 90020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0538681759847312e-05,
+      "loss": 1.2118,
+      "step": 90030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0535391095462174e-05,
+      "loss": 1.2749,
+      "step": 90040
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0532100431077035e-05,
+      "loss": 1.1735,
+      "step": 90050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0528809766691896e-05,
+      "loss": 1.2752,
+      "step": 90060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0525519102306758e-05,
+      "loss": 1.2521,
+      "step": 90070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.052222843792162e-05,
+      "loss": 1.3516,
+      "step": 90080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.051893777353648e-05,
+      "loss": 1.1798,
+      "step": 90090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.051564710915134e-05,
+      "loss": 1.2759,
+      "step": 90100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.05123564447662e-05,
+      "loss": 1.307,
+      "step": 90110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.050906578038106e-05,
+      "loss": 1.4349,
+      "step": 90120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0505775115995922e-05,
+      "loss": 1.2532,
+      "step": 90130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0502484451610783e-05,
+      "loss": 1.2077,
+      "step": 90140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0499193787225644e-05,
+      "loss": 1.3397,
+      "step": 90150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0495903122840502e-05,
+      "loss": 1.2849,
+      "step": 90160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0492612458455363e-05,
+      "loss": 1.4365,
+      "step": 90170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0489321794070225e-05,
+      "loss": 1.2841,
+      "step": 90180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0486031129685086e-05,
+      "loss": 1.1398,
+      "step": 90190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0482740465299947e-05,
+      "loss": 1.3568,
+      "step": 90200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0479449800914805e-05,
+      "loss": 1.1874,
+      "step": 90210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0476159136529666e-05,
+      "loss": 1.2535,
+      "step": 90220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0472868472144528e-05,
+      "loss": 1.2692,
+      "step": 90230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.046957780775939e-05,
+      "loss": 1.3612,
+      "step": 90240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.046628714337425e-05,
+      "loss": 1.3013,
+      "step": 90250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0462996478989108e-05,
+      "loss": 1.3323,
+      "step": 90260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.045970581460397e-05,
+      "loss": 1.1977,
+      "step": 90270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.045641515021883e-05,
+      "loss": 1.1408,
+      "step": 90280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0453124485833692e-05,
+      "loss": 1.3223,
+      "step": 90290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0449833821448553e-05,
+      "loss": 1.5364,
+      "step": 90300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.044654315706341e-05,
+      "loss": 1.2452,
+      "step": 90310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0443252492678272e-05,
+      "loss": 1.2744,
+      "step": 90320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0439961828293134e-05,
+      "loss": 1.2045,
+      "step": 90330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0436671163907995e-05,
+      "loss": 1.4867,
+      "step": 90340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0433380499522856e-05,
+      "loss": 1.2216,
+      "step": 90350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0430089835137714e-05,
+      "loss": 1.1649,
+      "step": 90360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0426799170752575e-05,
+      "loss": 1.1341,
+      "step": 90370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0423508506367436e-05,
+      "loss": 1.2863,
+      "step": 90380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0420217841982298e-05,
+      "loss": 1.0835,
+      "step": 90390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.041692717759716e-05,
+      "loss": 1.3585,
+      "step": 90400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0413636513212017e-05,
+      "loss": 1.3901,
+      "step": 90410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0410345848826878e-05,
+      "loss": 1.3056,
+      "step": 90420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.040705518444174e-05,
+      "loss": 1.1835,
+      "step": 90430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.04037645200566e-05,
+      "loss": 1.1203,
+      "step": 90440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0400473855671462e-05,
+      "loss": 1.2347,
+      "step": 90450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.039718319128632e-05,
+      "loss": 1.3381,
+      "step": 90460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.039389252690118e-05,
+      "loss": 1.2553,
+      "step": 90470
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0390601862516042e-05,
+      "loss": 1.4122,
+      "step": 90480
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0387311198130904e-05,
+      "loss": 1.278,
+      "step": 90490
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0384020533745765e-05,
+      "loss": 1.2651,
+      "step": 90500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0380729869360623e-05,
+      "loss": 1.2276,
+      "step": 90510
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0377439204975484e-05,
+      "loss": 1.2367,
+      "step": 90520
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0374148540590345e-05,
+      "loss": 1.1549,
+      "step": 90530
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0370857876205207e-05,
+      "loss": 1.251,
+      "step": 90540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0367567211820068e-05,
+      "loss": 1.2981,
+      "step": 90550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0364276547434926e-05,
+      "loss": 1.57,
+      "step": 90560
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0360985883049787e-05,
+      "loss": 1.4008,
+      "step": 90570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0357695218664648e-05,
+      "loss": 1.1793,
+      "step": 90580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.035440455427951e-05,
+      "loss": 1.3261,
+      "step": 90590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.035111388989437e-05,
+      "loss": 1.2466,
+      "step": 90600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.034782322550923e-05,
+      "loss": 1.1962,
+      "step": 90610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.034453256112409e-05,
+      "loss": 1.1101,
+      "step": 90620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.034124189673895e-05,
+      "loss": 1.2826,
+      "step": 90630
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0337951232353812e-05,
+      "loss": 1.3136,
+      "step": 90640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0334660567968674e-05,
+      "loss": 1.3026,
+      "step": 90650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.033136990358353e-05,
+      "loss": 1.0438,
+      "step": 90660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0328079239198393e-05,
+      "loss": 1.3414,
+      "step": 90670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0324788574813254e-05,
+      "loss": 1.3476,
+      "step": 90680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0321497910428115e-05,
+      "loss": 1.0761,
+      "step": 90690
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0318207246042977e-05,
+      "loss": 1.2522,
+      "step": 90700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0314916581657838e-05,
+      "loss": 1.191,
+      "step": 90710
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0311625917272696e-05,
+      "loss": 1.2324,
+      "step": 90720
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0308335252887557e-05,
+      "loss": 1.313,
+      "step": 90730
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.030504458850242e-05,
+      "loss": 1.3783,
+      "step": 90740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.030175392411728e-05,
+      "loss": 1.4511,
+      "step": 90750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.029846325973214e-05,
+      "loss": 1.2834,
+      "step": 90760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0295172595347002e-05,
+      "loss": 1.185,
+      "step": 90770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0291881930961863e-05,
+      "loss": 1.2393,
+      "step": 90780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0288591266576725e-05,
+      "loss": 1.237,
+      "step": 90790
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0285300602191583e-05,
+      "loss": 1.1889,
+      "step": 90800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0282009937806444e-05,
+      "loss": 1.4207,
+      "step": 90810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0278719273421305e-05,
+      "loss": 1.2576,
+      "step": 90820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0275428609036166e-05,
+      "loss": 1.2038,
+      "step": 90830
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0272137944651028e-05,
+      "loss": 1.0974,
+      "step": 90840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.026884728026589e-05,
+      "loss": 1.2495,
+      "step": 90850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.026555661588075e-05,
+      "loss": 1.4038,
+      "step": 90860
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0262265951495608e-05,
+      "loss": 1.2545,
+      "step": 90870
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.025897528711047e-05,
+      "loss": 1.2391,
+      "step": 90880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.025568462272533e-05,
+      "loss": 1.1384,
+      "step": 90890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0252393958340192e-05,
+      "loss": 1.2125,
+      "step": 90900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0249103293955053e-05,
+      "loss": 1.2546,
+      "step": 90910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.024581262956991e-05,
+      "loss": 1.3938,
+      "step": 90920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0242521965184772e-05,
+      "loss": 1.1863,
+      "step": 90930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0239231300799634e-05,
+      "loss": 1.2963,
+      "step": 90940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0235940636414495e-05,
+      "loss": 1.231,
+      "step": 90950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0232649972029356e-05,
+      "loss": 1.226,
+      "step": 90960
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0229359307644214e-05,
+      "loss": 1.2248,
+      "step": 90970
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0226068643259075e-05,
+      "loss": 1.2923,
+      "step": 90980
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0222777978873937e-05,
+      "loss": 1.3323,
+      "step": 90990
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0219487314488798e-05,
+      "loss": 1.229,
+      "step": 91000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.021619665010366e-05,
+      "loss": 1.2686,
+      "step": 91010
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0212905985718517e-05,
+      "loss": 1.2168,
+      "step": 91020
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0209615321333378e-05,
+      "loss": 1.2814,
+      "step": 91030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.020632465694824e-05,
+      "loss": 1.0863,
+      "step": 91040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.02030339925631e-05,
+      "loss": 1.4883,
+      "step": 91050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0199743328177962e-05,
+      "loss": 1.2742,
+      "step": 91060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.019645266379282e-05,
+      "loss": 1.2662,
+      "step": 91070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.019316199940768e-05,
+      "loss": 1.1974,
+      "step": 91080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0189871335022542e-05,
+      "loss": 1.1385,
+      "step": 91090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0186580670637404e-05,
+      "loss": 1.198,
+      "step": 91100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0183290006252265e-05,
+      "loss": 1.2309,
+      "step": 91110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0179999341867123e-05,
+      "loss": 1.2244,
+      "step": 91120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0176708677481984e-05,
+      "loss": 1.3656,
+      "step": 91130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0173418013096845e-05,
+      "loss": 1.1311,
+      "step": 91140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0170127348711707e-05,
+      "loss": 1.2336,
+      "step": 91150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0166836684326568e-05,
+      "loss": 1.1507,
+      "step": 91160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0163546019941426e-05,
+      "loss": 1.3557,
+      "step": 91170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0160255355556287e-05,
+      "loss": 1.1992,
+      "step": 91180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.015696469117115e-05,
+      "loss": 1.1407,
+      "step": 91190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.015367402678601e-05,
+      "loss": 1.2501,
+      "step": 91200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.015038336240087e-05,
+      "loss": 1.2807,
+      "step": 91210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.014709269801573e-05,
+      "loss": 1.1344,
+      "step": 91220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.014380203363059e-05,
+      "loss": 1.2944,
+      "step": 91230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.014051136924545e-05,
+      "loss": 1.2882,
+      "step": 91240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0137220704860313e-05,
+      "loss": 1.2742,
+      "step": 91250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0133930040475174e-05,
+      "loss": 1.2938,
+      "step": 91260
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.013063937609003e-05,
+      "loss": 1.1892,
+      "step": 91270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0127348711704893e-05,
+      "loss": 1.234,
+      "step": 91280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0124058047319754e-05,
+      "loss": 1.2275,
+      "step": 91290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0120767382934615e-05,
+      "loss": 1.1657,
+      "step": 91300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0117476718549477e-05,
+      "loss": 1.2312,
+      "step": 91310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0114186054164335e-05,
+      "loss": 1.1967,
+      "step": 91320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0110895389779196e-05,
+      "loss": 1.1632,
+      "step": 91330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0107604725394057e-05,
+      "loss": 1.3919,
+      "step": 91340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.010431406100892e-05,
+      "loss": 1.2681,
+      "step": 91350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.010102339662378e-05,
+      "loss": 1.3731,
+      "step": 91360
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0097732732238638e-05,
+      "loss": 1.1164,
+      "step": 91370
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.00944420678535e-05,
+      "loss": 1.4557,
+      "step": 91380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.009115140346836e-05,
+      "loss": 1.1205,
+      "step": 91390
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.008786073908322e-05,
+      "loss": 1.3029,
+      "step": 91400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0084570074698083e-05,
+      "loss": 1.2304,
+      "step": 91410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.008127941031294e-05,
+      "loss": 1.2322,
+      "step": 91420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0077988745927802e-05,
+      "loss": 1.3817,
+      "step": 91430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0074698081542663e-05,
+      "loss": 1.1258,
+      "step": 91440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0071407417157524e-05,
+      "loss": 1.1558,
+      "step": 91450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0068116752772386e-05,
+      "loss": 1.0838,
+      "step": 91460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0064826088387247e-05,
+      "loss": 1.1992,
+      "step": 91470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0061535424002105e-05,
+      "loss": 1.2132,
+      "step": 91480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0058244759616966e-05,
+      "loss": 1.2191,
+      "step": 91490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0054954095231827e-05,
+      "loss": 1.2477,
+      "step": 91500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.005166343084669e-05,
+      "loss": 1.3541,
+      "step": 91510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.004837276646155e-05,
+      "loss": 1.0018,
+      "step": 91520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.004508210207641e-05,
+      "loss": 1.0982,
+      "step": 91530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0041791437691272e-05,
+      "loss": 1.2258,
+      "step": 91540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0038500773306134e-05,
+      "loss": 1.0829,
+      "step": 91550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.003521010892099e-05,
+      "loss": 1.1886,
+      "step": 91560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0031919444535853e-05,
+      "loss": 1.1651,
+      "step": 91570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0028628780150714e-05,
+      "loss": 1.271,
+      "step": 91580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0025338115765575e-05,
+      "loss": 1.2311,
+      "step": 91590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0022047451380437e-05,
+      "loss": 1.473,
+      "step": 91600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0018756786995298e-05,
+      "loss": 1.3284,
+      "step": 91610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0015466122610156e-05,
+      "loss": 1.1959,
+      "step": 91620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0012175458225017e-05,
+      "loss": 1.2552,
+      "step": 91630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0008884793839878e-05,
+      "loss": 1.257,
+      "step": 91640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.000559412945474e-05,
+      "loss": 1.1596,
+      "step": 91650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.00023034650696e-05,
+      "loss": 1.2344,
+      "step": 91660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.999901280068446e-05,
+      "loss": 1.371,
+      "step": 91670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.999572213629932e-05,
+      "loss": 1.4177,
+      "step": 91680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.999243147191418e-05,
+      "loss": 1.2076,
+      "step": 91690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9989140807529042e-05,
+      "loss": 1.2194,
+      "step": 91700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9985850143143904e-05,
+      "loss": 1.2311,
+      "step": 91710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.998255947875876e-05,
+      "loss": 1.1451,
+      "step": 91720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9979268814373623e-05,
+      "loss": 1.2466,
+      "step": 91730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9975978149988484e-05,
+      "loss": 1.3617,
+      "step": 91740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9972687485603345e-05,
+      "loss": 1.2122,
+      "step": 91750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9969396821218207e-05,
+      "loss": 1.4077,
+      "step": 91760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9966106156833065e-05,
+      "loss": 1.2658,
+      "step": 91770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9962815492447926e-05,
+      "loss": 1.3746,
+      "step": 91780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9959524828062787e-05,
+      "loss": 1.2319,
+      "step": 91790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.995623416367765e-05,
+      "loss": 1.1798,
+      "step": 91800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.995294349929251e-05,
+      "loss": 1.3646,
+      "step": 91810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9949652834907368e-05,
+      "loss": 1.2311,
+      "step": 91820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.994636217052223e-05,
+      "loss": 1.1522,
+      "step": 91830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.994307150613709e-05,
+      "loss": 1.514,
+      "step": 91840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.993978084175195e-05,
+      "loss": 1.1779,
+      "step": 91850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9936490177366813e-05,
+      "loss": 1.184,
+      "step": 91860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.993319951298167e-05,
+      "loss": 1.2122,
+      "step": 91870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9929908848596532e-05,
+      "loss": 1.2514,
+      "step": 91880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9926618184211393e-05,
+      "loss": 1.214,
+      "step": 91890
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9923327519826254e-05,
+      "loss": 1.0765,
+      "step": 91900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9920036855441116e-05,
+      "loss": 1.3041,
+      "step": 91910
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9916746191055973e-05,
+      "loss": 1.2277,
+      "step": 91920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9913455526670835e-05,
+      "loss": 1.2682,
+      "step": 91930
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9910164862285696e-05,
+      "loss": 1.2071,
+      "step": 91940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9906874197900557e-05,
+      "loss": 1.279,
+      "step": 91950
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.990358353351542e-05,
+      "loss": 1.2351,
+      "step": 91960
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9900292869130276e-05,
+      "loss": 1.1,
+      "step": 91970
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9897002204745138e-05,
+      "loss": 1.2734,
+      "step": 91980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.989371154036e-05,
+      "loss": 1.2237,
+      "step": 91990
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.989042087597486e-05,
+      "loss": 1.2254,
+      "step": 92000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.988713021158972e-05,
+      "loss": 1.2917,
+      "step": 92010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.988383954720458e-05,
+      "loss": 1.2143,
+      "step": 92020
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.988054888281944e-05,
+      "loss": 1.2245,
+      "step": 92030
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9877258218434302e-05,
+      "loss": 1.0799,
+      "step": 92040
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9873967554049163e-05,
+      "loss": 1.0552,
+      "step": 92050
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9870676889664024e-05,
+      "loss": 1.1113,
+      "step": 92060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9867386225278882e-05,
+      "loss": 1.1334,
+      "step": 92070
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9864095560893744e-05,
+      "loss": 1.1856,
+      "step": 92080
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9860804896508605e-05,
+      "loss": 1.2279,
+      "step": 92090
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9857514232123466e-05,
+      "loss": 1.3042,
+      "step": 92100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9854223567738327e-05,
+      "loss": 1.11,
+      "step": 92110
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.985093290335319e-05,
+      "loss": 1.3077,
+      "step": 92120
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9847642238968046e-05,
+      "loss": 1.2107,
+      "step": 92130
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9844351574582908e-05,
+      "loss": 1.1905,
+      "step": 92140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.984106091019777e-05,
+      "loss": 1.2912,
+      "step": 92150
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.983777024581263e-05,
+      "loss": 1.1779,
+      "step": 92160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.983447958142749e-05,
+      "loss": 1.2994,
+      "step": 92170
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.983118891704235e-05,
+      "loss": 1.2075,
+      "step": 92180
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.982789825265721e-05,
+      "loss": 1.1589,
+      "step": 92190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9824607588272072e-05,
+      "loss": 1.308,
+      "step": 92200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9821316923886933e-05,
+      "loss": 1.2534,
+      "step": 92210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9818026259501794e-05,
+      "loss": 1.2425,
+      "step": 92220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9814735595116656e-05,
+      "loss": 1.1301,
+      "step": 92230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9811444930731514e-05,
+      "loss": 1.2385,
+      "step": 92240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9808154266346375e-05,
+      "loss": 1.1705,
+      "step": 92250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804863601961236e-05,
+      "loss": 1.1412,
+      "step": 92260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9801572937576097e-05,
+      "loss": 1.1227,
+      "step": 92270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.979828227319096e-05,
+      "loss": 1.2565,
+      "step": 92280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.979499160880582e-05,
+      "loss": 1.3326,
+      "step": 92290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.979170094442068e-05,
+      "loss": 1.2436,
+      "step": 92300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9788410280035543e-05,
+      "loss": 1.2854,
+      "step": 92310
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.97851196156504e-05,
+      "loss": 1.1893,
+      "step": 92320
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.978182895126526e-05,
+      "loss": 1.2176,
+      "step": 92330
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9778538286880123e-05,
+      "loss": 1.4215,
+      "step": 92340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9775247622494984e-05,
+      "loss": 1.1561,
+      "step": 92350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9771956958109845e-05,
+      "loss": 1.2559,
+      "step": 92360
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9768666293724707e-05,
+      "loss": 1.0954,
+      "step": 92370
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9765375629339565e-05,
+      "loss": 1.2989,
+      "step": 92380
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9762084964954426e-05,
+      "loss": 1.1427,
+      "step": 92390
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9758794300569287e-05,
+      "loss": 1.4496,
+      "step": 92400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.975550363618415e-05,
+      "loss": 1.2229,
+      "step": 92410
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.975221297179901e-05,
+      "loss": 1.1867,
+      "step": 92420
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9748922307413868e-05,
+      "loss": 1.1351,
+      "step": 92430
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.974563164302873e-05,
+      "loss": 1.2554,
+      "step": 92440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.974234097864359e-05,
+      "loss": 1.2397,
+      "step": 92450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.973905031425845e-05,
+      "loss": 1.4249,
+      "step": 92460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9735759649873313e-05,
+      "loss": 1.2314,
+      "step": 92470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.973246898548817e-05,
+      "loss": 1.1985,
+      "step": 92480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9729178321103032e-05,
+      "loss": 1.1936,
+      "step": 92490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9725887656717893e-05,
+      "loss": 1.1134,
+      "step": 92500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9722596992332754e-05,
+      "loss": 1.096,
+      "step": 92510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9719306327947616e-05,
+      "loss": 1.1168,
+      "step": 92520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9716015663562473e-05,
+      "loss": 1.2092,
+      "step": 92530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9712724999177335e-05,
+      "loss": 1.1552,
+      "step": 92540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9709434334792196e-05,
+      "loss": 1.3374,
+      "step": 92550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9706143670407057e-05,
+      "loss": 1.1151,
+      "step": 92560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.970285300602192e-05,
+      "loss": 1.1804,
+      "step": 92570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9699562341636776e-05,
+      "loss": 1.0951,
+      "step": 92580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9696271677251638e-05,
+      "loss": 1.1154,
+      "step": 92590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.96929810128665e-05,
+      "loss": 1.2605,
+      "step": 92600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.968969034848136e-05,
+      "loss": 1.3012,
+      "step": 92610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.968639968409622e-05,
+      "loss": 1.1651,
+      "step": 92620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.968310901971108e-05,
+      "loss": 1.4397,
+      "step": 92630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.967981835532594e-05,
+      "loss": 1.2034,
+      "step": 92640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9676527690940802e-05,
+      "loss": 1.3159,
+      "step": 92650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9673237026555663e-05,
+      "loss": 1.2345,
+      "step": 92660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9669946362170524e-05,
+      "loss": 1.3696,
+      "step": 92670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9666655697785382e-05,
+      "loss": 1.3293,
+      "step": 92680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9663365033400244e-05,
+      "loss": 1.3163,
+      "step": 92690
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9660074369015105e-05,
+      "loss": 1.2508,
+      "step": 92700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9656783704629966e-05,
+      "loss": 1.2015,
+      "step": 92710
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9653493040244827e-05,
+      "loss": 1.2538,
+      "step": 92720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9650202375859685e-05,
+      "loss": 1.1735,
+      "step": 92730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9646911711474547e-05,
+      "loss": 1.3791,
+      "step": 92740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9643621047089408e-05,
+      "loss": 1.1372,
+      "step": 92750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.964033038270427e-05,
+      "loss": 1.1886,
+      "step": 92760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.963703971831913e-05,
+      "loss": 1.1933,
+      "step": 92770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9633749053933988e-05,
+      "loss": 1.1691,
+      "step": 92780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.963045838954885e-05,
+      "loss": 1.249,
+      "step": 92790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.962716772516371e-05,
+      "loss": 1.3836,
+      "step": 92800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9623877060778572e-05,
+      "loss": 1.3041,
+      "step": 92810
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9620586396393433e-05,
+      "loss": 1.1521,
+      "step": 92820
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.961729573200829e-05,
+      "loss": 1.3243,
+      "step": 92830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9614005067623152e-05,
+      "loss": 1.2476,
+      "step": 92840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9610714403238014e-05,
+      "loss": 1.1995,
+      "step": 92850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9607423738852875e-05,
+      "loss": 1.3036,
+      "step": 92860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9604133074467736e-05,
+      "loss": 1.094,
+      "step": 92870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9600842410082594e-05,
+      "loss": 1.192,
+      "step": 92880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9597551745697455e-05,
+      "loss": 1.1404,
+      "step": 92890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9594261081312317e-05,
+      "loss": 1.2715,
+      "step": 92900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9590970416927178e-05,
+      "loss": 1.239,
+      "step": 92910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.958767975254204e-05,
+      "loss": 1.262,
+      "step": 92920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9584389088156897e-05,
+      "loss": 1.1132,
+      "step": 92930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.958109842377176e-05,
+      "loss": 1.2854,
+      "step": 92940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.957780775938662e-05,
+      "loss": 1.1024,
+      "step": 92950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.957451709500148e-05,
+      "loss": 1.2701,
+      "step": 92960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9571226430616342e-05,
+      "loss": 1.2683,
+      "step": 92970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9567935766231203e-05,
+      "loss": 1.1412,
+      "step": 92980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9564645101846065e-05,
+      "loss": 1.4183,
+      "step": 92990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9561354437460923e-05,
+      "loss": 1.3073,
+      "step": 93000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9558063773075784e-05,
+      "loss": 1.3388,
+      "step": 93010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9554773108690645e-05,
+      "loss": 1.2265,
+      "step": 93020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9551482444305506e-05,
+      "loss": 1.1731,
+      "step": 93030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9548191779920368e-05,
+      "loss": 1.2522,
+      "step": 93040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.954490111553523e-05,
+      "loss": 1.2517,
+      "step": 93050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.954161045115009e-05,
+      "loss": 1.111,
+      "step": 93060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.953831978676495e-05,
+      "loss": 1.21,
+      "step": 93070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.953502912237981e-05,
+      "loss": 1.1734,
+      "step": 93080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.953173845799467e-05,
+      "loss": 1.3036,
+      "step": 93090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9528447793609532e-05,
+      "loss": 1.4234,
+      "step": 93100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9525157129224393e-05,
+      "loss": 1.2176,
+      "step": 93110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9521866464839254e-05,
+      "loss": 1.1521,
+      "step": 93120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9518575800454112e-05,
+      "loss": 1.242,
+      "step": 93130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9515285136068973e-05,
+      "loss": 1.1756,
+      "step": 93140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9511994471683835e-05,
+      "loss": 1.1683,
+      "step": 93150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9508703807298696e-05,
+      "loss": 1.1517,
+      "step": 93160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9505413142913557e-05,
+      "loss": 1.3298,
+      "step": 93170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9502122478528415e-05,
+      "loss": 1.0399,
+      "step": 93180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9498831814143276e-05,
+      "loss": 1.1447,
+      "step": 93190
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9495541149758138e-05,
+      "loss": 1.2454,
+      "step": 93200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9492250485373e-05,
+      "loss": 1.1353,
+      "step": 93210
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.948895982098786e-05,
+      "loss": 1.2881,
+      "step": 93220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9485669156602718e-05,
+      "loss": 1.1783,
+      "step": 93230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.948237849221758e-05,
+      "loss": 1.2093,
+      "step": 93240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.947908782783244e-05,
+      "loss": 1.1782,
+      "step": 93250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9475797163447302e-05,
+      "loss": 1.4173,
+      "step": 93260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9472506499062163e-05,
+      "loss": 1.2849,
+      "step": 93270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.946921583467702e-05,
+      "loss": 1.0543,
+      "step": 93280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9465925170291882e-05,
+      "loss": 1.1794,
+      "step": 93290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9462634505906744e-05,
+      "loss": 1.258,
+      "step": 93300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9459343841521605e-05,
+      "loss": 1.3925,
+      "step": 93310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9456053177136466e-05,
+      "loss": 1.4091,
+      "step": 93320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9452762512751324e-05,
+      "loss": 1.0871,
+      "step": 93330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9449471848366185e-05,
+      "loss": 1.36,
+      "step": 93340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9446181183981047e-05,
+      "loss": 1.3165,
+      "step": 93350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9442890519595908e-05,
+      "loss": 1.2493,
+      "step": 93360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.943959985521077e-05,
+      "loss": 1.3423,
+      "step": 93370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9436309190825627e-05,
+      "loss": 1.1992,
+      "step": 93380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9433018526440488e-05,
+      "loss": 1.5882,
+      "step": 93390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.942972786205535e-05,
+      "loss": 1.1546,
+      "step": 93400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.942643719767021e-05,
+      "loss": 1.2474,
+      "step": 93410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9423146533285072e-05,
+      "loss": 1.3308,
+      "step": 93420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.941985586889993e-05,
+      "loss": 1.2843,
+      "step": 93430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.941656520451479e-05,
+      "loss": 1.2492,
+      "step": 93440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9413274540129652e-05,
+      "loss": 1.1106,
+      "step": 93450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9409983875744514e-05,
+      "loss": 1.2429,
+      "step": 93460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9406693211359375e-05,
+      "loss": 1.1362,
+      "step": 93470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9403402546974233e-05,
+      "loss": 1.3001,
+      "step": 93480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9400111882589094e-05,
+      "loss": 1.2094,
+      "step": 93490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9396821218203955e-05,
+      "loss": 1.3149,
+      "step": 93500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9393530553818817e-05,
+      "loss": 1.2661,
+      "step": 93510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9390239889433678e-05,
+      "loss": 1.2949,
+      "step": 93520
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938694922504854e-05,
+      "loss": 1.1748,
+      "step": 93530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9383658560663397e-05,
+      "loss": 1.4326,
+      "step": 93540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938036789627826e-05,
+      "loss": 1.2184,
+      "step": 93550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.937707723189312e-05,
+      "loss": 1.202,
+      "step": 93560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.937378656750798e-05,
+      "loss": 1.2469,
+      "step": 93570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9370495903122842e-05,
+      "loss": 1.2284,
+      "step": 93580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.93672052387377e-05,
+      "loss": 1.4266,
+      "step": 93590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.936391457435256e-05,
+      "loss": 1.2748,
+      "step": 93600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9360623909967423e-05,
+      "loss": 1.2082,
+      "step": 93610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9357333245582284e-05,
+      "loss": 1.2589,
+      "step": 93620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9354042581197145e-05,
+      "loss": 1.2203,
+      "step": 93630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9350751916812003e-05,
+      "loss": 1.2009,
+      "step": 93640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9347461252426864e-05,
+      "loss": 1.1955,
+      "step": 93650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9344170588041726e-05,
+      "loss": 1.0685,
+      "step": 93660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9340879923656587e-05,
+      "loss": 1.3198,
+      "step": 93670
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9337589259271448e-05,
+      "loss": 1.4448,
+      "step": 93680
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9334298594886306e-05,
+      "loss": 1.3117,
+      "step": 93690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9331007930501167e-05,
+      "loss": 1.2797,
+      "step": 93700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.932771726611603e-05,
+      "loss": 1.1884,
+      "step": 93710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.932442660173089e-05,
+      "loss": 1.1818,
+      "step": 93720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.932113593734575e-05,
+      "loss": 1.2039,
+      "step": 93730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9317845272960612e-05,
+      "loss": 1.1318,
+      "step": 93740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9314554608575474e-05,
+      "loss": 1.5055,
+      "step": 93750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9311263944190335e-05,
+      "loss": 1.2763,
+      "step": 93760
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9307973279805193e-05,
+      "loss": 1.2096,
+      "step": 93770
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9304682615420054e-05,
+      "loss": 1.2406,
+      "step": 93780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9301391951034915e-05,
+      "loss": 1.1299,
+      "step": 93790
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9298101286649776e-05,
+      "loss": 1.2517,
+      "step": 93800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9294810622264638e-05,
+      "loss": 1.0364,
+      "step": 93810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.92915199578795e-05,
+      "loss": 1.2419,
+      "step": 93820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.928822929349436e-05,
+      "loss": 1.1586,
+      "step": 93830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9284938629109218e-05,
+      "loss": 1.3277,
+      "step": 93840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.928164796472408e-05,
+      "loss": 1.1732,
+      "step": 93850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.927835730033894e-05,
+      "loss": 1.4053,
+      "step": 93860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9275066635953802e-05,
+      "loss": 1.1885,
+      "step": 93870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9271775971568663e-05,
+      "loss": 1.2307,
+      "step": 93880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.926848530718352e-05,
+      "loss": 1.235,
+      "step": 93890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9265194642798382e-05,
+      "loss": 1.2235,
+      "step": 93900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9261903978413244e-05,
+      "loss": 1.4084,
+      "step": 93910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9258613314028105e-05,
+      "loss": 1.1806,
+      "step": 93920
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9255322649642966e-05,
+      "loss": 1.3034,
+      "step": 93930
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9252031985257824e-05,
+      "loss": 1.2913,
+      "step": 93940
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9248741320872685e-05,
+      "loss": 1.2704,
+      "step": 93950
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9245450656487547e-05,
+      "loss": 1.1322,
+      "step": 93960
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9242159992102408e-05,
+      "loss": 1.1359,
+      "step": 93970
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.923886932771727e-05,
+      "loss": 1.2094,
+      "step": 93980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9235578663332127e-05,
+      "loss": 1.2812,
+      "step": 93990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9232287998946988e-05,
+      "loss": 1.1331,
+      "step": 94000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.922899733456185e-05,
+      "loss": 1.3624,
+      "step": 94010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.922570667017671e-05,
+      "loss": 1.2674,
+      "step": 94020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9222416005791572e-05,
+      "loss": 1.1604,
+      "step": 94030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.921912534140643e-05,
+      "loss": 1.1625,
+      "step": 94040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.921583467702129e-05,
+      "loss": 1.1125,
+      "step": 94050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9212544012636153e-05,
+      "loss": 1.2329,
+      "step": 94060
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9209253348251014e-05,
+      "loss": 1.2287,
+      "step": 94070
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9205962683865875e-05,
+      "loss": 1.2023,
+      "step": 94080
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9202672019480733e-05,
+      "loss": 1.1509,
+      "step": 94090
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9199381355095594e-05,
+      "loss": 1.1076,
+      "step": 94100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9196090690710455e-05,
+      "loss": 1.0395,
+      "step": 94110
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9192800026325317e-05,
+      "loss": 1.1308,
+      "step": 94120
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9189509361940178e-05,
+      "loss": 1.2334,
+      "step": 94130
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9186218697555036e-05,
+      "loss": 1.1008,
+      "step": 94140
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9182928033169897e-05,
+      "loss": 1.1623,
+      "step": 94150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.917963736878476e-05,
+      "loss": 1.2568,
+      "step": 94160
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.917634670439962e-05,
+      "loss": 1.3169,
+      "step": 94170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.917305604001448e-05,
+      "loss": 1.3158,
+      "step": 94180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.916976537562934e-05,
+      "loss": 1.1076,
+      "step": 94190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.91664747112442e-05,
+      "loss": 1.2215,
+      "step": 94200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.916318404685906e-05,
+      "loss": 1.2722,
+      "step": 94210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9159893382473923e-05,
+      "loss": 1.1664,
+      "step": 94220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9156602718088784e-05,
+      "loss": 1.317,
+      "step": 94230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9153312053703642e-05,
+      "loss": 1.2098,
+      "step": 94240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9150021389318503e-05,
+      "loss": 1.1219,
+      "step": 94250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9146730724933364e-05,
+      "loss": 1.1014,
+      "step": 94260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9143440060548226e-05,
+      "loss": 1.2034,
+      "step": 94270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9140149396163087e-05,
+      "loss": 1.1752,
+      "step": 94280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9136858731777945e-05,
+      "loss": 1.1272,
+      "step": 94290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9133568067392806e-05,
+      "loss": 1.2407,
+      "step": 94300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9130277403007667e-05,
+      "loss": 1.0379,
+      "step": 94310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.912698673862253e-05,
+      "loss": 1.2073,
+      "step": 94320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.912369607423739e-05,
+      "loss": 1.5615,
+      "step": 94330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9120405409852248e-05,
+      "loss": 1.2645,
+      "step": 94340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.911711474546711e-05,
+      "loss": 1.0964,
+      "step": 94350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.911382408108197e-05,
+      "loss": 1.1854,
+      "step": 94360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.911053341669683e-05,
+      "loss": 1.2069,
+      "step": 94370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9107242752311693e-05,
+      "loss": 1.3379,
+      "step": 94380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.910395208792655e-05,
+      "loss": 1.1209,
+      "step": 94390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9100661423541412e-05,
+      "loss": 1.3314,
+      "step": 94400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9097370759156273e-05,
+      "loss": 1.3335,
+      "step": 94410
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9094080094771134e-05,
+      "loss": 1.1835,
+      "step": 94420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9090789430385996e-05,
+      "loss": 1.3243,
+      "step": 94430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9087498766000857e-05,
+      "loss": 1.2454,
+      "step": 94440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9084208101615715e-05,
+      "loss": 1.152,
+      "step": 94450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9080917437230576e-05,
+      "loss": 1.7783,
+      "step": 94460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9077626772845437e-05,
+      "loss": 1.0932,
+      "step": 94470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.90743361084603e-05,
+      "loss": 1.1538,
+      "step": 94480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.907104544407516e-05,
+      "loss": 1.0713,
+      "step": 94490
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.906775477969002e-05,
+      "loss": 1.1922,
+      "step": 94500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9064464115304882e-05,
+      "loss": 1.373,
+      "step": 94510
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9061173450919744e-05,
+      "loss": 1.1818,
+      "step": 94520
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.90578827865346e-05,
+      "loss": 1.2801,
+      "step": 94530
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9054592122149463e-05,
+      "loss": 1.0877,
+      "step": 94540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9051301457764324e-05,
+      "loss": 1.0599,
+      "step": 94550
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9048010793379185e-05,
+      "loss": 1.1527,
+      "step": 94560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9044720128994047e-05,
+      "loss": 1.1585,
+      "step": 94570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9041429464608908e-05,
+      "loss": 1.1649,
+      "step": 94580
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9038138800223766e-05,
+      "loss": 1.1829,
+      "step": 94590
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9034848135838627e-05,
+      "loss": 1.2703,
+      "step": 94600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.903155747145349e-05,
+      "loss": 1.2237,
+      "step": 94610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.902826680706835e-05,
+      "loss": 1.2823,
+      "step": 94620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.902497614268321e-05,
+      "loss": 1.1381,
+      "step": 94630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.902168547829807e-05,
+      "loss": 1.3456,
+      "step": 94640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.901839481391293e-05,
+      "loss": 1.0649,
+      "step": 94650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.901510414952779e-05,
+      "loss": 1.2409,
+      "step": 94660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9011813485142653e-05,
+      "loss": 1.1221,
+      "step": 94670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9008522820757514e-05,
+      "loss": 1.3297,
+      "step": 94680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9005232156372372e-05,
+      "loss": 1.3712,
+      "step": 94690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9001941491987233e-05,
+      "loss": 1.1819,
+      "step": 94700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8998650827602094e-05,
+      "loss": 1.1284,
+      "step": 94710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8995360163216956e-05,
+      "loss": 1.3376,
+      "step": 94720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8992069498831817e-05,
+      "loss": 1.2274,
+      "step": 94730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8988778834446675e-05,
+      "loss": 1.2402,
+      "step": 94740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8985488170061536e-05,
+      "loss": 1.3121,
+      "step": 94750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8982197505676397e-05,
+      "loss": 1.3221,
+      "step": 94760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.897890684129126e-05,
+      "loss": 1.1949,
+      "step": 94770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.897561617690612e-05,
+      "loss": 1.3133,
+      "step": 94780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8972325512520978e-05,
+      "loss": 1.12,
+      "step": 94790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.896903484813584e-05,
+      "loss": 1.1592,
+      "step": 94800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.89657441837507e-05,
+      "loss": 1.2723,
+      "step": 94810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.896245351936556e-05,
+      "loss": 1.2633,
+      "step": 94820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8959162854980423e-05,
+      "loss": 1.3896,
+      "step": 94830
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.895587219059528e-05,
+      "loss": 1.4241,
+      "step": 94840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8952581526210142e-05,
+      "loss": 1.4235,
+      "step": 94850
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8949290861825003e-05,
+      "loss": 1.1301,
+      "step": 94860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8946000197439864e-05,
+      "loss": 1.0536,
+      "step": 94870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8942709533054726e-05,
+      "loss": 1.1325,
+      "step": 94880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8939418868669583e-05,
+      "loss": 1.1996,
+      "step": 94890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8936128204284445e-05,
+      "loss": 1.2488,
+      "step": 94900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8932837539899306e-05,
+      "loss": 1.3652,
+      "step": 94910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8929546875514167e-05,
+      "loss": 1.136,
+      "step": 94920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.892625621112903e-05,
+      "loss": 1.2888,
+      "step": 94930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.892296554674389e-05,
+      "loss": 1.2816,
+      "step": 94940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8919674882358748e-05,
+      "loss": 1.161,
+      "step": 94950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.891638421797361e-05,
+      "loss": 1.1406,
+      "step": 94960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.891309355358847e-05,
+      "loss": 1.2995,
+      "step": 94970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.890980288920333e-05,
+      "loss": 1.3197,
+      "step": 94980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8906512224818193e-05,
+      "loss": 1.1977,
+      "step": 94990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.890322156043305e-05,
+      "loss": 1.1362,
+      "step": 95000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8899930896047912e-05,
+      "loss": 1.3172,
+      "step": 95010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8896640231662773e-05,
+      "loss": 1.3049,
+      "step": 95020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8893349567277634e-05,
+      "loss": 1.2536,
+      "step": 95030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8890058902892496e-05,
+      "loss": 1.1828,
+      "step": 95040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8886768238507354e-05,
+      "loss": 1.249,
+      "step": 95050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8883477574122215e-05,
+      "loss": 1.1364,
+      "step": 95060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8880186909737076e-05,
+      "loss": 1.1906,
+      "step": 95070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8876896245351937e-05,
+      "loss": 1.4909,
+      "step": 95080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.88736055809668e-05,
+      "loss": 1.2471,
+      "step": 95090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8870314916581657e-05,
+      "loss": 1.1151,
+      "step": 95100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8867024252196518e-05,
+      "loss": 1.1006,
+      "step": 95110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.886373358781138e-05,
+      "loss": 1.4479,
+      "step": 95120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.886044292342624e-05,
+      "loss": 1.3112,
+      "step": 95130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.88571522590411e-05,
+      "loss": 1.2664,
+      "step": 95140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.885386159465596e-05,
+      "loss": 1.1908,
+      "step": 95150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.885057093027082e-05,
+      "loss": 1.2749,
+      "step": 95160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8847280265885682e-05,
+      "loss": 1.1106,
+      "step": 95170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8843989601500543e-05,
+      "loss": 1.2027,
+      "step": 95180
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8840698937115405e-05,
+      "loss": 1.3723,
+      "step": 95190
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8837408272730266e-05,
+      "loss": 1.3234,
+      "step": 95200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8834117608345124e-05,
+      "loss": 1.2382,
+      "step": 95210
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8830826943959985e-05,
+      "loss": 1.2148,
+      "step": 95220
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8827536279574846e-05,
+      "loss": 1.2712,
+      "step": 95230
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8824245615189708e-05,
+      "loss": 1.2461,
+      "step": 95240
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.882095495080457e-05,
+      "loss": 1.187,
+      "step": 95250
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.881766428641943e-05,
+      "loss": 1.4545,
+      "step": 95260
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.881437362203429e-05,
+      "loss": 1.2323,
+      "step": 95270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8811082957649153e-05,
+      "loss": 1.2668,
+      "step": 95280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.880779229326401e-05,
+      "loss": 1.1741,
+      "step": 95290
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8804501628878872e-05,
+      "loss": 1.3029,
+      "step": 95300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8801210964493733e-05,
+      "loss": 1.1263,
+      "step": 95310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8797920300108594e-05,
+      "loss": 1.1068,
+      "step": 95320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8794629635723456e-05,
+      "loss": 1.3632,
+      "step": 95330
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8791338971338317e-05,
+      "loss": 1.1049,
+      "step": 95340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8788048306953175e-05,
+      "loss": 1.2953,
+      "step": 95350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8784757642568036e-05,
+      "loss": 1.0703,
+      "step": 95360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8781466978182897e-05,
+      "loss": 1.015,
+      "step": 95370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.877817631379776e-05,
+      "loss": 1.1307,
+      "step": 95380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.877488564941262e-05,
+      "loss": 1.3595,
+      "step": 95390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8771594985027478e-05,
+      "loss": 1.2425,
+      "step": 95400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.876830432064234e-05,
+      "loss": 1.4585,
+      "step": 95410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.87650136562572e-05,
+      "loss": 1.1874,
+      "step": 95420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.876172299187206e-05,
+      "loss": 1.2589,
+      "step": 95430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8758432327486923e-05,
+      "loss": 1.3103,
+      "step": 95440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.875514166310178e-05,
+      "loss": 1.1464,
+      "step": 95450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8751850998716642e-05,
+      "loss": 1.1992,
+      "step": 95460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8748560334331503e-05,
+      "loss": 1.1933,
+      "step": 95470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8745269669946364e-05,
+      "loss": 1.3908,
+      "step": 95480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8741979005561226e-05,
+      "loss": 1.3745,
+      "step": 95490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8738688341176084e-05,
+      "loss": 1.0844,
+      "step": 95500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8735397676790945e-05,
+      "loss": 1.2547,
+      "step": 95510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8732107012405806e-05,
+      "loss": 1.3422,
+      "step": 95520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8728816348020667e-05,
+      "loss": 1.2891,
+      "step": 95530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.872552568363553e-05,
+      "loss": 1.363,
+      "step": 95540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8722235019250386e-05,
+      "loss": 1.3695,
+      "step": 95550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8718944354865248e-05,
+      "loss": 1.3027,
+      "step": 95560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.871565369048011e-05,
+      "loss": 1.2208,
+      "step": 95570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.871236302609497e-05,
+      "loss": 1.1873,
+      "step": 95580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.870907236170983e-05,
+      "loss": 1.1201,
+      "step": 95590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.870578169732469e-05,
+      "loss": 1.2521,
+      "step": 95600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.870249103293955e-05,
+      "loss": 1.188,
+      "step": 95610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8699200368554412e-05,
+      "loss": 1.2699,
+      "step": 95620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8695909704169273e-05,
+      "loss": 1.333,
+      "step": 95630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8692619039784135e-05,
+      "loss": 1.2478,
+      "step": 95640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8689328375398992e-05,
+      "loss": 1.2391,
+      "step": 95650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8686037711013854e-05,
+      "loss": 1.1677,
+      "step": 95660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8682747046628715e-05,
+      "loss": 1.1127,
+      "step": 95670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8679456382243576e-05,
+      "loss": 1.3283,
+      "step": 95680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8676165717858437e-05,
+      "loss": 1.2896,
+      "step": 95690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8672875053473295e-05,
+      "loss": 1.3054,
+      "step": 95700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8669584389088157e-05,
+      "loss": 1.2439,
+      "step": 95710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8666293724703018e-05,
+      "loss": 1.1583,
+      "step": 95720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.866300306031788e-05,
+      "loss": 1.2506,
+      "step": 95730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.865971239593274e-05,
+      "loss": 1.2667,
+      "step": 95740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8656421731547598e-05,
+      "loss": 1.2663,
+      "step": 95750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.865313106716246e-05,
+      "loss": 1.6353,
+      "step": 95760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.864984040277732e-05,
+      "loss": 1.271,
+      "step": 95770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8646549738392182e-05,
+      "loss": 1.1342,
+      "step": 95780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8643259074007043e-05,
+      "loss": 1.2175,
+      "step": 95790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.86399684096219e-05,
+      "loss": 1.2423,
+      "step": 95800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8636677745236763e-05,
+      "loss": 1.2548,
+      "step": 95810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8633387080851624e-05,
+      "loss": 1.1674,
+      "step": 95820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8630096416466485e-05,
+      "loss": 1.2313,
+      "step": 95830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8626805752081346e-05,
+      "loss": 1.1637,
+      "step": 95840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8623515087696204e-05,
+      "loss": 1.2637,
+      "step": 95850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8620224423311065e-05,
+      "loss": 1.3268,
+      "step": 95860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8616933758925927e-05,
+      "loss": 1.2233,
+      "step": 95870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8613643094540788e-05,
+      "loss": 1.3091,
+      "step": 95880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.861035243015565e-05,
+      "loss": 1.3282,
+      "step": 95890
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8607061765770507e-05,
+      "loss": 1.2992,
+      "step": 95900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.860377110138537e-05,
+      "loss": 1.3777,
+      "step": 95910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.860048043700023e-05,
+      "loss": 1.2132,
+      "step": 95920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.859718977261509e-05,
+      "loss": 1.3142,
+      "step": 95930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8593899108229952e-05,
+      "loss": 1.2441,
+      "step": 95940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8590608443844813e-05,
+      "loss": 1.2664,
+      "step": 95950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8587317779459675e-05,
+      "loss": 1.2356,
+      "step": 95960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8584027115074533e-05,
+      "loss": 1.1963,
+      "step": 95970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8580736450689394e-05,
+      "loss": 1.395,
+      "step": 95980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8577445786304255e-05,
+      "loss": 1.1635,
+      "step": 95990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8574155121919116e-05,
+      "loss": 1.2577,
+      "step": 96000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8570864457533978e-05,
+      "loss": 1.3555,
+      "step": 96010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.856757379314884e-05,
+      "loss": 1.0895,
+      "step": 96020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.85642831287637e-05,
+      "loss": 1.3013,
+      "step": 96030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.856099246437856e-05,
+      "loss": 1.1454,
+      "step": 96040
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.855770179999342e-05,
+      "loss": 1.3299,
+      "step": 96050
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.855441113560828e-05,
+      "loss": 1.1459,
+      "step": 96060
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8551120471223142e-05,
+      "loss": 1.2149,
+      "step": 96070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8547829806838003e-05,
+      "loss": 1.372,
+      "step": 96080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8544539142452864e-05,
+      "loss": 1.3108,
+      "step": 96090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8541248478067722e-05,
+      "loss": 1.2851,
+      "step": 96100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8537957813682584e-05,
+      "loss": 1.2288,
+      "step": 96110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8534667149297445e-05,
+      "loss": 1.1436,
+      "step": 96120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8531376484912306e-05,
+      "loss": 1.1093,
+      "step": 96130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8528085820527167e-05,
+      "loss": 1.2082,
+      "step": 96140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8524795156142025e-05,
+      "loss": 1.162,
+      "step": 96150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8521504491756887e-05,
+      "loss": 1.2848,
+      "step": 96160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8518213827371748e-05,
+      "loss": 1.2042,
+      "step": 96170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.851492316298661e-05,
+      "loss": 1.1379,
+      "step": 96180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.851163249860147e-05,
+      "loss": 1.0916,
+      "step": 96190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8508341834216328e-05,
+      "loss": 1.0789,
+      "step": 96200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.850505116983119e-05,
+      "loss": 1.2491,
+      "step": 96210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.850176050544605e-05,
+      "loss": 1.1866,
+      "step": 96220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8498469841060912e-05,
+      "loss": 1.1838,
+      "step": 96230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8495179176675773e-05,
+      "loss": 1.2344,
+      "step": 96240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.849188851229063e-05,
+      "loss": 1.3657,
+      "step": 96250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8488597847905492e-05,
+      "loss": 1.1887,
+      "step": 96260
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8485307183520354e-05,
+      "loss": 1.2772,
+      "step": 96270
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8482016519135215e-05,
+      "loss": 1.2092,
+      "step": 96280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8478725854750076e-05,
+      "loss": 1.0637,
+      "step": 96290
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8475435190364934e-05,
+      "loss": 1.0802,
+      "step": 96300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8472144525979795e-05,
+      "loss": 1.1716,
+      "step": 96310
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8468853861594657e-05,
+      "loss": 1.2896,
+      "step": 96320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8465563197209518e-05,
+      "loss": 1.1482,
+      "step": 96330
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.846227253282438e-05,
+      "loss": 1.3488,
+      "step": 96340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8458981868439237e-05,
+      "loss": 1.2768,
+      "step": 96350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.84556912040541e-05,
+      "loss": 1.0982,
+      "step": 96360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.845240053966896e-05,
+      "loss": 1.0846,
+      "step": 96370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.844910987528382e-05,
+      "loss": 1.1754,
+      "step": 96380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8445819210898682e-05,
+      "loss": 1.0696,
+      "step": 96390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8442528546513543e-05,
+      "loss": 1.246,
+      "step": 96400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.84392378821284e-05,
+      "loss": 1.0233,
+      "step": 96410
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8435947217743263e-05,
+      "loss": 1.1067,
+      "step": 96420
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8432656553358124e-05,
+      "loss": 1.1813,
+      "step": 96430
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8429365888972985e-05,
+      "loss": 1.2722,
+      "step": 96440
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8426075224587846e-05,
+      "loss": 1.3368,
+      "step": 96450
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8422784560202704e-05,
+      "loss": 1.2145,
+      "step": 96460
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8419493895817565e-05,
+      "loss": 1.353,
+      "step": 96470
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8416203231432427e-05,
+      "loss": 1.1658,
+      "step": 96480
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8412912567047288e-05,
+      "loss": 1.0179,
+      "step": 96490
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.840962190266215e-05,
+      "loss": 1.1533,
+      "step": 96500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8406331238277007e-05,
+      "loss": 1.2844,
+      "step": 96510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.840304057389187e-05,
+      "loss": 1.0568,
+      "step": 96520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.839974990950673e-05,
+      "loss": 1.1613,
+      "step": 96530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.839645924512159e-05,
+      "loss": 1.2048,
+      "step": 96540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8393168580736452e-05,
+      "loss": 1.2341,
+      "step": 96550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.838987791635131e-05,
+      "loss": 1.2105,
+      "step": 96560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.838658725196617e-05,
+      "loss": 1.2166,
+      "step": 96570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8383296587581033e-05,
+      "loss": 1.2013,
+      "step": 96580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8380005923195894e-05,
+      "loss": 1.2146,
+      "step": 96590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8376715258810755e-05,
+      "loss": 1.1539,
+      "step": 96600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8373424594425613e-05,
+      "loss": 1.0854,
+      "step": 96610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8370133930040474e-05,
+      "loss": 1.3026,
+      "step": 96620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8366843265655336e-05,
+      "loss": 1.1833,
+      "step": 96630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8363552601270197e-05,
+      "loss": 1.3184,
+      "step": 96640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8360261936885058e-05,
+      "loss": 1.0962,
+      "step": 96650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8356971272499916e-05,
+      "loss": 1.1303,
+      "step": 96660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8353680608114777e-05,
+      "loss": 1.3516,
+      "step": 96670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.835038994372964e-05,
+      "loss": 1.0489,
+      "step": 96680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.83470992793445e-05,
+      "loss": 1.1192,
+      "step": 96690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.834380861495936e-05,
+      "loss": 1.0901,
+      "step": 96700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8340517950574222e-05,
+      "loss": 1.4186,
+      "step": 96710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8337227286189084e-05,
+      "loss": 1.0639,
+      "step": 96720
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8333936621803945e-05,
+      "loss": 1.1962,
+      "step": 96730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8330645957418803e-05,
+      "loss": 1.2918,
+      "step": 96740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8327355293033664e-05,
+      "loss": 1.1323,
+      "step": 96750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8324064628648525e-05,
+      "loss": 1.2271,
+      "step": 96760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8320773964263387e-05,
+      "loss": 1.1792,
+      "step": 96770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8317483299878248e-05,
+      "loss": 1.3279,
+      "step": 96780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.831419263549311e-05,
+      "loss": 1.1523,
+      "step": 96790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.831090197110797e-05,
+      "loss": 1.3404,
+      "step": 96800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8307611306722828e-05,
+      "loss": 1.1904,
+      "step": 96810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.830432064233769e-05,
+      "loss": 1.3178,
+      "step": 96820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.830102997795255e-05,
+      "loss": 1.1702,
+      "step": 96830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8297739313567412e-05,
+      "loss": 1.5016,
+      "step": 96840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8294448649182273e-05,
+      "loss": 1.1403,
+      "step": 96850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.829115798479713e-05,
+      "loss": 1.1193,
+      "step": 96860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8287867320411992e-05,
+      "loss": 1.2266,
+      "step": 96870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8284576656026854e-05,
+      "loss": 1.1498,
+      "step": 96880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8281285991641715e-05,
+      "loss": 1.2109,
+      "step": 96890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8277995327256576e-05,
+      "loss": 1.183,
+      "step": 96900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8274704662871434e-05,
+      "loss": 1.2612,
+      "step": 96910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8271413998486295e-05,
+      "loss": 1.3026,
+      "step": 96920
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8268123334101157e-05,
+      "loss": 1.1913,
+      "step": 96930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8264832669716018e-05,
+      "loss": 1.1739,
+      "step": 96940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.826154200533088e-05,
+      "loss": 1.2977,
+      "step": 96950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8258251340945737e-05,
+      "loss": 1.1675,
+      "step": 96960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.82549606765606e-05,
+      "loss": 1.2647,
+      "step": 96970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.825167001217546e-05,
+      "loss": 1.3177,
+      "step": 96980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.824837934779032e-05,
+      "loss": 1.2739,
+      "step": 96990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8245088683405182e-05,
+      "loss": 1.154,
+      "step": 97000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.824179801902004e-05,
+      "loss": 1.2523,
+      "step": 97010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.82385073546349e-05,
+      "loss": 1.3062,
+      "step": 97020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8235216690249763e-05,
+      "loss": 1.2067,
+      "step": 97030
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8231926025864624e-05,
+      "loss": 1.2408,
+      "step": 97040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8228635361479485e-05,
+      "loss": 1.189,
+      "step": 97050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8225344697094343e-05,
+      "loss": 1.3152,
+      "step": 97060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8222054032709204e-05,
+      "loss": 1.152,
+      "step": 97070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8218763368324066e-05,
+      "loss": 1.1923,
+      "step": 97080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8215472703938927e-05,
+      "loss": 1.2007,
+      "step": 97090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8212182039553788e-05,
+      "loss": 1.1201,
+      "step": 97100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8208891375168646e-05,
+      "loss": 1.3222,
+      "step": 97110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8205600710783507e-05,
+      "loss": 1.0776,
+      "step": 97120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.820231004639837e-05,
+      "loss": 1.174,
+      "step": 97130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.819901938201323e-05,
+      "loss": 1.1016,
+      "step": 97140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.819572871762809e-05,
+      "loss": 1.2862,
+      "step": 97150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.819243805324295e-05,
+      "loss": 1.26,
+      "step": 97160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.818914738885781e-05,
+      "loss": 1.5339,
+      "step": 97170
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.818585672447267e-05,
+      "loss": 1.2624,
+      "step": 97180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8182566060087533e-05,
+      "loss": 1.1841,
+      "step": 97190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8179275395702394e-05,
+      "loss": 1.196,
+      "step": 97200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8175984731317252e-05,
+      "loss": 1.3517,
+      "step": 97210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8172694066932113e-05,
+      "loss": 1.2879,
+      "step": 97220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8169403402546974e-05,
+      "loss": 1.2059,
+      "step": 97230
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8166112738161836e-05,
+      "loss": 1.3493,
+      "step": 97240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8162822073776697e-05,
+      "loss": 1.2154,
+      "step": 97250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8159531409391555e-05,
+      "loss": 1.1942,
+      "step": 97260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8156240745006416e-05,
+      "loss": 1.0889,
+      "step": 97270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8152950080621277e-05,
+      "loss": 1.2032,
+      "step": 97280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.814965941623614e-05,
+      "loss": 0.9812,
+      "step": 97290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8146368751851e-05,
+      "loss": 1.1512,
+      "step": 97300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8143078087465858e-05,
+      "loss": 1.2702,
+      "step": 97310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.813978742308072e-05,
+      "loss": 1.3079,
+      "step": 97320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.813649675869558e-05,
+      "loss": 1.1277,
+      "step": 97330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.813320609431044e-05,
+      "loss": 1.3637,
+      "step": 97340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8129915429925303e-05,
+      "loss": 1.3938,
+      "step": 97350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.812662476554016e-05,
+      "loss": 1.2894,
+      "step": 97360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8123334101155022e-05,
+      "loss": 1.2306,
+      "step": 97370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8120043436769883e-05,
+      "loss": 1.1128,
+      "step": 97380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8116752772384745e-05,
+      "loss": 1.3237,
+      "step": 97390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8113462107999606e-05,
+      "loss": 1.2958,
+      "step": 97400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8110171443614467e-05,
+      "loss": 1.1348,
+      "step": 97410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8106880779229325e-05,
+      "loss": 1.3161,
+      "step": 97420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8103590114844186e-05,
+      "loss": 1.379,
+      "step": 97430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8100299450459047e-05,
+      "loss": 1.1967,
+      "step": 97440
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.809700878607391e-05,
+      "loss": 1.1731,
+      "step": 97450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.809371812168877e-05,
+      "loss": 1.2521,
+      "step": 97460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.809042745730363e-05,
+      "loss": 1.2431,
+      "step": 97470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8087136792918493e-05,
+      "loss": 1.1891,
+      "step": 97480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8083846128533354e-05,
+      "loss": 1.0818,
+      "step": 97490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.808055546414821e-05,
+      "loss": 1.176,
+      "step": 97500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8077264799763073e-05,
+      "loss": 1.1771,
+      "step": 97510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8073974135377934e-05,
+      "loss": 1.2975,
+      "step": 97520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8070683470992795e-05,
+      "loss": 1.1688,
+      "step": 97530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8067392806607657e-05,
+      "loss": 1.2158,
+      "step": 97540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8064102142222518e-05,
+      "loss": 1.3415,
+      "step": 97550
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8060811477837376e-05,
+      "loss": 1.2826,
+      "step": 97560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8057520813452237e-05,
+      "loss": 1.2303,
+      "step": 97570
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.80542301490671e-05,
+      "loss": 1.166,
+      "step": 97580
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.805093948468196e-05,
+      "loss": 1.1598,
+      "step": 97590
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.804764882029682e-05,
+      "loss": 1.1466,
+      "step": 97600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.804435815591168e-05,
+      "loss": 1.1727,
+      "step": 97610
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.804106749152654e-05,
+      "loss": 1.2104,
+      "step": 97620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.80377768271414e-05,
+      "loss": 1.1798,
+      "step": 97630
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8034486162756263e-05,
+      "loss": 1.2002,
+      "step": 97640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8031195498371124e-05,
+      "loss": 1.2635,
+      "step": 97650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8027904833985982e-05,
+      "loss": 1.21,
+      "step": 97660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8024614169600843e-05,
+      "loss": 1.2526,
+      "step": 97670
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8021323505215704e-05,
+      "loss": 1.2882,
+      "step": 97680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8018032840830566e-05,
+      "loss": 1.2249,
+      "step": 97690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8014742176445427e-05,
+      "loss": 1.2684,
+      "step": 97700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8011451512060285e-05,
+      "loss": 1.251,
+      "step": 97710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8008160847675146e-05,
+      "loss": 1.1639,
+      "step": 97720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8004870183290007e-05,
+      "loss": 1.3404,
+      "step": 97730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.800157951890487e-05,
+      "loss": 1.4196,
+      "step": 97740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.799828885451973e-05,
+      "loss": 1.2797,
+      "step": 97750
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7994998190134588e-05,
+      "loss": 1.2684,
+      "step": 97760
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.799170752574945e-05,
+      "loss": 1.2194,
+      "step": 97770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.798841686136431e-05,
+      "loss": 1.1756,
+      "step": 97780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.798512619697917e-05,
+      "loss": 1.2128,
+      "step": 97790
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7981835532594033e-05,
+      "loss": 1.3651,
+      "step": 97800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7978544868208894e-05,
+      "loss": 1.225,
+      "step": 97810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7975254203823752e-05,
+      "loss": 1.3137,
+      "step": 97820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7971963539438613e-05,
+      "loss": 1.159,
+      "step": 97830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7968672875053474e-05,
+      "loss": 1.7538,
+      "step": 97840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7965382210668336e-05,
+      "loss": 1.2463,
+      "step": 97850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7962091546283197e-05,
+      "loss": 1.3567,
+      "step": 97860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7958800881898055e-05,
+      "loss": 1.2491,
+      "step": 97870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7955510217512916e-05,
+      "loss": 1.1702,
+      "step": 97880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7952219553127777e-05,
+      "loss": 1.2,
+      "step": 97890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.794892888874264e-05,
+      "loss": 1.2813,
+      "step": 97900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.79456382243575e-05,
+      "loss": 1.2724,
+      "step": 97910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7942347559972358e-05,
+      "loss": 1.2174,
+      "step": 97920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.793905689558722e-05,
+      "loss": 1.2895,
+      "step": 97930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.793576623120208e-05,
+      "loss": 1.1667,
+      "step": 97940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.793247556681694e-05,
+      "loss": 1.2455,
+      "step": 97950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7929184902431803e-05,
+      "loss": 1.2732,
+      "step": 97960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.792589423804666e-05,
+      "loss": 1.3035,
+      "step": 97970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7922603573661522e-05,
+      "loss": 1.2958,
+      "step": 97980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7919312909276383e-05,
+      "loss": 1.1806,
+      "step": 97990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7916022244891245e-05,
+      "loss": 1.1155,
+      "step": 98000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7912731580506106e-05,
+      "loss": 1.2057,
+      "step": 98010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7909440916120964e-05,
+      "loss": 1.1727,
+      "step": 98020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7906150251735825e-05,
+      "loss": 1.2626,
+      "step": 98030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7902859587350686e-05,
+      "loss": 1.187,
+      "step": 98040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7899568922965548e-05,
+      "loss": 1.2904,
+      "step": 98050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.789627825858041e-05,
+      "loss": 1.1751,
+      "step": 98060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7892987594195267e-05,
+      "loss": 1.1854,
+      "step": 98070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7889696929810128e-05,
+      "loss": 1.2449,
+      "step": 98080
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.788640626542499e-05,
+      "loss": 1.2277,
+      "step": 98090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.788311560103985e-05,
+      "loss": 1.5043,
+      "step": 98100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7879824936654712e-05,
+      "loss": 1.1403,
+      "step": 98110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.787653427226957e-05,
+      "loss": 1.2584,
+      "step": 98120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.787324360788443e-05,
+      "loss": 1.2894,
+      "step": 98130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7869952943499292e-05,
+      "loss": 1.2039,
+      "step": 98140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7866662279114153e-05,
+      "loss": 1.1631,
+      "step": 98150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7863371614729015e-05,
+      "loss": 1.1297,
+      "step": 98160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7860080950343876e-05,
+      "loss": 1.2967,
+      "step": 98170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7856790285958734e-05,
+      "loss": 1.3348,
+      "step": 98180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7853499621573595e-05,
+      "loss": 1.162,
+      "step": 98190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7850208957188456e-05,
+      "loss": 1.0514,
+      "step": 98200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7846918292803318e-05,
+      "loss": 1.277,
+      "step": 98210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.784362762841818e-05,
+      "loss": 1.128,
+      "step": 98220
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.784033696403304e-05,
+      "loss": 1.2507,
+      "step": 98230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.78370462996479e-05,
+      "loss": 1.0734,
+      "step": 98240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7833755635262763e-05,
+      "loss": 1.2052,
+      "step": 98250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.783046497087762e-05,
+      "loss": 1.286,
+      "step": 98260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7827174306492482e-05,
+      "loss": 1.0647,
+      "step": 98270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7823883642107343e-05,
+      "loss": 1.2889,
+      "step": 98280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7820592977722204e-05,
+      "loss": 1.2157,
+      "step": 98290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7817302313337066e-05,
+      "loss": 1.2198,
+      "step": 98300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7814011648951927e-05,
+      "loss": 1.1518,
+      "step": 98310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7810720984566785e-05,
+      "loss": 1.2459,
+      "step": 98320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7807430320181646e-05,
+      "loss": 1.2817,
+      "step": 98330
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7804139655796507e-05,
+      "loss": 1.2063,
+      "step": 98340
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.780084899141137e-05,
+      "loss": 1.2945,
+      "step": 98350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.779755832702623e-05,
+      "loss": 1.4996,
+      "step": 98360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7794267662641088e-05,
+      "loss": 1.112,
+      "step": 98370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.779097699825595e-05,
+      "loss": 1.3652,
+      "step": 98380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.778768633387081e-05,
+      "loss": 1.3952,
+      "step": 98390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.778439566948567e-05,
+      "loss": 1.3735,
+      "step": 98400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7781105005100533e-05,
+      "loss": 1.4178,
+      "step": 98410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.777781434071539e-05,
+      "loss": 1.3281,
+      "step": 98420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7774523676330252e-05,
+      "loss": 1.2548,
+      "step": 98430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7771233011945113e-05,
+      "loss": 1.2133,
+      "step": 98440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7767942347559974e-05,
+      "loss": 1.1487,
+      "step": 98450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7764651683174836e-05,
+      "loss": 1.118,
+      "step": 98460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7761361018789694e-05,
+      "loss": 1.4422,
+      "step": 98470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7758070354404555e-05,
+      "loss": 1.3762,
+      "step": 98480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7754779690019416e-05,
+      "loss": 1.1938,
+      "step": 98490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7751489025634277e-05,
+      "loss": 1.1622,
+      "step": 98500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.774819836124914e-05,
+      "loss": 1.3365,
+      "step": 98510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7744907696863997e-05,
+      "loss": 1.1846,
+      "step": 98520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7741617032478858e-05,
+      "loss": 1.3458,
+      "step": 98530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.773832636809372e-05,
+      "loss": 1.1784,
+      "step": 98540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.773503570370858e-05,
+      "loss": 1.1923,
+      "step": 98550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.773174503932344e-05,
+      "loss": 1.3835,
+      "step": 98560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.77284543749383e-05,
+      "loss": 1.348,
+      "step": 98570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.772516371055316e-05,
+      "loss": 1.2085,
+      "step": 98580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7721873046168022e-05,
+      "loss": 1.3133,
+      "step": 98590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7718582381782883e-05,
+      "loss": 1.3084,
+      "step": 98600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7715291717397745e-05,
+      "loss": 1.188,
+      "step": 98610
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7712001053012602e-05,
+      "loss": 1.1517,
+      "step": 98620
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7708710388627464e-05,
+      "loss": 1.2832,
+      "step": 98630
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7705419724242325e-05,
+      "loss": 1.0787,
+      "step": 98640
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7702129059857186e-05,
+      "loss": 1.1935,
+      "step": 98650
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7698838395472048e-05,
+      "loss": 1.1586,
+      "step": 98660
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7695547731086905e-05,
+      "loss": 1.281,
+      "step": 98670
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7692257066701767e-05,
+      "loss": 1.196,
+      "step": 98680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7688966402316628e-05,
+      "loss": 1.0661,
+      "step": 98690
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.768567573793149e-05,
+      "loss": 1.1725,
+      "step": 98700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.768238507354635e-05,
+      "loss": 1.4364,
+      "step": 98710
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.767909440916121e-05,
+      "loss": 1.3365,
+      "step": 98720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.767580374477607e-05,
+      "loss": 1.1641,
+      "step": 98730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.767251308039093e-05,
+      "loss": 1.2455,
+      "step": 98740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7669222416005792e-05,
+      "loss": 1.2468,
+      "step": 98750
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7665931751620653e-05,
+      "loss": 1.195,
+      "step": 98760
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.766264108723551e-05,
+      "loss": 1.3065,
+      "step": 98770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7659350422850373e-05,
+      "loss": 1.1591,
+      "step": 98780
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7656059758465234e-05,
+      "loss": 1.1606,
+      "step": 98790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7652769094080095e-05,
+      "loss": 1.197,
+      "step": 98800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7649478429694956e-05,
+      "loss": 1.2272,
+      "step": 98810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7646187765309814e-05,
+      "loss": 1.241,
+      "step": 98820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7642897100924676e-05,
+      "loss": 1.2789,
+      "step": 98830
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7639606436539537e-05,
+      "loss": 1.1445,
+      "step": 98840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7636315772154398e-05,
+      "loss": 1.1262,
+      "step": 98850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.763302510776926e-05,
+      "loss": 1.2607,
+      "step": 98860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7629734443384117e-05,
+      "loss": 1.3447,
+      "step": 98870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.762644377899898e-05,
+      "loss": 1.3228,
+      "step": 98880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.762315311461384e-05,
+      "loss": 1.3557,
+      "step": 98890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.76198624502287e-05,
+      "loss": 1.2028,
+      "step": 98900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7616571785843562e-05,
+      "loss": 1.1463,
+      "step": 98910
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7613281121458424e-05,
+      "loss": 1.2857,
+      "step": 98920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7609990457073285e-05,
+      "loss": 1.2623,
+      "step": 98930
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7606699792688146e-05,
+      "loss": 1.3217,
+      "step": 98940
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7603409128303004e-05,
+      "loss": 1.1724,
+      "step": 98950
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7600118463917865e-05,
+      "loss": 1.2822,
+      "step": 98960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7596827799532727e-05,
+      "loss": 1.3403,
+      "step": 98970
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7593537135147588e-05,
+      "loss": 1.3582,
+      "step": 98980
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.759024647076245e-05,
+      "loss": 1.262,
+      "step": 98990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.758695580637731e-05,
+      "loss": 1.3934,
+      "step": 99000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.758366514199217e-05,
+      "loss": 1.2711,
+      "step": 99010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.758037447760703e-05,
+      "loss": 1.2406,
+      "step": 99020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.757708381322189e-05,
+      "loss": 1.3032,
+      "step": 99030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7573793148836752e-05,
+      "loss": 1.121,
+      "step": 99040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7570502484451613e-05,
+      "loss": 1.1274,
+      "step": 99050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7567211820066475e-05,
+      "loss": 1.1948,
+      "step": 99060
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7563921155681332e-05,
+      "loss": 1.2553,
+      "step": 99070
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7560630491296194e-05,
+      "loss": 1.1429,
+      "step": 99080
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7557339826911055e-05,
+      "loss": 1.2717,
+      "step": 99090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7554049162525916e-05,
+      "loss": 1.371,
+      "step": 99100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7550758498140777e-05,
+      "loss": 1.1019,
+      "step": 99110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7547467833755635e-05,
+      "loss": 1.2319,
+      "step": 99120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7544177169370497e-05,
+      "loss": 1.2099,
+      "step": 99130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7540886504985358e-05,
+      "loss": 1.1851,
+      "step": 99140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.753759584060022e-05,
+      "loss": 1.235,
+      "step": 99150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.753430517621508e-05,
+      "loss": 1.1839,
+      "step": 99160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7531014511829938e-05,
+      "loss": 1.365,
+      "step": 99170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.75277238474448e-05,
+      "loss": 1.3308,
+      "step": 99180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.752443318305966e-05,
+      "loss": 1.2767,
+      "step": 99190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7521142518674522e-05,
+      "loss": 1.1835,
+      "step": 99200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7517851854289383e-05,
+      "loss": 1.4123,
+      "step": 99210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7514561189904245e-05,
+      "loss": 1.058,
+      "step": 99220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7511270525519103e-05,
+      "loss": 1.1016,
+      "step": 99230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7507979861133964e-05,
+      "loss": 1.1811,
+      "step": 99240
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7504689196748825e-05,
+      "loss": 1.2189,
+      "step": 99250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7501398532363686e-05,
+      "loss": 1.0648,
+      "step": 99260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7498107867978548e-05,
+      "loss": 1.2447,
+      "step": 99270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7494817203593405e-05,
+      "loss": 1.2269,
+      "step": 99280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7491526539208267e-05,
+      "loss": 1.2757,
+      "step": 99290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7488235874823128e-05,
+      "loss": 1.4196,
+      "step": 99300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.748494521043799e-05,
+      "loss": 1.3087,
+      "step": 99310
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.748165454605285e-05,
+      "loss": 1.1878,
+      "step": 99320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.747836388166771e-05,
+      "loss": 1.2976,
+      "step": 99330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.747507321728257e-05,
+      "loss": 1.212,
+      "step": 99340
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.747178255289743e-05,
+      "loss": 1.2374,
+      "step": 99350
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7468491888512292e-05,
+      "loss": 1.3235,
+      "step": 99360
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7465201224127153e-05,
+      "loss": 1.0808,
+      "step": 99370
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.746191055974201e-05,
+      "loss": 1.2676,
+      "step": 99380
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7458619895356873e-05,
+      "loss": 1.2412,
+      "step": 99390
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7455329230971734e-05,
+      "loss": 1.2287,
+      "step": 99400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7452038566586595e-05,
+      "loss": 1.3157,
+      "step": 99410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7448747902201456e-05,
+      "loss": 1.2582,
+      "step": 99420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7445457237816314e-05,
+      "loss": 1.2246,
+      "step": 99430
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7442166573431176e-05,
+      "loss": 1.1849,
+      "step": 99440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7438875909046037e-05,
+      "loss": 1.4041,
+      "step": 99450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7435585244660898e-05,
+      "loss": 1.1917,
+      "step": 99460
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.743229458027576e-05,
+      "loss": 1.0926,
+      "step": 99470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7429003915890617e-05,
+      "loss": 1.342,
+      "step": 99480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.742571325150548e-05,
+      "loss": 1.1733,
+      "step": 99490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.742242258712034e-05,
+      "loss": 1.3021,
+      "step": 99500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.74191319227352e-05,
+      "loss": 1.1017,
+      "step": 99510
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7415841258350062e-05,
+      "loss": 1.3,
+      "step": 99520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.741255059396492e-05,
+      "loss": 1.1191,
+      "step": 99530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.740925992957978e-05,
+      "loss": 1.1093,
+      "step": 99540
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7405969265194643e-05,
+      "loss": 1.0193,
+      "step": 99550
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7402678600809504e-05,
+      "loss": 1.174,
+      "step": 99560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7399387936424365e-05,
+      "loss": 1.2442,
+      "step": 99570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7396097272039223e-05,
+      "loss": 1.2838,
+      "step": 99580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7392806607654084e-05,
+      "loss": 1.1569,
+      "step": 99590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7389515943268946e-05,
+      "loss": 1.2909,
+      "step": 99600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7386225278883807e-05,
+      "loss": 1.2543,
+      "step": 99610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7382934614498668e-05,
+      "loss": 1.2048,
+      "step": 99620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7379643950113526e-05,
+      "loss": 1.0434,
+      "step": 99630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7376353285728387e-05,
+      "loss": 1.3326,
+      "step": 99640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.737306262134325e-05,
+      "loss": 1.2214,
+      "step": 99650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.736977195695811e-05,
+      "loss": 1.2581,
+      "step": 99660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.736648129257297e-05,
+      "loss": 1.1625,
+      "step": 99670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7363190628187832e-05,
+      "loss": 1.3953,
+      "step": 99680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7359899963802694e-05,
+      "loss": 1.2427,
+      "step": 99690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7356609299417555e-05,
+      "loss": 1.2405,
+      "step": 99700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7353318635032413e-05,
+      "loss": 1.3108,
+      "step": 99710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7350027970647274e-05,
+      "loss": 1.4248,
+      "step": 99720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7346737306262135e-05,
+      "loss": 1.2347,
+      "step": 99730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7343446641876997e-05,
+      "loss": 1.1636,
+      "step": 99740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7340155977491858e-05,
+      "loss": 1.1047,
+      "step": 99750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.733686531310672e-05,
+      "loss": 1.4634,
+      "step": 99760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.733357464872158e-05,
+      "loss": 1.1665,
+      "step": 99770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.733028398433644e-05,
+      "loss": 1.3432,
+      "step": 99780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.73269933199513e-05,
+      "loss": 1.2389,
+      "step": 99790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.732370265556616e-05,
+      "loss": 1.3334,
+      "step": 99800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7320411991181022e-05,
+      "loss": 1.2301,
+      "step": 99810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7317121326795883e-05,
+      "loss": 1.1917,
+      "step": 99820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.731383066241074e-05,
+      "loss": 1.1567,
+      "step": 99830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7310539998025603e-05,
+      "loss": 1.1423,
+      "step": 99840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7307249333640464e-05,
+      "loss": 1.2313,
+      "step": 99850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7303958669255325e-05,
+      "loss": 1.2776,
+      "step": 99860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7300668004870186e-05,
+      "loss": 1.2756,
+      "step": 99870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7297377340485044e-05,
+      "loss": 1.163,
+      "step": 99880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7294086676099906e-05,
+      "loss": 1.3853,
+      "step": 99890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7290796011714767e-05,
+      "loss": 1.1669,
+      "step": 99900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7287505347329628e-05,
+      "loss": 1.1428,
+      "step": 99910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.728421468294449e-05,
+      "loss": 1.1954,
+      "step": 99920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7280924018559347e-05,
+      "loss": 1.1433,
+      "step": 99930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.727763335417421e-05,
+      "loss": 1.2514,
+      "step": 99940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.727434268978907e-05,
+      "loss": 1.3345,
+      "step": 99950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.727105202540393e-05,
+      "loss": 1.2051,
+      "step": 99960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7267761361018792e-05,
+      "loss": 1.1485,
+      "step": 99970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.726447069663365e-05,
+      "loss": 1.205,
+      "step": 99980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.726118003224851e-05,
+      "loss": 1.312,
+      "step": 99990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7257889367863373e-05,
+      "loss": 1.1152,
+      "step": 100000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7254598703478234e-05,
+      "loss": 1.2129,
+      "step": 100010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7251308039093095e-05,
+      "loss": 1.2382,
+      "step": 100020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7248017374707953e-05,
+      "loss": 1.2144,
+      "step": 100030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7244726710322814e-05,
+      "loss": 1.1781,
+      "step": 100040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7241436045937676e-05,
+      "loss": 1.4049,
+      "step": 100050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7238145381552537e-05,
+      "loss": 1.0361,
+      "step": 100060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7234854717167398e-05,
+      "loss": 1.1681,
+      "step": 100070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7231564052782256e-05,
+      "loss": 1.0509,
+      "step": 100080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7228273388397117e-05,
+      "loss": 1.1427,
+      "step": 100090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.722498272401198e-05,
+      "loss": 1.1899,
+      "step": 100100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.722169205962684e-05,
+      "loss": 1.2904,
+      "step": 100110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.72184013952417e-05,
+      "loss": 1.2435,
+      "step": 100120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.721511073085656e-05,
+      "loss": 1.1895,
+      "step": 100130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.721182006647142e-05,
+      "loss": 1.2814,
+      "step": 100140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.720852940208628e-05,
+      "loss": 1.208,
+      "step": 100150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7205238737701143e-05,
+      "loss": 1.3057,
+      "step": 100160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7201948073316004e-05,
+      "loss": 1.1531,
+      "step": 100170
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7198657408930862e-05,
+      "loss": 1.2377,
+      "step": 100180
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7195366744545723e-05,
+      "loss": 1.2816,
+      "step": 100190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7192076080160584e-05,
+      "loss": 1.0867,
+      "step": 100200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7188785415775446e-05,
+      "loss": 1.0678,
+      "step": 100210
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7185494751390307e-05,
+      "loss": 1.137,
+      "step": 100220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7182204087005165e-05,
+      "loss": 1.1503,
+      "step": 100230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7178913422620026e-05,
+      "loss": 1.1749,
+      "step": 100240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7175622758234887e-05,
+      "loss": 1.3965,
+      "step": 100250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.717233209384975e-05,
+      "loss": 1.271,
+      "step": 100260
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.716904142946461e-05,
+      "loss": 1.0924,
+      "step": 100270
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7165750765079468e-05,
+      "loss": 1.1546,
+      "step": 100280
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.716246010069433e-05,
+      "loss": 1.2237,
+      "step": 100290
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.715916943630919e-05,
+      "loss": 1.313,
+      "step": 100300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.715587877192405e-05,
+      "loss": 1.0695,
+      "step": 100310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7152588107538913e-05,
+      "loss": 1.1567,
+      "step": 100320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.714929744315377e-05,
+      "loss": 1.2238,
+      "step": 100330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7146006778768632e-05,
+      "loss": 1.0697,
+      "step": 100340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7142716114383493e-05,
+      "loss": 1.2075,
+      "step": 100350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7139425449998355e-05,
+      "loss": 1.1263,
+      "step": 100360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7136134785613216e-05,
+      "loss": 1.2485,
+      "step": 100370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7132844121228077e-05,
+      "loss": 1.2207,
+      "step": 100380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7129553456842935e-05,
+      "loss": 1.3056,
+      "step": 100390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7126262792457796e-05,
+      "loss": 1.2994,
+      "step": 100400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7122972128072658e-05,
+      "loss": 1.2184,
+      "step": 100410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.711968146368752e-05,
+      "loss": 1.205,
+      "step": 100420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.711639079930238e-05,
+      "loss": 1.3867,
+      "step": 100430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.711310013491724e-05,
+      "loss": 1.1897,
+      "step": 100440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7109809470532103e-05,
+      "loss": 1.0538,
+      "step": 100450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7106518806146964e-05,
+      "loss": 1.087,
+      "step": 100460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7103228141761822e-05,
+      "loss": 1.3082,
+      "step": 100470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7099937477376683e-05,
+      "loss": 1.6904,
+      "step": 100480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7096646812991544e-05,
+      "loss": 1.0905,
+      "step": 100490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7093356148606406e-05,
+      "loss": 1.4026,
+      "step": 100500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7090065484221267e-05,
+      "loss": 1.1485,
+      "step": 100510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7086774819836128e-05,
+      "loss": 1.3185,
+      "step": 100520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7083484155450986e-05,
+      "loss": 1.1235,
+      "step": 100530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7080193491065847e-05,
+      "loss": 1.2269,
+      "step": 100540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.707690282668071e-05,
+      "loss": 1.2871,
+      "step": 100550
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.707361216229557e-05,
+      "loss": 1.2762,
+      "step": 100560
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.707032149791043e-05,
+      "loss": 1.1669,
+      "step": 100570
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.706703083352529e-05,
+      "loss": 1.2086,
+      "step": 100580
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.706374016914015e-05,
+      "loss": 1.1783,
+      "step": 100590
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.706044950475501e-05,
+      "loss": 1.2131,
+      "step": 100600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7057158840369873e-05,
+      "loss": 1.127,
+      "step": 100610
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7053868175984734e-05,
+      "loss": 1.1487,
+      "step": 100620
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7050577511599595e-05,
+      "loss": 1.0139,
+      "step": 100630
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7047286847214453e-05,
+      "loss": 1.2491,
+      "step": 100640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7043996182829314e-05,
+      "loss": 1.0505,
+      "step": 100650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7040705518444176e-05,
+      "loss": 1.3907,
+      "step": 100660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7037414854059037e-05,
+      "loss": 1.2186,
+      "step": 100670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7034124189673898e-05,
+      "loss": 1.3084,
+      "step": 100680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7030833525288756e-05,
+      "loss": 1.2266,
+      "step": 100690
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7027542860903617e-05,
+      "loss": 1.2102,
+      "step": 100700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.702425219651848e-05,
+      "loss": 1.2547,
+      "step": 100710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.702096153213334e-05,
+      "loss": 1.1183,
+      "step": 100720
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.70176708677482e-05,
+      "loss": 1.1943,
+      "step": 100730
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.701438020336306e-05,
+      "loss": 1.2076,
+      "step": 100740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.701108953897792e-05,
+      "loss": 1.2684,
+      "step": 100750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.700779887459278e-05,
+      "loss": 1.1748,
+      "step": 100760
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7004508210207643e-05,
+      "loss": 1.3192,
+      "step": 100770
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7001217545822504e-05,
+      "loss": 1.0542,
+      "step": 100780
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6997926881437362e-05,
+      "loss": 1.2686,
+      "step": 100790
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6994636217052223e-05,
+      "loss": 1.2421,
+      "step": 100800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6991345552667085e-05,
+      "loss": 1.3416,
+      "step": 100810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6988054888281946e-05,
+      "loss": 1.2556,
+      "step": 100820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6984764223896807e-05,
+      "loss": 1.1711,
+      "step": 100830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6981473559511665e-05,
+      "loss": 1.254,
+      "step": 100840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6978182895126526e-05,
+      "loss": 1.4607,
+      "step": 100850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6974892230741387e-05,
+      "loss": 1.3408,
+      "step": 100860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.697160156635625e-05,
+      "loss": 1.203,
+      "step": 100870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.696831090197111e-05,
+      "loss": 1.3079,
+      "step": 100880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6965020237585968e-05,
+      "loss": 1.2564,
+      "step": 100890
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.696172957320083e-05,
+      "loss": 1.2161,
+      "step": 100900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.695843890881569e-05,
+      "loss": 1.3469,
+      "step": 100910
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.695514824443055e-05,
+      "loss": 1.2451,
+      "step": 100920
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6951857580045413e-05,
+      "loss": 1.4207,
+      "step": 100930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.694856691566027e-05,
+      "loss": 1.0964,
+      "step": 100940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6945276251275132e-05,
+      "loss": 1.2945,
+      "step": 100950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6941985586889993e-05,
+      "loss": 1.2838,
+      "step": 100960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6938694922504855e-05,
+      "loss": 1.3232,
+      "step": 100970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6935404258119716e-05,
+      "loss": 1.341,
+      "step": 100980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6932113593734574e-05,
+      "loss": 1.1804,
+      "step": 100990
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6928822929349435e-05,
+      "loss": 1.381,
+      "step": 101000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6925532264964296e-05,
+      "loss": 1.1712,
+      "step": 101010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6922241600579158e-05,
+      "loss": 1.117,
+      "step": 101020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.691895093619402e-05,
+      "loss": 1.1139,
+      "step": 101030
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6915660271808877e-05,
+      "loss": 1.2722,
+      "step": 101040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6912369607423738e-05,
+      "loss": 1.0725,
+      "step": 101050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.69090789430386e-05,
+      "loss": 1.2061,
+      "step": 101060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.690578827865346e-05,
+      "loss": 1.1766,
+      "step": 101070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6902497614268322e-05,
+      "loss": 1.2976,
+      "step": 101080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.689920694988318e-05,
+      "loss": 1.0884,
+      "step": 101090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.689591628549804e-05,
+      "loss": 1.0629,
+      "step": 101100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6892625621112902e-05,
+      "loss": 1.1687,
+      "step": 101110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6889334956727763e-05,
+      "loss": 1.323,
+      "step": 101120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6886044292342625e-05,
+      "loss": 1.2106,
+      "step": 101130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6882753627957486e-05,
+      "loss": 1.2684,
+      "step": 101140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6879462963572344e-05,
+      "loss": 1.1732,
+      "step": 101150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6876172299187205e-05,
+      "loss": 1.3113,
+      "step": 101160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6872881634802066e-05,
+      "loss": 1.1761,
+      "step": 101170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6869590970416928e-05,
+      "loss": 1.1305,
+      "step": 101180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.686630030603179e-05,
+      "loss": 1.3177,
+      "step": 101190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.686300964164665e-05,
+      "loss": 1.4025,
+      "step": 101200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.685971897726151e-05,
+      "loss": 1.2357,
+      "step": 101210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6856428312876373e-05,
+      "loss": 1.0606,
+      "step": 101220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.685313764849123e-05,
+      "loss": 1.1564,
+      "step": 101230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6849846984106092e-05,
+      "loss": 1.1983,
+      "step": 101240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6846556319720953e-05,
+      "loss": 1.2404,
+      "step": 101250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6843265655335814e-05,
+      "loss": 1.3078,
+      "step": 101260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6839974990950676e-05,
+      "loss": 1.0359,
+      "step": 101270
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6836684326565537e-05,
+      "loss": 1.0762,
+      "step": 101280
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6833393662180395e-05,
+      "loss": 1.3148,
+      "step": 101290
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6830102997795256e-05,
+      "loss": 1.3782,
+      "step": 101300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6826812333410117e-05,
+      "loss": 1.2186,
+      "step": 101310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.682352166902498e-05,
+      "loss": 1.2015,
+      "step": 101320
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.682023100463984e-05,
+      "loss": 1.1669,
+      "step": 101330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6816940340254698e-05,
+      "loss": 1.1357,
+      "step": 101340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.681364967586956e-05,
+      "loss": 1.2796,
+      "step": 101350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.681035901148442e-05,
+      "loss": 1.539,
+      "step": 101360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.680706834709928e-05,
+      "loss": 1.3675,
+      "step": 101370
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6803777682714143e-05,
+      "loss": 1.2174,
+      "step": 101380
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6800487018329e-05,
+      "loss": 1.1488,
+      "step": 101390
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6797196353943862e-05,
+      "loss": 1.1634,
+      "step": 101400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6793905689558723e-05,
+      "loss": 1.2484,
+      "step": 101410
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6790615025173585e-05,
+      "loss": 1.3619,
+      "step": 101420
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6787324360788446e-05,
+      "loss": 1.3648,
+      "step": 101430
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6784033696403304e-05,
+      "loss": 1.1274,
+      "step": 101440
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6780743032018165e-05,
+      "loss": 1.2066,
+      "step": 101450
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6777452367633026e-05,
+      "loss": 1.2658,
+      "step": 101460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6774161703247888e-05,
+      "loss": 1.4006,
+      "step": 101470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.677087103886275e-05,
+      "loss": 1.1801,
+      "step": 101480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6767580374477607e-05,
+      "loss": 1.3185,
+      "step": 101490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6764289710092468e-05,
+      "loss": 1.3078,
+      "step": 101500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.676099904570733e-05,
+      "loss": 1.3424,
+      "step": 101510
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.675770838132219e-05,
+      "loss": 1.2893,
+      "step": 101520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6754417716937052e-05,
+      "loss": 1.2491,
+      "step": 101530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.675112705255191e-05,
+      "loss": 1.0906,
+      "step": 101540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.674783638816677e-05,
+      "loss": 1.1583,
+      "step": 101550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6744545723781632e-05,
+      "loss": 1.1827,
+      "step": 101560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6741255059396493e-05,
+      "loss": 1.2676,
+      "step": 101570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6737964395011355e-05,
+      "loss": 1.0457,
+      "step": 101580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6734673730626213e-05,
+      "loss": 1.209,
+      "step": 101590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6731383066241074e-05,
+      "loss": 1.3608,
+      "step": 101600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6728092401855935e-05,
+      "loss": 1.1963,
+      "step": 101610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6724801737470796e-05,
+      "loss": 1.0833,
+      "step": 101620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6721511073085658e-05,
+      "loss": 1.282,
+      "step": 101630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6718220408700516e-05,
+      "loss": 0.8003,
+      "step": 101640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6714929744315377e-05,
+      "loss": 0.8454,
+      "step": 101650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6711639079930238e-05,
+      "loss": 0.8023,
+      "step": 101660
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.67083484155451e-05,
+      "loss": 0.6716,
+      "step": 101670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.670505775115996e-05,
+      "loss": 0.6376,
+      "step": 101680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.670176708677482e-05,
+      "loss": 0.639,
+      "step": 101690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.669847642238968e-05,
+      "loss": 0.6685,
+      "step": 101700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.669518575800454e-05,
+      "loss": 0.6554,
+      "step": 101710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6691895093619402e-05,
+      "loss": 0.7649,
+      "step": 101720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6688604429234264e-05,
+      "loss": 0.8104,
+      "step": 101730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.668531376484912e-05,
+      "loss": 0.6249,
+      "step": 101740
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6682023100463983e-05,
+      "loss": 0.6873,
+      "step": 101750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6678732436078844e-05,
+      "loss": 0.5126,
+      "step": 101760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6675441771693705e-05,
+      "loss": 0.6074,
+      "step": 101770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6672151107308566e-05,
+      "loss": 0.7252,
+      "step": 101780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6668860442923424e-05,
+      "loss": 0.7151,
+      "step": 101790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6665569778538286e-05,
+      "loss": 0.7431,
+      "step": 101800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6662279114153147e-05,
+      "loss": 0.6673,
+      "step": 101810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6658988449768008e-05,
+      "loss": 0.6918,
+      "step": 101820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.665569778538287e-05,
+      "loss": 0.5988,
+      "step": 101830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6652407120997727e-05,
+      "loss": 0.6893,
+      "step": 101840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.664911645661259e-05,
+      "loss": 0.5869,
+      "step": 101850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.664582579222745e-05,
+      "loss": 0.6773,
+      "step": 101860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.664253512784231e-05,
+      "loss": 0.6968,
+      "step": 101870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6639244463457172e-05,
+      "loss": 0.6606,
+      "step": 101880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6635953799072034e-05,
+      "loss": 0.5443,
+      "step": 101890
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6632663134686895e-05,
+      "loss": 0.6606,
+      "step": 101900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6629372470301756e-05,
+      "loss": 0.5671,
+      "step": 101910
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6626081805916614e-05,
+      "loss": 0.832,
+      "step": 101920
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6622791141531475e-05,
+      "loss": 0.6292,
+      "step": 101930
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6619500477146337e-05,
+      "loss": 0.7109,
+      "step": 101940
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6616209812761198e-05,
+      "loss": 0.6773,
+      "step": 101950
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.661291914837606e-05,
+      "loss": 0.7072,
+      "step": 101960
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.660962848399092e-05,
+      "loss": 0.5886,
+      "step": 101970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.660633781960578e-05,
+      "loss": 0.7092,
+      "step": 101980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.660304715522064e-05,
+      "loss": 0.6797,
+      "step": 101990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.65997564908355e-05,
+      "loss": 0.6144,
+      "step": 102000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6596465826450362e-05,
+      "loss": 0.6552,
+      "step": 102010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6593175162065223e-05,
+      "loss": 0.6162,
+      "step": 102020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6589884497680085e-05,
+      "loss": 0.6604,
+      "step": 102030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6586593833294946e-05,
+      "loss": 0.6976,
+      "step": 102040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6583303168909804e-05,
+      "loss": 0.631,
+      "step": 102050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6580012504524665e-05,
+      "loss": 0.5799,
+      "step": 102060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6576721840139526e-05,
+      "loss": 0.7438,
+      "step": 102070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6573431175754388e-05,
+      "loss": 0.4892,
+      "step": 102080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.657014051136925e-05,
+      "loss": 0.543,
+      "step": 102090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6566849846984107e-05,
+      "loss": 0.561,
+      "step": 102100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6563559182598968e-05,
+      "loss": 0.7682,
+      "step": 102110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.656026851821383e-05,
+      "loss": 0.6394,
+      "step": 102120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.655697785382869e-05,
+      "loss": 0.671,
+      "step": 102130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6553687189443552e-05,
+      "loss": 0.736,
+      "step": 102140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.655039652505841e-05,
+      "loss": 0.5631,
+      "step": 102150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.654710586067327e-05,
+      "loss": 0.5741,
+      "step": 102160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6543815196288132e-05,
+      "loss": 0.5736,
+      "step": 102170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6540524531902993e-05,
+      "loss": 0.6814,
+      "step": 102180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6537233867517855e-05,
+      "loss": 0.6533,
+      "step": 102190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6533943203132713e-05,
+      "loss": 0.6784,
+      "step": 102200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6530652538747574e-05,
+      "loss": 0.6127,
+      "step": 102210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6527361874362435e-05,
+      "loss": 0.6347,
+      "step": 102220
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6524071209977296e-05,
+      "loss": 0.6062,
+      "step": 102230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6520780545592158e-05,
+      "loss": 0.5738,
+      "step": 102240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6517489881207016e-05,
+      "loss": 0.711,
+      "step": 102250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6514199216821877e-05,
+      "loss": 0.7564,
+      "step": 102260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6510908552436738e-05,
+      "loss": 0.6693,
+      "step": 102270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.65076178880516e-05,
+      "loss": 0.8029,
+      "step": 102280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.650432722366646e-05,
+      "loss": 0.7096,
+      "step": 102290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.650103655928132e-05,
+      "loss": 0.7124,
+      "step": 102300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.649774589489618e-05,
+      "loss": 0.6617,
+      "step": 102310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.649445523051104e-05,
+      "loss": 0.6771,
+      "step": 102320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6491164566125902e-05,
+      "loss": 0.7474,
+      "step": 102330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6487873901740764e-05,
+      "loss": 0.5798,
+      "step": 102340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.648458323735562e-05,
+      "loss": 0.7444,
+      "step": 102350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6481292572970483e-05,
+      "loss": 0.6038,
+      "step": 102360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6478001908585344e-05,
+      "loss": 0.6045,
+      "step": 102370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6474711244200205e-05,
+      "loss": 0.6395,
+      "step": 102380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6471420579815067e-05,
+      "loss": 0.7798,
+      "step": 102390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6468129915429924e-05,
+      "loss": 0.6399,
+      "step": 102400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6464839251044786e-05,
+      "loss": 0.6279,
+      "step": 102410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6461548586659647e-05,
+      "loss": 0.736,
+      "step": 102420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6458257922274508e-05,
+      "loss": 0.6091,
+      "step": 102430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.645496725788937e-05,
+      "loss": 0.6473,
+      "step": 102440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6451676593504227e-05,
+      "loss": 0.6074,
+      "step": 102450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.644838592911909e-05,
+      "loss": 0.5667,
+      "step": 102460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.644509526473395e-05,
+      "loss": 0.5957,
+      "step": 102470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.644180460034881e-05,
+      "loss": 0.5413,
+      "step": 102480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6438513935963672e-05,
+      "loss": 0.7747,
+      "step": 102490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.643522327157853e-05,
+      "loss": 0.7033,
+      "step": 102500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.643193260719339e-05,
+      "loss": 0.6108,
+      "step": 102510
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6428641942808253e-05,
+      "loss": 0.657,
+      "step": 102520
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6425351278423114e-05,
+      "loss": 0.61,
+      "step": 102530
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6422060614037975e-05,
+      "loss": 0.6614,
+      "step": 102540
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6418769949652833e-05,
+      "loss": 0.7567,
+      "step": 102550
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6415479285267695e-05,
+      "loss": 0.6183,
+      "step": 102560
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6412188620882556e-05,
+      "loss": 0.6564,
+      "step": 102570
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6408897956497417e-05,
+      "loss": 0.6308,
+      "step": 102580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.640560729211228e-05,
+      "loss": 0.6029,
+      "step": 102590
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6402316627727136e-05,
+      "loss": 0.5318,
+      "step": 102600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6399025963341997e-05,
+      "loss": 0.5904,
+      "step": 102610
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.639573529895686e-05,
+      "loss": 0.5916,
+      "step": 102620
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.639244463457172e-05,
+      "loss": 0.618,
+      "step": 102630
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.638915397018658e-05,
+      "loss": 0.5831,
+      "step": 102640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6385863305801443e-05,
+      "loss": 0.7081,
+      "step": 102650
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6382572641416304e-05,
+      "loss": 0.6161,
+      "step": 102660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6379281977031165e-05,
+      "loss": 0.7869,
+      "step": 102670
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6375991312646023e-05,
+      "loss": 0.7567,
+      "step": 102680
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6372700648260884e-05,
+      "loss": 0.6973,
+      "step": 102690
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6369409983875745e-05,
+      "loss": 0.7946,
+      "step": 102700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6366119319490607e-05,
+      "loss": 0.5881,
+      "step": 102710
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6362828655105468e-05,
+      "loss": 0.6492,
+      "step": 102720
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.635953799072033e-05,
+      "loss": 0.7346,
+      "step": 102730
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.635624732633519e-05,
+      "loss": 0.5949,
+      "step": 102740
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.635295666195005e-05,
+      "loss": 0.6857,
+      "step": 102750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.634966599756491e-05,
+      "loss": 0.611,
+      "step": 102760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.634637533317977e-05,
+      "loss": 0.6959,
+      "step": 102770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6343084668794632e-05,
+      "loss": 0.7351,
+      "step": 102780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6339794004409494e-05,
+      "loss": 0.6479,
+      "step": 102790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.633650334002435e-05,
+      "loss": 0.7438,
+      "step": 102800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6333212675639213e-05,
+      "loss": 0.6586,
+      "step": 102810
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6329922011254074e-05,
+      "loss": 0.6377,
+      "step": 102820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6326631346868935e-05,
+      "loss": 0.5214,
+      "step": 102830
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6323340682483796e-05,
+      "loss": 0.7167,
+      "step": 102840
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6320050018098654e-05,
+      "loss": 0.6692,
+      "step": 102850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6316759353713516e-05,
+      "loss": 0.7712,
+      "step": 102860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6313468689328377e-05,
+      "loss": 0.6202,
+      "step": 102870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6310178024943238e-05,
+      "loss": 0.6702,
+      "step": 102880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.63068873605581e-05,
+      "loss": 0.6186,
+      "step": 102890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6303596696172957e-05,
+      "loss": 0.6627,
+      "step": 102900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.630030603178782e-05,
+      "loss": 0.6467,
+      "step": 102910
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.629701536740268e-05,
+      "loss": 0.6513,
+      "step": 102920
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.629372470301754e-05,
+      "loss": 0.6936,
+      "step": 102930
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6290434038632402e-05,
+      "loss": 0.6899,
+      "step": 102940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.628714337424726e-05,
+      "loss": 0.5898,
+      "step": 102950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.628385270986212e-05,
+      "loss": 0.5321,
+      "step": 102960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6280562045476983e-05,
+      "loss": 0.605,
+      "step": 102970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6277271381091844e-05,
+      "loss": 0.676,
+      "step": 102980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6273980716706705e-05,
+      "loss": 0.6381,
+      "step": 102990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6270690052321563e-05,
+      "loss": 0.632,
+      "step": 103000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6267399387936424e-05,
+      "loss": 0.5337,
+      "step": 103010
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6264108723551286e-05,
+      "loss": 0.6778,
+      "step": 103020
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6260818059166147e-05,
+      "loss": 0.5999,
+      "step": 103030
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6257527394781008e-05,
+      "loss": 0.6294,
+      "step": 103040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6254236730395866e-05,
+      "loss": 0.5599,
+      "step": 103050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6250946066010727e-05,
+      "loss": 0.702,
+      "step": 103060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.624765540162559e-05,
+      "loss": 0.6037,
+      "step": 103070
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.624436473724045e-05,
+      "loss": 0.6681,
+      "step": 103080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.624107407285531e-05,
+      "loss": 0.6857,
+      "step": 103090
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.623778340847017e-05,
+      "loss": 0.6378,
+      "step": 103100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.623449274408503e-05,
+      "loss": 0.5984,
+      "step": 103110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.623120207969989e-05,
+      "loss": 0.6642,
+      "step": 103120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6227911415314753e-05,
+      "loss": 0.6613,
+      "step": 103130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6224620750929614e-05,
+      "loss": 0.6869,
+      "step": 103140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6221330086544472e-05,
+      "loss": 0.5562,
+      "step": 103150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6218039422159333e-05,
+      "loss": 0.7086,
+      "step": 103160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6214748757774195e-05,
+      "loss": 0.6895,
+      "step": 103170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6211458093389056e-05,
+      "loss": 0.6667,
+      "step": 103180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6208167429003917e-05,
+      "loss": 0.6418,
+      "step": 103190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6204876764618775e-05,
+      "loss": 0.611,
+      "step": 103200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6201586100233636e-05,
+      "loss": 0.6769,
+      "step": 103210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6198295435848498e-05,
+      "loss": 0.6916,
+      "step": 103220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.619500477146336e-05,
+      "loss": 0.6244,
+      "step": 103230
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.619171410707822e-05,
+      "loss": 0.6014,
+      "step": 103240
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6188423442693078e-05,
+      "loss": 0.6122,
+      "step": 103250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.618513277830794e-05,
+      "loss": 0.6065,
+      "step": 103260
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.61818421139228e-05,
+      "loss": 0.5952,
+      "step": 103270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6178551449537662e-05,
+      "loss": 0.6481,
+      "step": 103280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6175260785152523e-05,
+      "loss": 0.5461,
+      "step": 103290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.617197012076738e-05,
+      "loss": 0.6971,
+      "step": 103300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6168679456382242e-05,
+      "loss": 0.6625,
+      "step": 103310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6165388791997103e-05,
+      "loss": 0.572,
+      "step": 103320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6162098127611965e-05,
+      "loss": 0.7083,
+      "step": 103330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6158807463226826e-05,
+      "loss": 0.6405,
+      "step": 103340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6155516798841687e-05,
+      "loss": 0.6336,
+      "step": 103350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6152226134456545e-05,
+      "loss": 0.6515,
+      "step": 103360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6148935470071406e-05,
+      "loss": 0.593,
+      "step": 103370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6145644805686268e-05,
+      "loss": 0.7508,
+      "step": 103380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.614235414130113e-05,
+      "loss": 0.6606,
+      "step": 103390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.613906347691599e-05,
+      "loss": 0.6968,
+      "step": 103400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.613577281253085e-05,
+      "loss": 0.7216,
+      "step": 103410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6132482148145713e-05,
+      "loss": 0.6408,
+      "step": 103420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6129191483760574e-05,
+      "loss": 0.5234,
+      "step": 103430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6125900819375432e-05,
+      "loss": 0.6381,
+      "step": 103440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6122610154990293e-05,
+      "loss": 0.7466,
+      "step": 103450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6119319490605154e-05,
+      "loss": 0.739,
+      "step": 103460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6116028826220016e-05,
+      "loss": 0.7953,
+      "step": 103470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6112738161834877e-05,
+      "loss": 0.6486,
+      "step": 103480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6109447497449738e-05,
+      "loss": 0.6209,
+      "step": 103490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.61061568330646e-05,
+      "loss": 0.5908,
+      "step": 103500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6102866168679457e-05,
+      "loss": 0.576,
+      "step": 103510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.609957550429432e-05,
+      "loss": 0.6812,
+      "step": 103520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.609628483990918e-05,
+      "loss": 0.6939,
+      "step": 103530
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.609299417552404e-05,
+      "loss": 0.5983,
+      "step": 103540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6089703511138902e-05,
+      "loss": 0.583,
+      "step": 103550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.608641284675376e-05,
+      "loss": 0.5508,
+      "step": 103560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.608312218236862e-05,
+      "loss": 0.541,
+      "step": 103570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6079831517983483e-05,
+      "loss": 0.5847,
+      "step": 103580
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6076540853598344e-05,
+      "loss": 0.6478,
+      "step": 103590
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6073250189213205e-05,
+      "loss": 0.6215,
+      "step": 103600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6069959524828063e-05,
+      "loss": 0.8286,
+      "step": 103610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6066668860442924e-05,
+      "loss": 0.614,
+      "step": 103620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6063378196057786e-05,
+      "loss": 0.6176,
+      "step": 103630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6060087531672647e-05,
+      "loss": 0.6172,
+      "step": 103640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6056796867287508e-05,
+      "loss": 0.6284,
+      "step": 103650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6053506202902366e-05,
+      "loss": 0.6869,
+      "step": 103660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6050215538517227e-05,
+      "loss": 0.5509,
+      "step": 103670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.604692487413209e-05,
+      "loss": 0.5568,
+      "step": 103680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.604363420974695e-05,
+      "loss": 0.6749,
+      "step": 103690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.604034354536181e-05,
+      "loss": 0.6686,
+      "step": 103700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.603705288097667e-05,
+      "loss": 0.6105,
+      "step": 103710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.603376221659153e-05,
+      "loss": 0.6598,
+      "step": 103720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.603047155220639e-05,
+      "loss": 0.5564,
+      "step": 103730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6027180887821253e-05,
+      "loss": 0.6142,
+      "step": 103740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6023890223436114e-05,
+      "loss": 0.7481,
+      "step": 103750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6020599559050972e-05,
+      "loss": 0.7998,
+      "step": 103760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6017308894665833e-05,
+      "loss": 0.4538,
+      "step": 103770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6014018230280695e-05,
+      "loss": 0.561,
+      "step": 103780
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6010727565895556e-05,
+      "loss": 0.6757,
+      "step": 103790
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6007436901510417e-05,
+      "loss": 0.6678,
+      "step": 103800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6004146237125275e-05,
+      "loss": 0.6816,
+      "step": 103810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6000855572740136e-05,
+      "loss": 0.6323,
+      "step": 103820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5997564908354998e-05,
+      "loss": 0.616,
+      "step": 103830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.599427424396986e-05,
+      "loss": 0.4726,
+      "step": 103840
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.599098357958472e-05,
+      "loss": 0.6194,
+      "step": 103850
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5987692915199578e-05,
+      "loss": 0.617,
+      "step": 103860
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598440225081444e-05,
+      "loss": 0.612,
+      "step": 103870
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.59811115864293e-05,
+      "loss": 0.6108,
+      "step": 103880
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5977820922044162e-05,
+      "loss": 0.6383,
+      "step": 103890
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5974530257659023e-05,
+      "loss": 0.5957,
+      "step": 103900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.597123959327388e-05,
+      "loss": 0.6309,
+      "step": 103910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5967948928888742e-05,
+      "loss": 0.6115,
+      "step": 103920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5964658264503603e-05,
+      "loss": 0.5816,
+      "step": 103930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5961367600118465e-05,
+      "loss": 0.6088,
+      "step": 103940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5958076935733326e-05,
+      "loss": 0.6039,
+      "step": 103950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5954786271348184e-05,
+      "loss": 0.6349,
+      "step": 103960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5951495606963045e-05,
+      "loss": 0.6741,
+      "step": 103970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5948204942577906e-05,
+      "loss": 0.7622,
+      "step": 103980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5944914278192768e-05,
+      "loss": 0.7129,
+      "step": 103990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.594162361380763e-05,
+      "loss": 0.6835,
+      "step": 104000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5938332949422487e-05,
+      "loss": 0.543,
+      "step": 104010
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5935042285037348e-05,
+      "loss": 0.6238,
+      "step": 104020
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.593175162065221e-05,
+      "loss": 0.8505,
+      "step": 104030
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.592846095626707e-05,
+      "loss": 0.7334,
+      "step": 104040
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5925170291881932e-05,
+      "loss": 0.666,
+      "step": 104050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.592187962749679e-05,
+      "loss": 0.6759,
+      "step": 104060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.591858896311165e-05,
+      "loss": 0.6675,
+      "step": 104070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5915298298726512e-05,
+      "loss": 0.6907,
+      "step": 104080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5912007634341374e-05,
+      "loss": 0.7961,
+      "step": 104090
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5908716969956235e-05,
+      "loss": 0.654,
+      "step": 104100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5905426305571096e-05,
+      "loss": 0.4853,
+      "step": 104110
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5902135641185954e-05,
+      "loss": 0.6887,
+      "step": 104120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5898844976800815e-05,
+      "loss": 0.6518,
+      "step": 104130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5895554312415677e-05,
+      "loss": 0.7304,
+      "step": 104140
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5892263648030538e-05,
+      "loss": 0.562,
+      "step": 104150
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.58889729836454e-05,
+      "loss": 0.6698,
+      "step": 104160
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.588568231926026e-05,
+      "loss": 0.5897,
+      "step": 104170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.588239165487512e-05,
+      "loss": 0.6052,
+      "step": 104180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5879100990489983e-05,
+      "loss": 0.583,
+      "step": 104190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.587581032610484e-05,
+      "loss": 0.6188,
+      "step": 104200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5872519661719702e-05,
+      "loss": 0.6274,
+      "step": 104210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5869228997334563e-05,
+      "loss": 0.6464,
+      "step": 104220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5865938332949425e-05,
+      "loss": 0.636,
+      "step": 104230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5862647668564286e-05,
+      "loss": 0.7223,
+      "step": 104240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5859357004179147e-05,
+      "loss": 0.5945,
+      "step": 104250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5856066339794005e-05,
+      "loss": 0.6285,
+      "step": 104260
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5852775675408866e-05,
+      "loss": 0.6429,
+      "step": 104270
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5849485011023727e-05,
+      "loss": 0.8498,
+      "step": 104280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.584619434663859e-05,
+      "loss": 0.6514,
+      "step": 104290
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.584290368225345e-05,
+      "loss": 0.5507,
+      "step": 104300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5839613017868308e-05,
+      "loss": 0.5373,
+      "step": 104310
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.583632235348317e-05,
+      "loss": 0.559,
+      "step": 104320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.583303168909803e-05,
+      "loss": 0.5966,
+      "step": 104330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5829741024712892e-05,
+      "loss": 0.5996,
+      "step": 104340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5826450360327753e-05,
+      "loss": 0.603,
+      "step": 104350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.582315969594261e-05,
+      "loss": 0.5952,
+      "step": 104360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5819869031557472e-05,
+      "loss": 0.6657,
+      "step": 104370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5816578367172333e-05,
+      "loss": 0.6128,
+      "step": 104380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5813287702787195e-05,
+      "loss": 0.6972,
+      "step": 104390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5809997038402056e-05,
+      "loss": 0.6562,
+      "step": 104400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5806706374016914e-05,
+      "loss": 0.6321,
+      "step": 104410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5803415709631775e-05,
+      "loss": 0.7163,
+      "step": 104420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5800125045246636e-05,
+      "loss": 0.6729,
+      "step": 104430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5796834380861498e-05,
+      "loss": 0.7282,
+      "step": 104440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.579354371647636e-05,
+      "loss": 0.5639,
+      "step": 104450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5790253052091217e-05,
+      "loss": 0.683,
+      "step": 104460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5786962387706078e-05,
+      "loss": 0.6031,
+      "step": 104470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.578367172332094e-05,
+      "loss": 0.7958,
+      "step": 104480
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.57803810589358e-05,
+      "loss": 0.687,
+      "step": 104490
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5777090394550662e-05,
+      "loss": 0.6194,
+      "step": 104500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.577379973016552e-05,
+      "loss": 0.6693,
+      "step": 104510
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.577050906578038e-05,
+      "loss": 0.6104,
+      "step": 104520
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5767218401395242e-05,
+      "loss": 0.7652,
+      "step": 104530
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5763927737010104e-05,
+      "loss": 0.6108,
+      "step": 104540
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5760637072624965e-05,
+      "loss": 0.6599,
+      "step": 104550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5757346408239823e-05,
+      "loss": 0.6688,
+      "step": 104560
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5754055743854684e-05,
+      "loss": 0.5318,
+      "step": 104570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5750765079469545e-05,
+      "loss": 0.5902,
+      "step": 104580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5747474415084406e-05,
+      "loss": 0.6898,
+      "step": 104590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5744183750699268e-05,
+      "loss": 0.5931,
+      "step": 104600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5740893086314126e-05,
+      "loss": 0.6691,
+      "step": 104610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5737602421928987e-05,
+      "loss": 0.7421,
+      "step": 104620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5734311757543848e-05,
+      "loss": 0.6584,
+      "step": 104630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.573102109315871e-05,
+      "loss": 0.6117,
+      "step": 104640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.572773042877357e-05,
+      "loss": 0.5696,
+      "step": 104650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.572443976438843e-05,
+      "loss": 0.6759,
+      "step": 104660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.572114910000329e-05,
+      "loss": 0.6694,
+      "step": 104670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.571785843561815e-05,
+      "loss": 0.6313,
+      "step": 104680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5714567771233012e-05,
+      "loss": 0.6892,
+      "step": 104690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5711277106847874e-05,
+      "loss": 0.5982,
+      "step": 104700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.570798644246273e-05,
+      "loss": 0.5609,
+      "step": 104710
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5704695778077593e-05,
+      "loss": 0.6023,
+      "step": 104720
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5701405113692454e-05,
+      "loss": 0.6294,
+      "step": 104730
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5698114449307315e-05,
+      "loss": 0.5699,
+      "step": 104740
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5694823784922177e-05,
+      "loss": 0.603,
+      "step": 104750
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5691533120537034e-05,
+      "loss": 0.6493,
+      "step": 104760
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5688242456151896e-05,
+      "loss": 0.6207,
+      "step": 104770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5684951791766757e-05,
+      "loss": 0.5915,
+      "step": 104780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5681661127381618e-05,
+      "loss": 0.6312,
+      "step": 104790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.567837046299648e-05,
+      "loss": 0.6211,
+      "step": 104800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5675079798611337e-05,
+      "loss": 0.5555,
+      "step": 104810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.56717891342262e-05,
+      "loss": 0.7065,
+      "step": 104820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.566849846984106e-05,
+      "loss": 0.652,
+      "step": 104830
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.566520780545592e-05,
+      "loss": 0.7161,
+      "step": 104840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5661917141070782e-05,
+      "loss": 0.6017,
+      "step": 104850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5658626476685644e-05,
+      "loss": 0.6708,
+      "step": 104860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5655335812300505e-05,
+      "loss": 0.6032,
+      "step": 104870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5652045147915366e-05,
+      "loss": 0.6593,
+      "step": 104880
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5648754483530224e-05,
+      "loss": 0.6098,
+      "step": 104890
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5645463819145085e-05,
+      "loss": 0.6146,
+      "step": 104900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5642173154759947e-05,
+      "loss": 0.5723,
+      "step": 104910
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5638882490374808e-05,
+      "loss": 0.6766,
+      "step": 104920
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.563559182598967e-05,
+      "loss": 0.6958,
+      "step": 104930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.563230116160453e-05,
+      "loss": 0.6237,
+      "step": 104940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5629010497219392e-05,
+      "loss": 0.7206,
+      "step": 104950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5625719832834253e-05,
+      "loss": 0.668,
+      "step": 104960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.562242916844911e-05,
+      "loss": 0.6595,
+      "step": 104970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5619138504063972e-05,
+      "loss": 0.6092,
+      "step": 104980
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5615847839678833e-05,
+      "loss": 0.6217,
+      "step": 104990
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5612557175293695e-05,
+      "loss": 0.5899,
+      "step": 105000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5609266510908556e-05,
+      "loss": 0.6119,
+      "step": 105010
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5605975846523414e-05,
+      "loss": 0.6662,
+      "step": 105020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5602685182138275e-05,
+      "loss": 0.7349,
+      "step": 105030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5599394517753136e-05,
+      "loss": 0.6537,
+      "step": 105040
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5596103853367998e-05,
+      "loss": 0.6071,
+      "step": 105050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.559281318898286e-05,
+      "loss": 0.6169,
+      "step": 105060
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5589522524597717e-05,
+      "loss": 0.5333,
+      "step": 105070
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5586231860212578e-05,
+      "loss": 0.6201,
+      "step": 105080
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.558294119582744e-05,
+      "loss": 0.7377,
+      "step": 105090
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.55796505314423e-05,
+      "loss": 0.5531,
+      "step": 105100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5576359867057162e-05,
+      "loss": 0.5943,
+      "step": 105110
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.557306920267202e-05,
+      "loss": 0.6136,
+      "step": 105120
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.556977853828688e-05,
+      "loss": 0.638,
+      "step": 105130
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5566487873901742e-05,
+      "loss": 0.6521,
+      "step": 105140
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5563197209516604e-05,
+      "loss": 0.6184,
+      "step": 105150
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5559906545131465e-05,
+      "loss": 0.6038,
+      "step": 105160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5556615880746323e-05,
+      "loss": 0.6981,
+      "step": 105170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5553325216361184e-05,
+      "loss": 0.6312,
+      "step": 105180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5550034551976045e-05,
+      "loss": 0.6537,
+      "step": 105190
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5546743887590907e-05,
+      "loss": 0.7473,
+      "step": 105200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5543453223205768e-05,
+      "loss": 0.6738,
+      "step": 105210
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5540162558820626e-05,
+      "loss": 0.6851,
+      "step": 105220
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5536871894435487e-05,
+      "loss": 0.6569,
+      "step": 105230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5533581230050348e-05,
+      "loss": 0.7346,
+      "step": 105240
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.553029056566521e-05,
+      "loss": 0.6304,
+      "step": 105250
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.552699990128007e-05,
+      "loss": 0.6611,
+      "step": 105260
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.552370923689493e-05,
+      "loss": 0.6012,
+      "step": 105270
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.552041857250979e-05,
+      "loss": 0.6162,
+      "step": 105280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.551712790812465e-05,
+      "loss": 0.5946,
+      "step": 105290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5513837243739512e-05,
+      "loss": 0.5633,
+      "step": 105300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5510546579354374e-05,
+      "loss": 0.6235,
+      "step": 105310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.550725591496923e-05,
+      "loss": 0.5901,
+      "step": 105320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5503965250584093e-05,
+      "loss": 0.5712,
+      "step": 105330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5500674586198954e-05,
+      "loss": 0.5425,
+      "step": 105340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5497383921813815e-05,
+      "loss": 0.5879,
+      "step": 105350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5494093257428677e-05,
+      "loss": 0.7418,
+      "step": 105360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5490802593043534e-05,
+      "loss": 0.5497,
+      "step": 105370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5487511928658396e-05,
+      "loss": 0.7003,
+      "step": 105380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5484221264273257e-05,
+      "loss": 0.5959,
+      "step": 105390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5480930599888118e-05,
+      "loss": 0.6372,
+      "step": 105400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.547763993550298e-05,
+      "loss": 0.6484,
+      "step": 105410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5474349271117837e-05,
+      "loss": 0.6382,
+      "step": 105420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.54710586067327e-05,
+      "loss": 0.5692,
+      "step": 105430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.546776794234756e-05,
+      "loss": 0.603,
+      "step": 105440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.546447727796242e-05,
+      "loss": 0.6688,
+      "step": 105450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5461186613577283e-05,
+      "loss": 0.6045,
+      "step": 105460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.545789594919214e-05,
+      "loss": 0.6251,
+      "step": 105470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5454605284807e-05,
+      "loss": 0.6357,
+      "step": 105480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5451314620421863e-05,
+      "loss": 0.5845,
+      "step": 105490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5448023956036724e-05,
+      "loss": 0.5992,
+      "step": 105500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5444733291651585e-05,
+      "loss": 0.6877,
+      "step": 105510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5441442627266443e-05,
+      "loss": 0.6032,
+      "step": 105520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5438151962881305e-05,
+      "loss": 0.7201,
+      "step": 105530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5434861298496166e-05,
+      "loss": 0.5516,
+      "step": 105540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5431570634111027e-05,
+      "loss": 0.7096,
+      "step": 105550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.542827996972589e-05,
+      "loss": 0.6239,
+      "step": 105560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5424989305340746e-05,
+      "loss": 0.6078,
+      "step": 105570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5421698640955608e-05,
+      "loss": 0.633,
+      "step": 105580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.541840797657047e-05,
+      "loss": 0.4978,
+      "step": 105590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.541511731218533e-05,
+      "loss": 0.592,
+      "step": 105600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.541182664780019e-05,
+      "loss": 0.5928,
+      "step": 105610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5408535983415053e-05,
+      "loss": 0.5565,
+      "step": 105620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5405245319029914e-05,
+      "loss": 0.6135,
+      "step": 105630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5401954654644775e-05,
+      "loss": 0.6363,
+      "step": 105640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5398663990259633e-05,
+      "loss": 0.5154,
+      "step": 105650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5395373325874494e-05,
+      "loss": 0.5817,
+      "step": 105660
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5392082661489356e-05,
+      "loss": 0.6347,
+      "step": 105670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5388791997104217e-05,
+      "loss": 0.6054,
+      "step": 105680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5385501332719078e-05,
+      "loss": 0.6286,
+      "step": 105690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.538221066833394e-05,
+      "loss": 0.6954,
+      "step": 105700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.53789200039488e-05,
+      "loss": 0.6416,
+      "step": 105710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.537562933956366e-05,
+      "loss": 0.6792,
+      "step": 105720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.537233867517852e-05,
+      "loss": 0.6558,
+      "step": 105730
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.536904801079338e-05,
+      "loss": 0.6064,
+      "step": 105740
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5365757346408242e-05,
+      "loss": 0.5972,
+      "step": 105750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5362466682023104e-05,
+      "loss": 0.6604,
+      "step": 105760
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.535917601763796e-05,
+      "loss": 0.5234,
+      "step": 105770
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5355885353252823e-05,
+      "loss": 0.6611,
+      "step": 105780
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5352594688867684e-05,
+      "loss": 0.577,
+      "step": 105790
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5349304024482545e-05,
+      "loss": 0.6272,
+      "step": 105800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5346013360097407e-05,
+      "loss": 0.636,
+      "step": 105810
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5342722695712264e-05,
+      "loss": 0.648,
+      "step": 105820
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5339432031327126e-05,
+      "loss": 0.5476,
+      "step": 105830
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5336141366941987e-05,
+      "loss": 0.6152,
+      "step": 105840
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5332850702556848e-05,
+      "loss": 0.6135,
+      "step": 105850
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.532956003817171e-05,
+      "loss": 0.6741,
+      "step": 105860
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5326269373786567e-05,
+      "loss": 0.5836,
+      "step": 105870
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.532297870940143e-05,
+      "loss": 0.5958,
+      "step": 105880
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.531968804501629e-05,
+      "loss": 0.6468,
+      "step": 105890
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.531639738063115e-05,
+      "loss": 0.6115,
+      "step": 105900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5313106716246012e-05,
+      "loss": 0.6417,
+      "step": 105910
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.530981605186087e-05,
+      "loss": 0.6123,
+      "step": 105920
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.530652538747573e-05,
+      "loss": 0.7024,
+      "step": 105930
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5303234723090593e-05,
+      "loss": 0.6822,
+      "step": 105940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5299944058705454e-05,
+      "loss": 0.7493,
+      "step": 105950
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5296653394320315e-05,
+      "loss": 0.6321,
+      "step": 105960
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5293362729935173e-05,
+      "loss": 0.6779,
+      "step": 105970
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5290072065550035e-05,
+      "loss": 0.6501,
+      "step": 105980
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5286781401164896e-05,
+      "loss": 0.6841,
+      "step": 105990
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5283490736779757e-05,
+      "loss": 0.6455,
+      "step": 106000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.528020007239462e-05,
+      "loss": 0.6969,
+      "step": 106010
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5276909408009476e-05,
+      "loss": 0.6222,
+      "step": 106020
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5273618743624337e-05,
+      "loss": 0.666,
+      "step": 106030
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.52703280792392e-05,
+      "loss": 0.6786,
+      "step": 106040
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.526703741485406e-05,
+      "loss": 0.6208,
+      "step": 106050
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.526374675046892e-05,
+      "loss": 0.5827,
+      "step": 106060
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.526045608608378e-05,
+      "loss": 0.6616,
+      "step": 106070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.525716542169864e-05,
+      "loss": 0.5095,
+      "step": 106080
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5253874757313502e-05,
+      "loss": 0.648,
+      "step": 106090
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5250584092928363e-05,
+      "loss": 0.5888,
+      "step": 106100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5247293428543224e-05,
+      "loss": 0.6365,
+      "step": 106110
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5244002764158082e-05,
+      "loss": 0.6607,
+      "step": 106120
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5240712099772943e-05,
+      "loss": 0.5803,
+      "step": 106130
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5237421435387805e-05,
+      "loss": 0.5606,
+      "step": 106140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5234130771002666e-05,
+      "loss": 0.5804,
+      "step": 106150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5230840106617527e-05,
+      "loss": 0.6517,
+      "step": 106160
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5227549442232387e-05,
+      "loss": 0.6283,
+      "step": 106170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5224258777847248e-05,
+      "loss": 0.6938,
+      "step": 106180
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.522096811346211e-05,
+      "loss": 0.6138,
+      "step": 106190
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5217677449076969e-05,
+      "loss": 0.6047,
+      "step": 106200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.521438678469183e-05,
+      "loss": 0.6139,
+      "step": 106210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.521109612030669e-05,
+      "loss": 0.6761,
+      "step": 106220
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5207805455921551e-05,
+      "loss": 0.4652,
+      "step": 106230
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5204514791536412e-05,
+      "loss": 0.6102,
+      "step": 106240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5201224127151274e-05,
+      "loss": 0.6218,
+      "step": 106250
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5197933462766135e-05,
+      "loss": 0.6658,
+      "step": 106260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5194642798380993e-05,
+      "loss": 0.6647,
+      "step": 106270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5191352133995854e-05,
+      "loss": 0.5971,
+      "step": 106280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5188061469610715e-05,
+      "loss": 0.642,
+      "step": 106290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5184770805225576e-05,
+      "loss": 0.5413,
+      "step": 106300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5181480140840438e-05,
+      "loss": 0.5983,
+      "step": 106310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5178189476455299e-05,
+      "loss": 0.6499,
+      "step": 106320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5174898812070157e-05,
+      "loss": 0.5481,
+      "step": 106330
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5171608147685018e-05,
+      "loss": 0.7074,
+      "step": 106340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.516831748329988e-05,
+      "loss": 0.6264,
+      "step": 106350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.516502681891474e-05,
+      "loss": 0.5971,
+      "step": 106360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5161736154529602e-05,
+      "loss": 0.6523,
+      "step": 106370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.515844549014446e-05,
+      "loss": 0.6673,
+      "step": 106380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5155154825759321e-05,
+      "loss": 0.6572,
+      "step": 106390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5151864161374182e-05,
+      "loss": 0.6019,
+      "step": 106400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5148573496989044e-05,
+      "loss": 0.504,
+      "step": 106410
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5145282832603905e-05,
+      "loss": 0.5888,
+      "step": 106420
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5141992168218763e-05,
+      "loss": 0.6519,
+      "step": 106430
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5138701503833624e-05,
+      "loss": 0.6204,
+      "step": 106440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5135410839448485e-05,
+      "loss": 0.5067,
+      "step": 106450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5132120175063347e-05,
+      "loss": 0.6627,
+      "step": 106460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5128829510678208e-05,
+      "loss": 0.5367,
+      "step": 106470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5125538846293066e-05,
+      "loss": 0.6126,
+      "step": 106480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5122248181907927e-05,
+      "loss": 0.6112,
+      "step": 106490
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5118957517522788e-05,
+      "loss": 0.7184,
+      "step": 106500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.511566685313765e-05,
+      "loss": 0.5687,
+      "step": 106510
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.511237618875251e-05,
+      "loss": 0.5853,
+      "step": 106520
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.510908552436737e-05,
+      "loss": 0.5308,
+      "step": 106530
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.510579485998223e-05,
+      "loss": 0.6837,
+      "step": 106540
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5102504195597091e-05,
+      "loss": 0.6105,
+      "step": 106550
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5099213531211952e-05,
+      "loss": 0.6221,
+      "step": 106560
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5095922866826814e-05,
+      "loss": 0.707,
+      "step": 106570
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5092632202441673e-05,
+      "loss": 0.5596,
+      "step": 106580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5089341538056535e-05,
+      "loss": 0.6387,
+      "step": 106590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5086050873671396e-05,
+      "loss": 0.7362,
+      "step": 106600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5082760209286257e-05,
+      "loss": 0.4968,
+      "step": 106610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5079469544901117e-05,
+      "loss": 0.5953,
+      "step": 106620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5076178880515976e-05,
+      "loss": 0.6773,
+      "step": 106630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5072888216130838e-05,
+      "loss": 0.5673,
+      "step": 106640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5069597551745699e-05,
+      "loss": 0.6716,
+      "step": 106650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.506630688736056e-05,
+      "loss": 0.5342,
+      "step": 106660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5063016222975421e-05,
+      "loss": 0.6156,
+      "step": 106670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.505972555859028e-05,
+      "loss": 0.7236,
+      "step": 106680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.505643489420514e-05,
+      "loss": 0.7063,
+      "step": 106690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5053144229820002e-05,
+      "loss": 0.5708,
+      "step": 106700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5049853565434863e-05,
+      "loss": 0.6839,
+      "step": 106710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5046562901049724e-05,
+      "loss": 0.7135,
+      "step": 106720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5043272236664582e-05,
+      "loss": 0.6786,
+      "step": 106730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5039981572279443e-05,
+      "loss": 0.6896,
+      "step": 106740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5036690907894305e-05,
+      "loss": 0.6048,
+      "step": 106750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5033400243509166e-05,
+      "loss": 0.5825,
+      "step": 106760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5030109579124027e-05,
+      "loss": 0.6751,
+      "step": 106770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5026818914738885e-05,
+      "loss": 0.5283,
+      "step": 106780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5023528250353746e-05,
+      "loss": 0.6046,
+      "step": 106790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5020237585968608e-05,
+      "loss": 0.6593,
+      "step": 106800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5016946921583469e-05,
+      "loss": 0.6266,
+      "step": 106810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.501365625719833e-05,
+      "loss": 0.5445,
+      "step": 106820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5010365592813188e-05,
+      "loss": 0.7184,
+      "step": 106830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.500707492842805e-05,
+      "loss": 0.5068,
+      "step": 106840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.500378426404291e-05,
+      "loss": 0.5409,
+      "step": 106850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5000493599657772e-05,
+      "loss": 0.6305,
+      "step": 106860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4997202935272633e-05,
+      "loss": 0.6449,
+      "step": 106870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4993912270887491e-05,
+      "loss": 0.767,
+      "step": 106880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4990621606502352e-05,
+      "loss": 0.7384,
+      "step": 106890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4987330942117214e-05,
+      "loss": 0.6324,
+      "step": 106900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4984040277732075e-05,
+      "loss": 0.6185,
+      "step": 106910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4980749613346936e-05,
+      "loss": 0.5375,
+      "step": 106920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4977458948961796e-05,
+      "loss": 0.7713,
+      "step": 106930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4974168284576657e-05,
+      "loss": 0.5374,
+      "step": 106940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4970877620191518e-05,
+      "loss": 0.5524,
+      "step": 106950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4967586955806378e-05,
+      "loss": 0.8046,
+      "step": 106960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4964296291421239e-05,
+      "loss": 0.6435,
+      "step": 106970
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4961005627036099e-05,
+      "loss": 0.6278,
+      "step": 106980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.495771496265096e-05,
+      "loss": 0.576,
+      "step": 106990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4954424298265821e-05,
+      "loss": 0.4677,
+      "step": 107000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4951133633880682e-05,
+      "loss": 0.595,
+      "step": 107010
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4947842969495544e-05,
+      "loss": 0.6839,
+      "step": 107020
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4944552305110402e-05,
+      "loss": 0.628,
+      "step": 107030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4941261640725263e-05,
+      "loss": 0.5689,
+      "step": 107040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4937970976340124e-05,
+      "loss": 0.5607,
+      "step": 107050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4934680311954985e-05,
+      "loss": 0.6072,
+      "step": 107060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4931389647569847e-05,
+      "loss": 0.5672,
+      "step": 107070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4928098983184705e-05,
+      "loss": 0.6015,
+      "step": 107080
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4924808318799566e-05,
+      "loss": 0.6661,
+      "step": 107090
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4921517654414427e-05,
+      "loss": 0.713,
+      "step": 107100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4918226990029288e-05,
+      "loss": 0.5409,
+      "step": 107110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.491493632564415e-05,
+      "loss": 0.528,
+      "step": 107120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4911645661259007e-05,
+      "loss": 0.6287,
+      "step": 107130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4908354996873869e-05,
+      "loss": 0.6434,
+      "step": 107140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.490506433248873e-05,
+      "loss": 0.6112,
+      "step": 107150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4901773668103591e-05,
+      "loss": 0.6665,
+      "step": 107160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4898483003718453e-05,
+      "loss": 0.5951,
+      "step": 107170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.489519233933331e-05,
+      "loss": 0.667,
+      "step": 107180
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4891901674948172e-05,
+      "loss": 0.5396,
+      "step": 107190
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4888611010563033e-05,
+      "loss": 0.8024,
+      "step": 107200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4885320346177894e-05,
+      "loss": 0.6657,
+      "step": 107210
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4882029681792755e-05,
+      "loss": 0.4794,
+      "step": 107220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4878739017407613e-05,
+      "loss": 0.6309,
+      "step": 107230
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4875448353022475e-05,
+      "loss": 0.5824,
+      "step": 107240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4872157688637336e-05,
+      "loss": 0.5346,
+      "step": 107250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4868867024252197e-05,
+      "loss": 0.5938,
+      "step": 107260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4865576359867058e-05,
+      "loss": 0.6072,
+      "step": 107270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4862285695481918e-05,
+      "loss": 0.5784,
+      "step": 107280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.485899503109678e-05,
+      "loss": 0.5854,
+      "step": 107290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4855704366711639e-05,
+      "loss": 0.655,
+      "step": 107300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.48524137023265e-05,
+      "loss": 0.6488,
+      "step": 107310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4849123037941361e-05,
+      "loss": 0.5793,
+      "step": 107320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4845832373556221e-05,
+      "loss": 0.625,
+      "step": 107330
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4842541709171082e-05,
+      "loss": 0.7529,
+      "step": 107340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4839251044785943e-05,
+      "loss": 0.6488,
+      "step": 107350
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4835960380400805e-05,
+      "loss": 0.5857,
+      "step": 107360
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4832669716015666e-05,
+      "loss": 0.6474,
+      "step": 107370
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4829379051630524e-05,
+      "loss": 0.6691,
+      "step": 107380
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4826088387245385e-05,
+      "loss": 0.5757,
+      "step": 107390
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4822797722860246e-05,
+      "loss": 0.5475,
+      "step": 107400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4819507058475108e-05,
+      "loss": 0.6124,
+      "step": 107410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4816216394089969e-05,
+      "loss": 0.7058,
+      "step": 107420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4812925729704827e-05,
+      "loss": 0.5872,
+      "step": 107430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4809635065319688e-05,
+      "loss": 0.6787,
+      "step": 107440
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.480634440093455e-05,
+      "loss": 0.6488,
+      "step": 107450
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.480305373654941e-05,
+      "loss": 0.5995,
+      "step": 107460
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4799763072164272e-05,
+      "loss": 0.6746,
+      "step": 107470
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.479647240777913e-05,
+      "loss": 0.6091,
+      "step": 107480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4793181743393991e-05,
+      "loss": 0.5744,
+      "step": 107490
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4789891079008852e-05,
+      "loss": 0.5903,
+      "step": 107500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4786600414623714e-05,
+      "loss": 0.5873,
+      "step": 107510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4783309750238575e-05,
+      "loss": 0.5947,
+      "step": 107520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4780019085853433e-05,
+      "loss": 0.5457,
+      "step": 107530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4776728421468294e-05,
+      "loss": 0.5669,
+      "step": 107540
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4773437757083155e-05,
+      "loss": 0.5018,
+      "step": 107550
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4770147092698017e-05,
+      "loss": 0.6236,
+      "step": 107560
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4766856428312878e-05,
+      "loss": 0.6375,
+      "step": 107570
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4763565763927736e-05,
+      "loss": 0.545,
+      "step": 107580
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4760275099542597e-05,
+      "loss": 0.5927,
+      "step": 107590
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4756984435157458e-05,
+      "loss": 0.5987,
+      "step": 107600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.475369377077232e-05,
+      "loss": 0.6613,
+      "step": 107610
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.475040310638718e-05,
+      "loss": 0.7101,
+      "step": 107620
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.474711244200204e-05,
+      "loss": 0.6126,
+      "step": 107630
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4743821777616902e-05,
+      "loss": 0.7442,
+      "step": 107640
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4740531113231761e-05,
+      "loss": 0.6776,
+      "step": 107650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4737240448846622e-05,
+      "loss": 0.6023,
+      "step": 107660
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4733949784461484e-05,
+      "loss": 0.6973,
+      "step": 107670
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4730659120076343e-05,
+      "loss": 0.6868,
+      "step": 107680
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4727368455691205e-05,
+      "loss": 0.843,
+      "step": 107690
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4724077791306066e-05,
+      "loss": 0.6208,
+      "step": 107700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4720787126920927e-05,
+      "loss": 0.6396,
+      "step": 107710
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4717496462535788e-05,
+      "loss": 0.6246,
+      "step": 107720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4714205798150648e-05,
+      "loss": 0.7195,
+      "step": 107730
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4710915133765507e-05,
+      "loss": 0.6495,
+      "step": 107740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4707624469380369e-05,
+      "loss": 0.6821,
+      "step": 107750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.470433380499523e-05,
+      "loss": 0.7387,
+      "step": 107760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4701043140610091e-05,
+      "loss": 0.6023,
+      "step": 107770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4697752476224953e-05,
+      "loss": 0.4915,
+      "step": 107780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.469446181183981e-05,
+      "loss": 0.6942,
+      "step": 107790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4691171147454672e-05,
+      "loss": 0.6199,
+      "step": 107800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4687880483069533e-05,
+      "loss": 0.682,
+      "step": 107810
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4684589818684394e-05,
+      "loss": 0.6885,
+      "step": 107820
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4681299154299256e-05,
+      "loss": 0.6481,
+      "step": 107830
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4678008489914113e-05,
+      "loss": 0.6203,
+      "step": 107840
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4674717825528975e-05,
+      "loss": 0.6309,
+      "step": 107850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4671427161143836e-05,
+      "loss": 0.6209,
+      "step": 107860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4668136496758697e-05,
+      "loss": 0.5954,
+      "step": 107870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4664845832373558e-05,
+      "loss": 0.6242,
+      "step": 107880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4661555167988416e-05,
+      "loss": 0.5108,
+      "step": 107890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4658264503603278e-05,
+      "loss": 0.5393,
+      "step": 107900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4654973839218139e-05,
+      "loss": 0.4689,
+      "step": 107910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4651683174833e-05,
+      "loss": 0.6673,
+      "step": 107920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4648392510447861e-05,
+      "loss": 0.5632,
+      "step": 107930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.464510184606272e-05,
+      "loss": 0.5676,
+      "step": 107940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.464181118167758e-05,
+      "loss": 0.6331,
+      "step": 107950
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4638520517292442e-05,
+      "loss": 0.5867,
+      "step": 107960
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4635229852907303e-05,
+      "loss": 0.6357,
+      "step": 107970
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4631939188522164e-05,
+      "loss": 0.6808,
+      "step": 107980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4628648524137022e-05,
+      "loss": 0.6393,
+      "step": 107990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4625357859751884e-05,
+      "loss": 0.6031,
+      "step": 108000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4622067195366745e-05,
+      "loss": 0.6813,
+      "step": 108010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4618776530981606e-05,
+      "loss": 0.6004,
+      "step": 108020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4615485866596467e-05,
+      "loss": 0.5662,
+      "step": 108030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4612195202211327e-05,
+      "loss": 0.6139,
+      "step": 108040
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4608904537826188e-05,
+      "loss": 0.6107,
+      "step": 108050
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.460561387344105e-05,
+      "loss": 0.6642,
+      "step": 108060
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4602323209055909e-05,
+      "loss": 0.5458,
+      "step": 108070
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.459903254467077e-05,
+      "loss": 0.6788,
+      "step": 108080
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.459574188028563e-05,
+      "loss": 0.6003,
+      "step": 108090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4592451215900491e-05,
+      "loss": 0.5497,
+      "step": 108100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4589160551515352e-05,
+      "loss": 0.713,
+      "step": 108110
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4585869887130214e-05,
+      "loss": 0.6974,
+      "step": 108120
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4582579222745075e-05,
+      "loss": 0.5497,
+      "step": 108130
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4579288558359933e-05,
+      "loss": 0.678,
+      "step": 108140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4575997893974794e-05,
+      "loss": 0.5919,
+      "step": 108150
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4572707229589655e-05,
+      "loss": 0.5775,
+      "step": 108160
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4569416565204517e-05,
+      "loss": 0.6645,
+      "step": 108170
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4566125900819378e-05,
+      "loss": 0.6363,
+      "step": 108180
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4562835236434236e-05,
+      "loss": 0.6124,
+      "step": 108190
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4559544572049097e-05,
+      "loss": 0.5933,
+      "step": 108200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4556253907663958e-05,
+      "loss": 0.5345,
+      "step": 108210
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.455296324327882e-05,
+      "loss": 0.4977,
+      "step": 108220
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.454967257889368e-05,
+      "loss": 0.6447,
+      "step": 108230
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4546381914508539e-05,
+      "loss": 0.5523,
+      "step": 108240
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.45430912501234e-05,
+      "loss": 0.572,
+      "step": 108250
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4539800585738261e-05,
+      "loss": 0.6466,
+      "step": 108260
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4536509921353122e-05,
+      "loss": 0.528,
+      "step": 108270
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4533219256967984e-05,
+      "loss": 0.5649,
+      "step": 108280
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4529928592582842e-05,
+      "loss": 0.536,
+      "step": 108290
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4526637928197703e-05,
+      "loss": 0.5175,
+      "step": 108300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4523347263812564e-05,
+      "loss": 0.6269,
+      "step": 108310
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4520056599427425e-05,
+      "loss": 0.6073,
+      "step": 108320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4516765935042287e-05,
+      "loss": 0.7673,
+      "step": 108330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4513475270657145e-05,
+      "loss": 0.6256,
+      "step": 108340
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4510184606272006e-05,
+      "loss": 0.6503,
+      "step": 108350
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4506893941886867e-05,
+      "loss": 0.6321,
+      "step": 108360
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4503603277501728e-05,
+      "loss": 0.6417,
+      "step": 108370
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.450031261311659e-05,
+      "loss": 0.5386,
+      "step": 108380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.449702194873145e-05,
+      "loss": 0.6004,
+      "step": 108390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.449373128434631e-05,
+      "loss": 0.5614,
+      "step": 108400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.449044061996117e-05,
+      "loss": 0.527,
+      "step": 108410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4487149955576031e-05,
+      "loss": 0.5456,
+      "step": 108420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4483859291190893e-05,
+      "loss": 0.6611,
+      "step": 108430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4480568626805752e-05,
+      "loss": 0.5552,
+      "step": 108440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4477277962420613e-05,
+      "loss": 0.6199,
+      "step": 108450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4473987298035475e-05,
+      "loss": 0.5393,
+      "step": 108460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4470696633650336e-05,
+      "loss": 0.7704,
+      "step": 108470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4467405969265197e-05,
+      "loss": 0.7769,
+      "step": 108480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4464115304880055e-05,
+      "loss": 0.675,
+      "step": 108490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4460824640494916e-05,
+      "loss": 0.6109,
+      "step": 108500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4457533976109778e-05,
+      "loss": 0.6127,
+      "step": 108510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4454243311724639e-05,
+      "loss": 0.5678,
+      "step": 108520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.44509526473395e-05,
+      "loss": 0.5766,
+      "step": 108530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4447661982954358e-05,
+      "loss": 0.779,
+      "step": 108540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.444437131856922e-05,
+      "loss": 0.5822,
+      "step": 108550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.444108065418408e-05,
+      "loss": 0.7392,
+      "step": 108560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4437789989798942e-05,
+      "loss": 0.5216,
+      "step": 108570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4434499325413803e-05,
+      "loss": 0.6992,
+      "step": 108580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4431208661028661e-05,
+      "loss": 0.5974,
+      "step": 108590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4427917996643522e-05,
+      "loss": 0.5855,
+      "step": 108600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4424627332258384e-05,
+      "loss": 0.6436,
+      "step": 108610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4421336667873245e-05,
+      "loss": 0.5846,
+      "step": 108620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4418046003488106e-05,
+      "loss": 0.5495,
+      "step": 108630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4414755339102964e-05,
+      "loss": 0.7064,
+      "step": 108640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4411464674717825e-05,
+      "loss": 0.6365,
+      "step": 108650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4408174010332687e-05,
+      "loss": 0.5315,
+      "step": 108660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4404883345947548e-05,
+      "loss": 0.6512,
+      "step": 108670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4401592681562409e-05,
+      "loss": 0.5136,
+      "step": 108680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4398302017177267e-05,
+      "loss": 0.6602,
+      "step": 108690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4395011352792128e-05,
+      "loss": 0.6407,
+      "step": 108700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.439172068840699e-05,
+      "loss": 0.567,
+      "step": 108710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.438843002402185e-05,
+      "loss": 0.7118,
+      "step": 108720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4385139359636712e-05,
+      "loss": 0.6937,
+      "step": 108730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4381848695251572e-05,
+      "loss": 0.6237,
+      "step": 108740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4378558030866431e-05,
+      "loss": 0.6461,
+      "step": 108750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4375267366481292e-05,
+      "loss": 0.523,
+      "step": 108760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4371976702096154e-05,
+      "loss": 0.6688,
+      "step": 108770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4368686037711015e-05,
+      "loss": 0.5869,
+      "step": 108780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4365395373325875e-05,
+      "loss": 0.6677,
+      "step": 108790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4362104708940736e-05,
+      "loss": 0.5642,
+      "step": 108800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4358814044555597e-05,
+      "loss": 0.6493,
+      "step": 108810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4355523380170458e-05,
+      "loss": 0.5352,
+      "step": 108820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4352232715785318e-05,
+      "loss": 0.6053,
+      "step": 108830
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4348942051400177e-05,
+      "loss": 0.6554,
+      "step": 108840
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4345651387015039e-05,
+      "loss": 0.6324,
+      "step": 108850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.43423607226299e-05,
+      "loss": 0.5975,
+      "step": 108860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4339070058244761e-05,
+      "loss": 0.7074,
+      "step": 108870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4335779393859623e-05,
+      "loss": 0.6446,
+      "step": 108880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.433248872947448e-05,
+      "loss": 0.4926,
+      "step": 108890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4329198065089342e-05,
+      "loss": 0.7453,
+      "step": 108900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4325907400704203e-05,
+      "loss": 0.7113,
+      "step": 108910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4322616736319064e-05,
+      "loss": 0.6352,
+      "step": 108920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4319326071933925e-05,
+      "loss": 0.6833,
+      "step": 108930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4316035407548783e-05,
+      "loss": 0.5683,
+      "step": 108940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4312744743163645e-05,
+      "loss": 0.6808,
+      "step": 108950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4309454078778506e-05,
+      "loss": 0.6863,
+      "step": 108960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4306163414393367e-05,
+      "loss": 0.545,
+      "step": 108970
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4302872750008228e-05,
+      "loss": 0.4793,
+      "step": 108980
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4299582085623086e-05,
+      "loss": 0.5779,
+      "step": 108990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4296291421237948e-05,
+      "loss": 0.5953,
+      "step": 109000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4293000756852809e-05,
+      "loss": 0.5429,
+      "step": 109010
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.428971009246767e-05,
+      "loss": 0.6428,
+      "step": 109020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4286419428082531e-05,
+      "loss": 0.691,
+      "step": 109030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.428312876369739e-05,
+      "loss": 0.6162,
+      "step": 109040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.427983809931225e-05,
+      "loss": 0.6617,
+      "step": 109050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4276547434927112e-05,
+      "loss": 0.5542,
+      "step": 109060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4273256770541973e-05,
+      "loss": 0.6119,
+      "step": 109070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4269966106156834e-05,
+      "loss": 0.695,
+      "step": 109080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4266675441771692e-05,
+      "loss": 0.6649,
+      "step": 109090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4263384777386553e-05,
+      "loss": 0.5343,
+      "step": 109100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4260094113001415e-05,
+      "loss": 0.5948,
+      "step": 109110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4256803448616276e-05,
+      "loss": 0.745,
+      "step": 109120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4253512784231137e-05,
+      "loss": 0.6005,
+      "step": 109130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4250222119845999e-05,
+      "loss": 0.6674,
+      "step": 109140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4246931455460858e-05,
+      "loss": 0.6051,
+      "step": 109150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.424364079107572e-05,
+      "loss": 0.6233,
+      "step": 109160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4240350126690579e-05,
+      "loss": 0.6449,
+      "step": 109170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.423705946230544e-05,
+      "loss": 0.6668,
+      "step": 109180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4233768797920301e-05,
+      "loss": 0.6082,
+      "step": 109190
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4230478133535161e-05,
+      "loss": 0.5859,
+      "step": 109200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4227187469150022e-05,
+      "loss": 0.6918,
+      "step": 109210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4223896804764884e-05,
+      "loss": 0.7226,
+      "step": 109220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4220606140379745e-05,
+      "loss": 0.6278,
+      "step": 109230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4217315475994606e-05,
+      "loss": 0.6365,
+      "step": 109240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4214024811609464e-05,
+      "loss": 0.7514,
+      "step": 109250
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4210734147224325e-05,
+      "loss": 0.527,
+      "step": 109260
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4207443482839187e-05,
+      "loss": 0.6014,
+      "step": 109270
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4204152818454048e-05,
+      "loss": 0.5504,
+      "step": 109280
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4200862154068909e-05,
+      "loss": 0.545,
+      "step": 109290
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4197571489683767e-05,
+      "loss": 0.6932,
+      "step": 109300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4194280825298628e-05,
+      "loss": 0.5905,
+      "step": 109310
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.419099016091349e-05,
+      "loss": 0.6195,
+      "step": 109320
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.418769949652835e-05,
+      "loss": 0.6707,
+      "step": 109330
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4184408832143212e-05,
+      "loss": 0.6778,
+      "step": 109340
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.418111816775807e-05,
+      "loss": 0.5608,
+      "step": 109350
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4177827503372931e-05,
+      "loss": 0.5545,
+      "step": 109360
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4174536838987792e-05,
+      "loss": 0.5625,
+      "step": 109370
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4171246174602654e-05,
+      "loss": 0.6823,
+      "step": 109380
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4167955510217515e-05,
+      "loss": 0.521,
+      "step": 109390
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4164664845832373e-05,
+      "loss": 0.6659,
+      "step": 109400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4161374181447234e-05,
+      "loss": 0.6435,
+      "step": 109410
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4158083517062095e-05,
+      "loss": 0.6402,
+      "step": 109420
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4154792852676957e-05,
+      "loss": 0.6549,
+      "step": 109430
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4151502188291818e-05,
+      "loss": 0.5846,
+      "step": 109440
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4148211523906676e-05,
+      "loss": 0.6144,
+      "step": 109450
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4144920859521537e-05,
+      "loss": 0.6733,
+      "step": 109460
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4141630195136398e-05,
+      "loss": 0.5287,
+      "step": 109470
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.413833953075126e-05,
+      "loss": 0.6145,
+      "step": 109480
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4135048866366121e-05,
+      "loss": 0.471,
+      "step": 109490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.413175820198098e-05,
+      "loss": 0.5582,
+      "step": 109500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.412846753759584e-05,
+      "loss": 0.7074,
+      "step": 109510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4125176873210701e-05,
+      "loss": 0.4876,
+      "step": 109520
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4121886208825563e-05,
+      "loss": 0.6796,
+      "step": 109530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4118595544440424e-05,
+      "loss": 0.6655,
+      "step": 109540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4115304880055283e-05,
+      "loss": 0.6359,
+      "step": 109550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4112014215670145e-05,
+      "loss": 0.63,
+      "step": 109560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4108723551285006e-05,
+      "loss": 0.6295,
+      "step": 109570
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4105432886899867e-05,
+      "loss": 0.6486,
+      "step": 109580
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4102142222514727e-05,
+      "loss": 0.6666,
+      "step": 109590
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4098851558129586e-05,
+      "loss": 0.5956,
+      "step": 109600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4095560893744448e-05,
+      "loss": 0.5613,
+      "step": 109610
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4092270229359309e-05,
+      "loss": 0.5178,
+      "step": 109620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.408897956497417e-05,
+      "loss": 0.5365,
+      "step": 109630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4085688900589031e-05,
+      "loss": 0.7026,
+      "step": 109640
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.408239823620389e-05,
+      "loss": 0.615,
+      "step": 109650
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.407910757181875e-05,
+      "loss": 0.5502,
+      "step": 109660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4075816907433612e-05,
+      "loss": 0.7484,
+      "step": 109670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4072526243048473e-05,
+      "loss": 0.5472,
+      "step": 109680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4069235578663334e-05,
+      "loss": 0.7023,
+      "step": 109690
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4065944914278192e-05,
+      "loss": 0.6192,
+      "step": 109700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4062654249893054e-05,
+      "loss": 0.7426,
+      "step": 109710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4059363585507915e-05,
+      "loss": 0.7007,
+      "step": 109720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4056072921122776e-05,
+      "loss": 0.6215,
+      "step": 109730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4052782256737637e-05,
+      "loss": 0.7379,
+      "step": 109740
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4049491592352495e-05,
+      "loss": 0.6724,
+      "step": 109750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4046200927967356e-05,
+      "loss": 0.6384,
+      "step": 109760
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4042910263582218e-05,
+      "loss": 0.6321,
+      "step": 109770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4039619599197079e-05,
+      "loss": 0.5603,
+      "step": 109780
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.403632893481194e-05,
+      "loss": 0.542,
+      "step": 109790
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4033038270426798e-05,
+      "loss": 0.6402,
+      "step": 109800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.402974760604166e-05,
+      "loss": 0.6225,
+      "step": 109810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.402645694165652e-05,
+      "loss": 0.6242,
+      "step": 109820
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4023166277271382e-05,
+      "loss": 0.6065,
+      "step": 109830
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4019875612886243e-05,
+      "loss": 0.5682,
+      "step": 109840
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4016584948501101e-05,
+      "loss": 0.6535,
+      "step": 109850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4013294284115962e-05,
+      "loss": 0.5752,
+      "step": 109860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4010003619730824e-05,
+      "loss": 0.6251,
+      "step": 109870
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4006712955345685e-05,
+      "loss": 0.6026,
+      "step": 109880
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4003422290960546e-05,
+      "loss": 0.5791,
+      "step": 109890
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4000131626575406e-05,
+      "loss": 0.5493,
+      "step": 109900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3996840962190267e-05,
+      "loss": 0.5442,
+      "step": 109910
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3993550297805128e-05,
+      "loss": 0.6027,
+      "step": 109920
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3990259633419988e-05,
+      "loss": 0.6575,
+      "step": 109930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3986968969034849e-05,
+      "loss": 0.6069,
+      "step": 109940
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3983678304649709e-05,
+      "loss": 0.6416,
+      "step": 109950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.398038764026457e-05,
+      "loss": 0.6219,
+      "step": 109960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3977096975879431e-05,
+      "loss": 0.6539,
+      "step": 109970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3973806311494292e-05,
+      "loss": 0.6129,
+      "step": 109980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3970515647109154e-05,
+      "loss": 0.5628,
+      "step": 109990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3967224982724012e-05,
+      "loss": 0.693,
+      "step": 110000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3963934318338873e-05,
+      "loss": 0.6255,
+      "step": 110010
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3960643653953734e-05,
+      "loss": 0.6866,
+      "step": 110020
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3957352989568595e-05,
+      "loss": 0.5474,
+      "step": 110030
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3954062325183457e-05,
+      "loss": 0.6088,
+      "step": 110040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3950771660798315e-05,
+      "loss": 0.6889,
+      "step": 110050
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3947480996413176e-05,
+      "loss": 0.619,
+      "step": 110060
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3944190332028037e-05,
+      "loss": 0.6427,
+      "step": 110070
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3940899667642898e-05,
+      "loss": 0.7508,
+      "step": 110080
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.393760900325776e-05,
+      "loss": 0.6294,
+      "step": 110090
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3934318338872618e-05,
+      "loss": 0.6031,
+      "step": 110100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3931027674487479e-05,
+      "loss": 0.5989,
+      "step": 110110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.392773701010234e-05,
+      "loss": 0.7076,
+      "step": 110120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3924446345717201e-05,
+      "loss": 0.5876,
+      "step": 110130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3921155681332063e-05,
+      "loss": 0.7265,
+      "step": 110140
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.391786501694692e-05,
+      "loss": 0.7241,
+      "step": 110150
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3914574352561782e-05,
+      "loss": 0.5879,
+      "step": 110160
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3911283688176643e-05,
+      "loss": 0.6045,
+      "step": 110170
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3907993023791504e-05,
+      "loss": 0.6097,
+      "step": 110180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3904702359406366e-05,
+      "loss": 0.6225,
+      "step": 110190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3901411695021223e-05,
+      "loss": 0.7044,
+      "step": 110200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3898121030636085e-05,
+      "loss": 0.6749,
+      "step": 110210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3894830366250946e-05,
+      "loss": 0.6416,
+      "step": 110220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3891539701865807e-05,
+      "loss": 0.6041,
+      "step": 110230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3888249037480669e-05,
+      "loss": 0.5805,
+      "step": 110240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3884958373095528e-05,
+      "loss": 0.6985,
+      "step": 110250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.388166770871039e-05,
+      "loss": 0.5752,
+      "step": 110260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3878377044325249e-05,
+      "loss": 0.7314,
+      "step": 110270
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.387508637994011e-05,
+      "loss": 0.6452,
+      "step": 110280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3871795715554971e-05,
+      "loss": 0.6306,
+      "step": 110290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3868505051169831e-05,
+      "loss": 0.6029,
+      "step": 110300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3865214386784692e-05,
+      "loss": 0.5846,
+      "step": 110310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3861923722399554e-05,
+      "loss": 0.5311,
+      "step": 110320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3858633058014415e-05,
+      "loss": 0.5705,
+      "step": 110330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3855342393629276e-05,
+      "loss": 0.5635,
+      "step": 110340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3852051729244134e-05,
+      "loss": 0.7269,
+      "step": 110350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3848761064858995e-05,
+      "loss": 0.6134,
+      "step": 110360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3845470400473857e-05,
+      "loss": 0.6357,
+      "step": 110370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3842179736088718e-05,
+      "loss": 0.5159,
+      "step": 110380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3838889071703579e-05,
+      "loss": 0.595,
+      "step": 110390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3835598407318437e-05,
+      "loss": 0.4709,
+      "step": 110400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3832307742933298e-05,
+      "loss": 0.6512,
+      "step": 110410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.382901707854816e-05,
+      "loss": 0.655,
+      "step": 110420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.382572641416302e-05,
+      "loss": 0.5645,
+      "step": 110430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3822435749777882e-05,
+      "loss": 0.5619,
+      "step": 110440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.381914508539274e-05,
+      "loss": 0.6197,
+      "step": 110450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3815854421007601e-05,
+      "loss": 0.6685,
+      "step": 110460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3812563756622462e-05,
+      "loss": 0.6529,
+      "step": 110470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3809273092237324e-05,
+      "loss": 0.6172,
+      "step": 110480
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3805982427852185e-05,
+      "loss": 0.6699,
+      "step": 110490
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3802691763467043e-05,
+      "loss": 0.6508,
+      "step": 110500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3799401099081904e-05,
+      "loss": 0.5631,
+      "step": 110510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3796110434696765e-05,
+      "loss": 0.6443,
+      "step": 110520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3792819770311627e-05,
+      "loss": 0.8406,
+      "step": 110530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3789529105926488e-05,
+      "loss": 0.646,
+      "step": 110540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3786238441541346e-05,
+      "loss": 0.5631,
+      "step": 110550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3782947777156207e-05,
+      "loss": 0.5434,
+      "step": 110560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3779657112771068e-05,
+      "loss": 0.515,
+      "step": 110570
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.377636644838593e-05,
+      "loss": 0.5737,
+      "step": 110580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3773075784000791e-05,
+      "loss": 0.7187,
+      "step": 110590
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3769785119615652e-05,
+      "loss": 0.676,
+      "step": 110600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3766494455230512e-05,
+      "loss": 0.5512,
+      "step": 110610
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3763203790845371e-05,
+      "loss": 0.6665,
+      "step": 110620
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3759913126460233e-05,
+      "loss": 0.6071,
+      "step": 110630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3756622462075094e-05,
+      "loss": 0.7161,
+      "step": 110640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3753331797689955e-05,
+      "loss": 0.6019,
+      "step": 110650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3750041133304815e-05,
+      "loss": 0.6388,
+      "step": 110660
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3746750468919676e-05,
+      "loss": 0.6203,
+      "step": 110670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3743459804534537e-05,
+      "loss": 0.5445,
+      "step": 110680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3740169140149398e-05,
+      "loss": 0.6677,
+      "step": 110690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3736878475764258e-05,
+      "loss": 0.7374,
+      "step": 110700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3733587811379118e-05,
+      "loss": 0.5792,
+      "step": 110710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3730297146993979e-05,
+      "loss": 0.6878,
+      "step": 110720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.372700648260884e-05,
+      "loss": 0.6609,
+      "step": 110730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3723715818223701e-05,
+      "loss": 0.677,
+      "step": 110740
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3720425153838563e-05,
+      "loss": 0.5704,
+      "step": 110750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.371713448945342e-05,
+      "loss": 0.5138,
+      "step": 110760
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3713843825068282e-05,
+      "loss": 0.6285,
+      "step": 110770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3710553160683143e-05,
+      "loss": 0.6441,
+      "step": 110780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3707262496298004e-05,
+      "loss": 0.5305,
+      "step": 110790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3703971831912866e-05,
+      "loss": 0.506,
+      "step": 110800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3700681167527723e-05,
+      "loss": 0.6059,
+      "step": 110810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3697390503142585e-05,
+      "loss": 0.6173,
+      "step": 110820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3694099838757446e-05,
+      "loss": 0.5736,
+      "step": 110830
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3690809174372307e-05,
+      "loss": 0.5789,
+      "step": 110840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3687518509987169e-05,
+      "loss": 0.7957,
+      "step": 110850
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3684227845602026e-05,
+      "loss": 0.6067,
+      "step": 110860
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3680937181216888e-05,
+      "loss": 0.6559,
+      "step": 110870
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3677646516831749e-05,
+      "loss": 0.7269,
+      "step": 110880
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.367435585244661e-05,
+      "loss": 0.5471,
+      "step": 110890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3671065188061472e-05,
+      "loss": 0.6413,
+      "step": 110900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.366777452367633e-05,
+      "loss": 0.6489,
+      "step": 110910
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.366448385929119e-05,
+      "loss": 0.4946,
+      "step": 110920
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3661193194906052e-05,
+      "loss": 0.786,
+      "step": 110930
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3657902530520913e-05,
+      "loss": 0.6366,
+      "step": 110940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3654611866135774e-05,
+      "loss": 0.6014,
+      "step": 110950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3651321201750632e-05,
+      "loss": 0.6209,
+      "step": 110960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3648030537365494e-05,
+      "loss": 0.5678,
+      "step": 110970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3644739872980355e-05,
+      "loss": 0.6529,
+      "step": 110980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3641449208595216e-05,
+      "loss": 0.6825,
+      "step": 110990
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3638158544210077e-05,
+      "loss": 0.7272,
+      "step": 111000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3634867879824937e-05,
+      "loss": 0.627,
+      "step": 111010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3631577215439798e-05,
+      "loss": 0.6088,
+      "step": 111020
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.362828655105466e-05,
+      "loss": 0.6575,
+      "step": 111030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3624995886669519e-05,
+      "loss": 0.6789,
+      "step": 111040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.362170522228438e-05,
+      "loss": 0.5845,
+      "step": 111050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.361841455789924e-05,
+      "loss": 0.6029,
+      "step": 111060
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3615123893514101e-05,
+      "loss": 0.5499,
+      "step": 111070
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3611833229128962e-05,
+      "loss": 0.5977,
+      "step": 111080
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3608542564743824e-05,
+      "loss": 0.6104,
+      "step": 111090
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3605251900358685e-05,
+      "loss": 0.6164,
+      "step": 111100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3601961235973543e-05,
+      "loss": 0.6495,
+      "step": 111110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3598670571588404e-05,
+      "loss": 0.5968,
+      "step": 111120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3595379907203265e-05,
+      "loss": 0.6832,
+      "step": 111130
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3592089242818127e-05,
+      "loss": 0.546,
+      "step": 111140
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3588798578432988e-05,
+      "loss": 0.5881,
+      "step": 111150
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3585507914047846e-05,
+      "loss": 0.7229,
+      "step": 111160
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3582217249662707e-05,
+      "loss": 0.5377,
+      "step": 111170
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3578926585277568e-05,
+      "loss": 0.5694,
+      "step": 111180
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.357563592089243e-05,
+      "loss": 0.5205,
+      "step": 111190
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3572345256507291e-05,
+      "loss": 0.5973,
+      "step": 111200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3569054592122149e-05,
+      "loss": 0.6969,
+      "step": 111210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.356576392773701e-05,
+      "loss": 0.5726,
+      "step": 111220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3562473263351871e-05,
+      "loss": 0.6338,
+      "step": 111230
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3559182598966733e-05,
+      "loss": 0.6007,
+      "step": 111240
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3555891934581594e-05,
+      "loss": 0.5538,
+      "step": 111250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3552601270196452e-05,
+      "loss": 0.6353,
+      "step": 111260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3549310605811313e-05,
+      "loss": 0.6343,
+      "step": 111270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3546019941426174e-05,
+      "loss": 0.6477,
+      "step": 111280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3542729277041036e-05,
+      "loss": 0.6721,
+      "step": 111290
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3539438612655897e-05,
+      "loss": 0.6399,
+      "step": 111300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3536147948270755e-05,
+      "loss": 0.7591,
+      "step": 111310
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3532857283885616e-05,
+      "loss": 0.6994,
+      "step": 111320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3529566619500477e-05,
+      "loss": 0.6759,
+      "step": 111330
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3526275955115338e-05,
+      "loss": 0.5825,
+      "step": 111340
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.35229852907302e-05,
+      "loss": 0.5069,
+      "step": 111350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.351969462634506e-05,
+      "loss": 0.5751,
+      "step": 111360
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.351640396195992e-05,
+      "loss": 0.7514,
+      "step": 111370
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.351311329757478e-05,
+      "loss": 0.6662,
+      "step": 111380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3509822633189641e-05,
+      "loss": 0.5861,
+      "step": 111390
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3506531968804503e-05,
+      "loss": 0.637,
+      "step": 111400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3503241304419362e-05,
+      "loss": 0.6926,
+      "step": 111410
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3499950640034224e-05,
+      "loss": 0.5212,
+      "step": 111420
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3496659975649085e-05,
+      "loss": 0.5106,
+      "step": 111430
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3493369311263946e-05,
+      "loss": 0.5595,
+      "step": 111440
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3490078646878807e-05,
+      "loss": 0.6185,
+      "step": 111450
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3486787982493665e-05,
+      "loss": 0.6163,
+      "step": 111460
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3483497318108526e-05,
+      "loss": 0.4841,
+      "step": 111470
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3480206653723388e-05,
+      "loss": 0.5776,
+      "step": 111480
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3476915989338249e-05,
+      "loss": 0.6083,
+      "step": 111490
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.347362532495311e-05,
+      "loss": 0.5621,
+      "step": 111500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3470334660567968e-05,
+      "loss": 0.5508,
+      "step": 111510
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.346704399618283e-05,
+      "loss": 0.643,
+      "step": 111520
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.346375333179769e-05,
+      "loss": 0.5201,
+      "step": 111530
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3460462667412552e-05,
+      "loss": 0.6108,
+      "step": 111540
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3457172003027413e-05,
+      "loss": 0.6431,
+      "step": 111550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3453881338642271e-05,
+      "loss": 0.6266,
+      "step": 111560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3450590674257132e-05,
+      "loss": 0.6169,
+      "step": 111570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3447300009871994e-05,
+      "loss": 0.6065,
+      "step": 111580
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3444009345486855e-05,
+      "loss": 0.6329,
+      "step": 111590
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3440718681101716e-05,
+      "loss": 0.6418,
+      "step": 111600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3437428016716574e-05,
+      "loss": 0.5683,
+      "step": 111610
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3434137352331435e-05,
+      "loss": 0.497,
+      "step": 111620
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3430846687946297e-05,
+      "loss": 0.6611,
+      "step": 111630
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3427556023561158e-05,
+      "loss": 0.6169,
+      "step": 111640
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3424265359176019e-05,
+      "loss": 0.5874,
+      "step": 111650
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3420974694790877e-05,
+      "loss": 0.573,
+      "step": 111660
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3417684030405738e-05,
+      "loss": 0.8301,
+      "step": 111670
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.34143933660206e-05,
+      "loss": 0.6115,
+      "step": 111680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.341110270163546e-05,
+      "loss": 0.6616,
+      "step": 111690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3407812037250322e-05,
+      "loss": 0.6983,
+      "step": 111700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3404521372865182e-05,
+      "loss": 0.6552,
+      "step": 111710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3401230708480041e-05,
+      "loss": 0.5686,
+      "step": 111720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3397940044094902e-05,
+      "loss": 0.5896,
+      "step": 111730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3394649379709764e-05,
+      "loss": 0.5652,
+      "step": 111740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3391358715324625e-05,
+      "loss": 0.6421,
+      "step": 111750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3388068050939485e-05,
+      "loss": 0.6894,
+      "step": 111760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3384777386554346e-05,
+      "loss": 0.7828,
+      "step": 111770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3381486722169207e-05,
+      "loss": 0.7339,
+      "step": 111780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3378196057784068e-05,
+      "loss": 0.4906,
+      "step": 111790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3374905393398928e-05,
+      "loss": 0.5263,
+      "step": 111800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3371614729013788e-05,
+      "loss": 0.5871,
+      "step": 111810
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3368324064628649e-05,
+      "loss": 0.8017,
+      "step": 111820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.336503340024351e-05,
+      "loss": 0.5972,
+      "step": 111830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3361742735858371e-05,
+      "loss": 0.7354,
+      "step": 111840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3358452071473233e-05,
+      "loss": 0.5844,
+      "step": 111850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.335516140708809e-05,
+      "loss": 0.5694,
+      "step": 111860
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3351870742702952e-05,
+      "loss": 0.6545,
+      "step": 111870
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3348580078317813e-05,
+      "loss": 0.6039,
+      "step": 111880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3345289413932674e-05,
+      "loss": 0.6375,
+      "step": 111890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3341998749547536e-05,
+      "loss": 0.7388,
+      "step": 111900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3338708085162393e-05,
+      "loss": 0.5561,
+      "step": 111910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3335417420777255e-05,
+      "loss": 0.604,
+      "step": 111920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3332126756392116e-05,
+      "loss": 0.5034,
+      "step": 111930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3328836092006977e-05,
+      "loss": 0.5534,
+      "step": 111940
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3325545427621839e-05,
+      "loss": 0.5752,
+      "step": 111950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3322254763236696e-05,
+      "loss": 0.5216,
+      "step": 111960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3318964098851558e-05,
+      "loss": 0.6168,
+      "step": 111970
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3315673434466419e-05,
+      "loss": 0.6846,
+      "step": 111980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.331238277008128e-05,
+      "loss": 0.639,
+      "step": 111990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3309092105696141e-05,
+      "loss": 0.5868,
+      "step": 112000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3305801441311003e-05,
+      "loss": 0.5234,
+      "step": 112010
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.330251077692586e-05,
+      "loss": 0.5784,
+      "step": 112020
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3299220112540722e-05,
+      "loss": 0.7536,
+      "step": 112030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3295929448155583e-05,
+      "loss": 0.5812,
+      "step": 112040
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3292638783770444e-05,
+      "loss": 0.5587,
+      "step": 112050
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3289348119385306e-05,
+      "loss": 0.6512,
+      "step": 112060
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3286057455000164e-05,
+      "loss": 0.5928,
+      "step": 112070
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3282766790615025e-05,
+      "loss": 0.6222,
+      "step": 112080
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3279476126229886e-05,
+      "loss": 0.6263,
+      "step": 112090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3276185461844747e-05,
+      "loss": 0.5792,
+      "step": 112100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3272894797459609e-05,
+      "loss": 0.7267,
+      "step": 112110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3269604133074468e-05,
+      "loss": 0.6516,
+      "step": 112120
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.326631346868933e-05,
+      "loss": 0.6409,
+      "step": 112130
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3263022804304189e-05,
+      "loss": 0.5648,
+      "step": 112140
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.325973213991905e-05,
+      "loss": 0.64,
+      "step": 112150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3256441475533912e-05,
+      "loss": 0.8713,
+      "step": 112160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3253150811148771e-05,
+      "loss": 0.6709,
+      "step": 112170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3249860146763632e-05,
+      "loss": 0.7926,
+      "step": 112180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3246569482378494e-05,
+      "loss": 0.6371,
+      "step": 112190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3243278817993355e-05,
+      "loss": 0.708,
+      "step": 112200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3239988153608216e-05,
+      "loss": 0.5884,
+      "step": 112210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3236697489223074e-05,
+      "loss": 0.6534,
+      "step": 112220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3233406824837935e-05,
+      "loss": 0.6131,
+      "step": 112230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3230116160452797e-05,
+      "loss": 0.6364,
+      "step": 112240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3226825496067658e-05,
+      "loss": 0.5778,
+      "step": 112250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.322353483168252e-05,
+      "loss": 0.477,
+      "step": 112260
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3220244167297377e-05,
+      "loss": 0.7149,
+      "step": 112270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3216953502912238e-05,
+      "loss": 0.5613,
+      "step": 112280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.32136628385271e-05,
+      "loss": 0.7003,
+      "step": 112290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3210372174141961e-05,
+      "loss": 0.5429,
+      "step": 112300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3207081509756822e-05,
+      "loss": 0.7378,
+      "step": 112310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.320379084537168e-05,
+      "loss": 0.6089,
+      "step": 112320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3200500180986541e-05,
+      "loss": 0.6017,
+      "step": 112330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3197209516601403e-05,
+      "loss": 0.7725,
+      "step": 112340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3193918852216264e-05,
+      "loss": 0.7033,
+      "step": 112350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3190628187831125e-05,
+      "loss": 0.6928,
+      "step": 112360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3187337523445983e-05,
+      "loss": 0.6742,
+      "step": 112370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3184046859060844e-05,
+      "loss": 0.5725,
+      "step": 112380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3180756194675705e-05,
+      "loss": 0.6198,
+      "step": 112390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3177465530290567e-05,
+      "loss": 0.6016,
+      "step": 112400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3174174865905428e-05,
+      "loss": 0.579,
+      "step": 112410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3170884201520286e-05,
+      "loss": 0.5921,
+      "step": 112420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3167593537135147e-05,
+      "loss": 0.6469,
+      "step": 112430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3164302872750008e-05,
+      "loss": 0.5425,
+      "step": 112440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.316101220836487e-05,
+      "loss": 0.6831,
+      "step": 112450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3157721543979731e-05,
+      "loss": 0.532,
+      "step": 112460
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.315443087959459e-05,
+      "loss": 0.5027,
+      "step": 112470
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.315114021520945e-05,
+      "loss": 0.654,
+      "step": 112480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3147849550824311e-05,
+      "loss": 0.6497,
+      "step": 112490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3144558886439173e-05,
+      "loss": 0.5773,
+      "step": 112500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3141268222054034e-05,
+      "loss": 0.729,
+      "step": 112510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3137977557668893e-05,
+      "loss": 0.5943,
+      "step": 112520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3134686893283755e-05,
+      "loss": 0.5529,
+      "step": 112530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3131396228898616e-05,
+      "loss": 0.6024,
+      "step": 112540
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3128105564513477e-05,
+      "loss": 0.6308,
+      "step": 112550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3124814900128337e-05,
+      "loss": 0.6948,
+      "step": 112560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3121524235743196e-05,
+      "loss": 0.5716,
+      "step": 112570
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3118233571358058e-05,
+      "loss": 0.4974,
+      "step": 112580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3114942906972919e-05,
+      "loss": 0.6111,
+      "step": 112590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.311165224258778e-05,
+      "loss": 0.7586,
+      "step": 112600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3108361578202642e-05,
+      "loss": 0.5789,
+      "step": 112610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.31050709138175e-05,
+      "loss": 0.5785,
+      "step": 112620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.310178024943236e-05,
+      "loss": 0.5374,
+      "step": 112630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3098489585047222e-05,
+      "loss": 0.581,
+      "step": 112640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3095198920662083e-05,
+      "loss": 0.5089,
+      "step": 112650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3091908256276944e-05,
+      "loss": 0.7092,
+      "step": 112660
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3088617591891802e-05,
+      "loss": 0.5213,
+      "step": 112670
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3085326927506664e-05,
+      "loss": 0.6906,
+      "step": 112680
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3082036263121525e-05,
+      "loss": 0.6793,
+      "step": 112690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3078745598736386e-05,
+      "loss": 0.5667,
+      "step": 112700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3075454934351247e-05,
+      "loss": 0.6155,
+      "step": 112710
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3072164269966105e-05,
+      "loss": 0.6278,
+      "step": 112720
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3068873605580967e-05,
+      "loss": 0.6737,
+      "step": 112730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3065582941195828e-05,
+      "loss": 0.7448,
+      "step": 112740
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3062292276810689e-05,
+      "loss": 0.6315,
+      "step": 112750
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.305900161242555e-05,
+      "loss": 0.5982,
+      "step": 112760
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3055710948040408e-05,
+      "loss": 0.6212,
+      "step": 112770
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.305242028365527e-05,
+      "loss": 0.6487,
+      "step": 112780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.304912961927013e-05,
+      "loss": 0.6337,
+      "step": 112790
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3045838954884992e-05,
+      "loss": 0.6473,
+      "step": 112800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3042548290499853e-05,
+      "loss": 0.5783,
+      "step": 112810
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3039257626114711e-05,
+      "loss": 0.6002,
+      "step": 112820
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3035966961729572e-05,
+      "loss": 0.4789,
+      "step": 112830
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3032676297344434e-05,
+      "loss": 0.703,
+      "step": 112840
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3029385632959295e-05,
+      "loss": 0.511,
+      "step": 112850
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3026094968574156e-05,
+      "loss": 0.5362,
+      "step": 112860
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3022804304189016e-05,
+      "loss": 0.6014,
+      "step": 112870
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3019513639803877e-05,
+      "loss": 0.6997,
+      "step": 112880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3016222975418738e-05,
+      "loss": 0.693,
+      "step": 112890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3012932311033598e-05,
+      "loss": 0.6639,
+      "step": 112900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.300964164664846e-05,
+      "loss": 0.6888,
+      "step": 112910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3006350982263319e-05,
+      "loss": 0.6065,
+      "step": 112920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.300306031787818e-05,
+      "loss": 0.5453,
+      "step": 112930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2999769653493041e-05,
+      "loss": 0.6721,
+      "step": 112940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2996478989107903e-05,
+      "loss": 0.5564,
+      "step": 112950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2993188324722764e-05,
+      "loss": 0.5931,
+      "step": 112960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2989897660337622e-05,
+      "loss": 0.6092,
+      "step": 112970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2986606995952483e-05,
+      "loss": 0.4825,
+      "step": 112980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2983316331567344e-05,
+      "loss": 0.5394,
+      "step": 112990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2980025667182206e-05,
+      "loss": 0.6212,
+      "step": 113000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2976735002797067e-05,
+      "loss": 0.5932,
+      "step": 113010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2973444338411925e-05,
+      "loss": 0.5121,
+      "step": 113020
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2970153674026786e-05,
+      "loss": 0.5851,
+      "step": 113030
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2966863009641647e-05,
+      "loss": 0.5242,
+      "step": 113040
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2963572345256508e-05,
+      "loss": 0.5951,
+      "step": 113050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.296028168087137e-05,
+      "loss": 0.7146,
+      "step": 113060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2956991016486228e-05,
+      "loss": 0.5325,
+      "step": 113070
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2953700352101089e-05,
+      "loss": 0.647,
+      "step": 113080
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.295040968771595e-05,
+      "loss": 0.6513,
+      "step": 113090
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2947119023330811e-05,
+      "loss": 0.7113,
+      "step": 113100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2943828358945673e-05,
+      "loss": 0.5278,
+      "step": 113110
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.294053769456053e-05,
+      "loss": 0.6411,
+      "step": 113120
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2937247030175392e-05,
+      "loss": 0.6745,
+      "step": 113130
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2933956365790253e-05,
+      "loss": 0.782,
+      "step": 113140
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2930665701405114e-05,
+      "loss": 0.6174,
+      "step": 113150
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2927375037019976e-05,
+      "loss": 0.572,
+      "step": 113160
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2924084372634834e-05,
+      "loss": 0.6714,
+      "step": 113170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2920793708249695e-05,
+      "loss": 0.5693,
+      "step": 113180
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2917503043864556e-05,
+      "loss": 0.59,
+      "step": 113190
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2914212379479417e-05,
+      "loss": 0.5258,
+      "step": 113200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2910921715094279e-05,
+      "loss": 0.666,
+      "step": 113210
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2907631050709138e-05,
+      "loss": 0.5333,
+      "step": 113220
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2904340386324e-05,
+      "loss": 0.5947,
+      "step": 113230
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2901049721938859e-05,
+      "loss": 0.5301,
+      "step": 113240
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.289775905755372e-05,
+      "loss": 0.5506,
+      "step": 113250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2894468393168582e-05,
+      "loss": 0.5468,
+      "step": 113260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2891177728783441e-05,
+      "loss": 0.6167,
+      "step": 113270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2887887064398302e-05,
+      "loss": 0.6753,
+      "step": 113280
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2884596400013164e-05,
+      "loss": 0.6898,
+      "step": 113290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2881305735628025e-05,
+      "loss": 0.5816,
+      "step": 113300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2878015071242886e-05,
+      "loss": 0.6344,
+      "step": 113310
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2874724406857744e-05,
+      "loss": 0.7021,
+      "step": 113320
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2871433742472605e-05,
+      "loss": 0.6643,
+      "step": 113330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2868143078087467e-05,
+      "loss": 0.654,
+      "step": 113340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2864852413702328e-05,
+      "loss": 0.5425,
+      "step": 113350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2861561749317189e-05,
+      "loss": 0.6153,
+      "step": 113360
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2858271084932047e-05,
+      "loss": 0.5325,
+      "step": 113370
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2854980420546908e-05,
+      "loss": 0.6325,
+      "step": 113380
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.285168975616177e-05,
+      "loss": 0.6741,
+      "step": 113390
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.284839909177663e-05,
+      "loss": 0.5605,
+      "step": 113400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2845108427391492e-05,
+      "loss": 0.6659,
+      "step": 113410
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2841817763006353e-05,
+      "loss": 0.5799,
+      "step": 113420
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2838527098621211e-05,
+      "loss": 0.6345,
+      "step": 113430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2835236434236073e-05,
+      "loss": 0.6449,
+      "step": 113440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2831945769850934e-05,
+      "loss": 0.6063,
+      "step": 113450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2828655105465795e-05,
+      "loss": 0.589,
+      "step": 113460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2825364441080656e-05,
+      "loss": 0.5039,
+      "step": 113470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2822073776695514e-05,
+      "loss": 0.5694,
+      "step": 113480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2818783112310375e-05,
+      "loss": 0.545,
+      "step": 113490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2815492447925237e-05,
+      "loss": 0.715,
+      "step": 113500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2812201783540098e-05,
+      "loss": 0.7445,
+      "step": 113510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.280891111915496e-05,
+      "loss": 0.7959,
+      "step": 113520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2805620454769817e-05,
+      "loss": 0.676,
+      "step": 113530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2802329790384678e-05,
+      "loss": 0.671,
+      "step": 113540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.279903912599954e-05,
+      "loss": 0.5916,
+      "step": 113550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2795748461614401e-05,
+      "loss": 0.6574,
+      "step": 113560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2792457797229262e-05,
+      "loss": 0.7065,
+      "step": 113570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2789167132844122e-05,
+      "loss": 0.6054,
+      "step": 113580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2785876468458981e-05,
+      "loss": 0.7333,
+      "step": 113590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2782585804073843e-05,
+      "loss": 0.5755,
+      "step": 113600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2779295139688704e-05,
+      "loss": 0.5814,
+      "step": 113610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2776004475303565e-05,
+      "loss": 0.4811,
+      "step": 113620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2772713810918425e-05,
+      "loss": 0.6427,
+      "step": 113630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2769423146533286e-05,
+      "loss": 0.6327,
+      "step": 113640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2766132482148147e-05,
+      "loss": 0.5975,
+      "step": 113650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2762841817763009e-05,
+      "loss": 0.648,
+      "step": 113660
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2759551153377868e-05,
+      "loss": 0.6929,
+      "step": 113670
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2756260488992728e-05,
+      "loss": 0.7022,
+      "step": 113680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2752969824607589e-05,
+      "loss": 0.578,
+      "step": 113690
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.274967916022245e-05,
+      "loss": 0.7357,
+      "step": 113700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2746388495837311e-05,
+      "loss": 0.8557,
+      "step": 113710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2743097831452173e-05,
+      "loss": 0.74,
+      "step": 113720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.273980716706703e-05,
+      "loss": 0.6428,
+      "step": 113730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2736516502681892e-05,
+      "loss": 0.6093,
+      "step": 113740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2733225838296753e-05,
+      "loss": 0.7518,
+      "step": 113750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2729935173911614e-05,
+      "loss": 0.7348,
+      "step": 113760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2726644509526476e-05,
+      "loss": 0.6673,
+      "step": 113770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2723353845141334e-05,
+      "loss": 0.5879,
+      "step": 113780
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2720063180756195e-05,
+      "loss": 0.6668,
+      "step": 113790
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2716772516371056e-05,
+      "loss": 0.6176,
+      "step": 113800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2713481851985917e-05,
+      "loss": 0.6797,
+      "step": 113810
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2710191187600779e-05,
+      "loss": 0.6064,
+      "step": 113820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2706900523215637e-05,
+      "loss": 0.6138,
+      "step": 113830
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2703609858830498e-05,
+      "loss": 0.6577,
+      "step": 113840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2700319194445359e-05,
+      "loss": 0.5099,
+      "step": 113850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.269702853006022e-05,
+      "loss": 0.6281,
+      "step": 113860
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2693737865675082e-05,
+      "loss": 0.5114,
+      "step": 113870
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.269044720128994e-05,
+      "loss": 0.6586,
+      "step": 113880
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.26871565369048e-05,
+      "loss": 0.6138,
+      "step": 113890
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2683865872519662e-05,
+      "loss": 0.6028,
+      "step": 113900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2680575208134523e-05,
+      "loss": 0.6277,
+      "step": 113910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2677284543749385e-05,
+      "loss": 0.5482,
+      "step": 113920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2673993879364242e-05,
+      "loss": 0.5981,
+      "step": 113930
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2670703214979104e-05,
+      "loss": 0.607,
+      "step": 113940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2667412550593965e-05,
+      "loss": 0.871,
+      "step": 113950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2664121886208826e-05,
+      "loss": 0.6012,
+      "step": 113960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2660831221823687e-05,
+      "loss": 0.6136,
+      "step": 113970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2657540557438547e-05,
+      "loss": 0.5495,
+      "step": 113980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2654249893053408e-05,
+      "loss": 0.6481,
+      "step": 113990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.265095922866827e-05,
+      "loss": 0.5562,
+      "step": 114000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.264766856428313e-05,
+      "loss": 0.7117,
+      "step": 114010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.264437789989799e-05,
+      "loss": 0.5334,
+      "step": 114020
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.264108723551285e-05,
+      "loss": 0.6393,
+      "step": 114030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2637796571127711e-05,
+      "loss": 0.6722,
+      "step": 114040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2634505906742573e-05,
+      "loss": 0.6501,
+      "step": 114050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2631215242357434e-05,
+      "loss": 0.6374,
+      "step": 114060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2627924577972295e-05,
+      "loss": 0.5745,
+      "step": 114070
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2624633913587153e-05,
+      "loss": 0.6001,
+      "step": 114080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2621343249202014e-05,
+      "loss": 0.4902,
+      "step": 114090
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2618052584816875e-05,
+      "loss": 0.5067,
+      "step": 114100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2614761920431737e-05,
+      "loss": 0.508,
+      "step": 114110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2611471256046598e-05,
+      "loss": 0.5945,
+      "step": 114120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2608180591661456e-05,
+      "loss": 0.6753,
+      "step": 114130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2604889927276317e-05,
+      "loss": 0.5618,
+      "step": 114140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2601599262891178e-05,
+      "loss": 0.5453,
+      "step": 114150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.259830859850604e-05,
+      "loss": 0.5979,
+      "step": 114160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2595017934120901e-05,
+      "loss": 0.4532,
+      "step": 114170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2591727269735759e-05,
+      "loss": 0.605,
+      "step": 114180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.258843660535062e-05,
+      "loss": 0.5326,
+      "step": 114190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2585145940965481e-05,
+      "loss": 0.5833,
+      "step": 114200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2581855276580343e-05,
+      "loss": 0.6023,
+      "step": 114210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2578564612195204e-05,
+      "loss": 0.7187,
+      "step": 114220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2575273947810062e-05,
+      "loss": 0.6607,
+      "step": 114230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2571983283424923e-05,
+      "loss": 0.5693,
+      "step": 114240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2568692619039784e-05,
+      "loss": 0.686,
+      "step": 114250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2565401954654646e-05,
+      "loss": 0.631,
+      "step": 114260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2562111290269507e-05,
+      "loss": 0.6134,
+      "step": 114270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2558820625884365e-05,
+      "loss": 0.6325,
+      "step": 114280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2555529961499226e-05,
+      "loss": 0.6729,
+      "step": 114290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2552239297114087e-05,
+      "loss": 0.6853,
+      "step": 114300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2548948632728949e-05,
+      "loss": 0.5889,
+      "step": 114310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.254565796834381e-05,
+      "loss": 0.5332,
+      "step": 114320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.254236730395867e-05,
+      "loss": 0.6645,
+      "step": 114330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.253907663957353e-05,
+      "loss": 0.6146,
+      "step": 114340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.253578597518839e-05,
+      "loss": 0.6711,
+      "step": 114350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2532495310803252e-05,
+      "loss": 0.5074,
+      "step": 114360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2529204646418113e-05,
+      "loss": 0.6895,
+      "step": 114370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2525913982032972e-05,
+      "loss": 0.6265,
+      "step": 114380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2522623317647834e-05,
+      "loss": 0.524,
+      "step": 114390
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2519332653262695e-05,
+      "loss": 0.619,
+      "step": 114400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2516041988877556e-05,
+      "loss": 0.5819,
+      "step": 114410
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2512751324492417e-05,
+      "loss": 0.6127,
+      "step": 114420
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2509460660107275e-05,
+      "loss": 0.7815,
+      "step": 114430
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2506169995722137e-05,
+      "loss": 0.581,
+      "step": 114440
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2502879331336998e-05,
+      "loss": 0.5219,
+      "step": 114450
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2499588666951859e-05,
+      "loss": 0.8813,
+      "step": 114460
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2496298002566719e-05,
+      "loss": 0.6381,
+      "step": 114470
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.249300733818158e-05,
+      "loss": 0.5923,
+      "step": 114480
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2489716673796441e-05,
+      "loss": 0.5843,
+      "step": 114490
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.24864260094113e-05,
+      "loss": 0.5931,
+      "step": 114500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2483135345026162e-05,
+      "loss": 0.5629,
+      "step": 114510
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2479844680641022e-05,
+      "loss": 0.7616,
+      "step": 114520
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2476554016255883e-05,
+      "loss": 0.805,
+      "step": 114530
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2473263351870744e-05,
+      "loss": 0.6237,
+      "step": 114540
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2469972687485604e-05,
+      "loss": 0.6957,
+      "step": 114550
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2466682023100465e-05,
+      "loss": 0.6348,
+      "step": 114560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2463391358715325e-05,
+      "loss": 0.6378,
+      "step": 114570
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2460100694330186e-05,
+      "loss": 0.6977,
+      "step": 114580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2456810029945047e-05,
+      "loss": 0.6545,
+      "step": 114590
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2453519365559907e-05,
+      "loss": 0.5416,
+      "step": 114600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2450228701174768e-05,
+      "loss": 0.5974,
+      "step": 114610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2446938036789628e-05,
+      "loss": 0.518,
+      "step": 114620
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2443647372404489e-05,
+      "loss": 0.5796,
+      "step": 114630
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.244035670801935e-05,
+      "loss": 0.6519,
+      "step": 114640
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.243706604363421e-05,
+      "loss": 0.5679,
+      "step": 114650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2433775379249071e-05,
+      "loss": 0.7903,
+      "step": 114660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.243048471486393e-05,
+      "loss": 0.5288,
+      "step": 114670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2427194050478792e-05,
+      "loss": 0.6491,
+      "step": 114680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2423903386093653e-05,
+      "loss": 0.6823,
+      "step": 114690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2420612721708513e-05,
+      "loss": 0.577,
+      "step": 114700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2417322057323374e-05,
+      "loss": 0.5662,
+      "step": 114710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2414031392938235e-05,
+      "loss": 0.5907,
+      "step": 114720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2410740728553095e-05,
+      "loss": 0.6554,
+      "step": 114730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2407450064167956e-05,
+      "loss": 0.563,
+      "step": 114740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2404159399782817e-05,
+      "loss": 0.4882,
+      "step": 114750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2400868735397678e-05,
+      "loss": 0.4805,
+      "step": 114760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2397578071012538e-05,
+      "loss": 0.6289,
+      "step": 114770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.23942874066274e-05,
+      "loss": 0.6267,
+      "step": 114780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.239099674224226e-05,
+      "loss": 0.5155,
+      "step": 114790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.238770607785712e-05,
+      "loss": 0.5546,
+      "step": 114800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2384415413471981e-05,
+      "loss": 0.5363,
+      "step": 114810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2381124749086841e-05,
+      "loss": 0.5877,
+      "step": 114820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2377834084701702e-05,
+      "loss": 0.6345,
+      "step": 114830
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2374543420316564e-05,
+      "loss": 0.626,
+      "step": 114840
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2371252755931423e-05,
+      "loss": 0.567,
+      "step": 114850
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2367962091546284e-05,
+      "loss": 0.5343,
+      "step": 114860
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2364671427161144e-05,
+      "loss": 0.7325,
+      "step": 114870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2361380762776005e-05,
+      "loss": 0.5889,
+      "step": 114880
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2358090098390867e-05,
+      "loss": 0.5979,
+      "step": 114890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2354799434005726e-05,
+      "loss": 0.6908,
+      "step": 114900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2351508769620587e-05,
+      "loss": 0.6037,
+      "step": 114910
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2348218105235447e-05,
+      "loss": 0.6469,
+      "step": 114920
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2344927440850308e-05,
+      "loss": 0.6685,
+      "step": 114930
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.234163677646517e-05,
+      "loss": 0.5892,
+      "step": 114940
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2338346112080029e-05,
+      "loss": 0.6062,
+      "step": 114950
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.233505544769489e-05,
+      "loss": 0.665,
+      "step": 114960
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.233176478330975e-05,
+      "loss": 0.6831,
+      "step": 114970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2328474118924611e-05,
+      "loss": 0.6764,
+      "step": 114980
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2325183454539472e-05,
+      "loss": 0.5549,
+      "step": 114990
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2321892790154332e-05,
+      "loss": 0.6168,
+      "step": 115000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2318602125769193e-05,
+      "loss": 0.5821,
+      "step": 115010
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2315311461384053e-05,
+      "loss": 0.6075,
+      "step": 115020
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2312020796998914e-05,
+      "loss": 0.6645,
+      "step": 115030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2308730132613775e-05,
+      "loss": 0.5916,
+      "step": 115040
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2305439468228635e-05,
+      "loss": 0.6428,
+      "step": 115050
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2302148803843496e-05,
+      "loss": 0.715,
+      "step": 115060
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2298858139458356e-05,
+      "loss": 0.6149,
+      "step": 115070
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2295567475073217e-05,
+      "loss": 0.634,
+      "step": 115080
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2292276810688078e-05,
+      "loss": 0.7139,
+      "step": 115090
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.228898614630294e-05,
+      "loss": 0.5713,
+      "step": 115100
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2285695481917799e-05,
+      "loss": 0.5169,
+      "step": 115110
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.228240481753266e-05,
+      "loss": 0.6809,
+      "step": 115120
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2279114153147522e-05,
+      "loss": 0.6676,
+      "step": 115130
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2275823488762383e-05,
+      "loss": 0.6587,
+      "step": 115140
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2272532824377243e-05,
+      "loss": 0.6888,
+      "step": 115150
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2269242159992104e-05,
+      "loss": 0.5504,
+      "step": 115160
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2265951495606965e-05,
+      "loss": 0.5338,
+      "step": 115170
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2262660831221825e-05,
+      "loss": 0.754,
+      "step": 115180
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2259370166836686e-05,
+      "loss": 0.6076,
+      "step": 115190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2256079502451545e-05,
+      "loss": 0.6602,
+      "step": 115200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2252788838066407e-05,
+      "loss": 0.5869,
+      "step": 115210
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2249498173681268e-05,
+      "loss": 0.6216,
+      "step": 115220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2246207509296128e-05,
+      "loss": 0.7138,
+      "step": 115230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2242916844910989e-05,
+      "loss": 0.6601,
+      "step": 115240
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2239626180525848e-05,
+      "loss": 0.6542,
+      "step": 115250
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.223633551614071e-05,
+      "loss": 0.6134,
+      "step": 115260
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2233044851755571e-05,
+      "loss": 0.6149,
+      "step": 115270
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.222975418737043e-05,
+      "loss": 0.5197,
+      "step": 115280
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2226463522985292e-05,
+      "loss": 0.6158,
+      "step": 115290
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2223172858600151e-05,
+      "loss": 0.6768,
+      "step": 115300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2219882194215013e-05,
+      "loss": 0.5831,
+      "step": 115310
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2216591529829874e-05,
+      "loss": 0.5937,
+      "step": 115320
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2213300865444733e-05,
+      "loss": 0.5672,
+      "step": 115330
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2210010201059595e-05,
+      "loss": 0.5999,
+      "step": 115340
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2206719536674454e-05,
+      "loss": 0.5281,
+      "step": 115350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2203428872289316e-05,
+      "loss": 0.6362,
+      "step": 115360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2200138207904177e-05,
+      "loss": 0.4748,
+      "step": 115370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2196847543519036e-05,
+      "loss": 0.6755,
+      "step": 115380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2193556879133898e-05,
+      "loss": 0.5473,
+      "step": 115390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2190266214748757e-05,
+      "loss": 0.5235,
+      "step": 115400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2186975550363619e-05,
+      "loss": 0.6053,
+      "step": 115410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.218368488597848e-05,
+      "loss": 0.6008,
+      "step": 115420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.218039422159334e-05,
+      "loss": 0.6766,
+      "step": 115430
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.21771035572082e-05,
+      "loss": 0.6843,
+      "step": 115440
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.217381289282306e-05,
+      "loss": 0.6498,
+      "step": 115450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2170522228437921e-05,
+      "loss": 0.5504,
+      "step": 115460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2167231564052783e-05,
+      "loss": 0.6135,
+      "step": 115470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2163940899667644e-05,
+      "loss": 0.5765,
+      "step": 115480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2160650235282504e-05,
+      "loss": 0.6137,
+      "step": 115490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2157359570897365e-05,
+      "loss": 0.7461,
+      "step": 115500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2154068906512226e-05,
+      "loss": 0.5906,
+      "step": 115510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2150778242127087e-05,
+      "loss": 0.646,
+      "step": 115520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2147487577741947e-05,
+      "loss": 0.6036,
+      "step": 115530
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2144196913356808e-05,
+      "loss": 0.6832,
+      "step": 115540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2140906248971668e-05,
+      "loss": 0.645,
+      "step": 115550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2137615584586529e-05,
+      "loss": 0.525,
+      "step": 115560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.213432492020139e-05,
+      "loss": 0.5743,
+      "step": 115570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.213103425581625e-05,
+      "loss": 0.672,
+      "step": 115580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2127743591431111e-05,
+      "loss": 0.5995,
+      "step": 115590
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.212445292704597e-05,
+      "loss": 0.6273,
+      "step": 115600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2121162262660832e-05,
+      "loss": 0.6456,
+      "step": 115610
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2117871598275693e-05,
+      "loss": 0.588,
+      "step": 115620
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2114580933890553e-05,
+      "loss": 0.6208,
+      "step": 115630
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2111290269505414e-05,
+      "loss": 0.5937,
+      "step": 115640
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2107999605120274e-05,
+      "loss": 0.6518,
+      "step": 115650
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2104708940735135e-05,
+      "loss": 0.6141,
+      "step": 115660
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2101418276349996e-05,
+      "loss": 0.6069,
+      "step": 115670
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2098127611964856e-05,
+      "loss": 0.7575,
+      "step": 115680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2094836947579717e-05,
+      "loss": 0.5213,
+      "step": 115690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2091546283194577e-05,
+      "loss": 0.6397,
+      "step": 115700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2088255618809438e-05,
+      "loss": 0.6309,
+      "step": 115710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.20849649544243e-05,
+      "loss": 0.6252,
+      "step": 115720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2081674290039159e-05,
+      "loss": 0.5315,
+      "step": 115730
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.207838362565402e-05,
+      "loss": 0.6012,
+      "step": 115740
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.207509296126888e-05,
+      "loss": 0.6454,
+      "step": 115750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2071802296883741e-05,
+      "loss": 0.655,
+      "step": 115760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2068511632498602e-05,
+      "loss": 0.6825,
+      "step": 115770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2065220968113462e-05,
+      "loss": 0.6137,
+      "step": 115780
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2061930303728323e-05,
+      "loss": 0.6146,
+      "step": 115790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2058639639343183e-05,
+      "loss": 0.8435,
+      "step": 115800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2055348974958044e-05,
+      "loss": 0.6721,
+      "step": 115810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2052058310572905e-05,
+      "loss": 0.6633,
+      "step": 115820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2048767646187765e-05,
+      "loss": 0.6324,
+      "step": 115830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2045476981802626e-05,
+      "loss": 0.6644,
+      "step": 115840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2042186317417487e-05,
+      "loss": 0.6568,
+      "step": 115850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2038895653032348e-05,
+      "loss": 0.6906,
+      "step": 115860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2035604988647208e-05,
+      "loss": 0.7064,
+      "step": 115870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.203231432426207e-05,
+      "loss": 0.5727,
+      "step": 115880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.202902365987693e-05,
+      "loss": 0.5865,
+      "step": 115890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2025732995491792e-05,
+      "loss": 0.64,
+      "step": 115900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2022442331106651e-05,
+      "loss": 0.6369,
+      "step": 115910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2019151666721513e-05,
+      "loss": 0.6307,
+      "step": 115920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2015861002336372e-05,
+      "loss": 0.6211,
+      "step": 115930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2012570337951234e-05,
+      "loss": 0.6287,
+      "step": 115940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2009279673566095e-05,
+      "loss": 0.6109,
+      "step": 115950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2005989009180954e-05,
+      "loss": 0.5673,
+      "step": 115960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2002698344795816e-05,
+      "loss": 0.6331,
+      "step": 115970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1999407680410675e-05,
+      "loss": 0.5089,
+      "step": 115980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1996117016025536e-05,
+      "loss": 0.5196,
+      "step": 115990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1992826351640398e-05,
+      "loss": 0.6145,
+      "step": 116000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1989535687255257e-05,
+      "loss": 0.6471,
+      "step": 116010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1986245022870119e-05,
+      "loss": 0.59,
+      "step": 116020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1982954358484978e-05,
+      "loss": 0.5295,
+      "step": 116030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.197966369409984e-05,
+      "loss": 0.6184,
+      "step": 116040
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.19763730297147e-05,
+      "loss": 0.593,
+      "step": 116050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.197308236532956e-05,
+      "loss": 0.5615,
+      "step": 116060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1969791700944422e-05,
+      "loss": 0.7247,
+      "step": 116070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1966501036559281e-05,
+      "loss": 0.6441,
+      "step": 116080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1963210372174142e-05,
+      "loss": 0.5939,
+      "step": 116090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1959919707789004e-05,
+      "loss": 0.5887,
+      "step": 116100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1956629043403863e-05,
+      "loss": 0.5293,
+      "step": 116110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1953338379018724e-05,
+      "loss": 0.6012,
+      "step": 116120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1950047714633584e-05,
+      "loss": 0.6019,
+      "step": 116130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1946757050248445e-05,
+      "loss": 0.7598,
+      "step": 116140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1943466385863307e-05,
+      "loss": 0.6562,
+      "step": 116150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1940175721478166e-05,
+      "loss": 0.6723,
+      "step": 116160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1936885057093027e-05,
+      "loss": 0.6011,
+      "step": 116170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1933594392707887e-05,
+      "loss": 0.751,
+      "step": 116180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1930303728322748e-05,
+      "loss": 0.6068,
+      "step": 116190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.192701306393761e-05,
+      "loss": 0.697,
+      "step": 116200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1923722399552469e-05,
+      "loss": 0.6081,
+      "step": 116210
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.192043173516733e-05,
+      "loss": 0.653,
+      "step": 116220
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1917141070782192e-05,
+      "loss": 0.6138,
+      "step": 116230
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1913850406397053e-05,
+      "loss": 0.6903,
+      "step": 116240
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1910559742011914e-05,
+      "loss": 0.6265,
+      "step": 116250
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1907269077626774e-05,
+      "loss": 0.6549,
+      "step": 116260
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1903978413241635e-05,
+      "loss": 0.744,
+      "step": 116270
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1900687748856495e-05,
+      "loss": 0.549,
+      "step": 116280
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1897397084471356e-05,
+      "loss": 0.51,
+      "step": 116290
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1894106420086217e-05,
+      "loss": 0.7077,
+      "step": 116300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1890815755701077e-05,
+      "loss": 0.592,
+      "step": 116310
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1887525091315938e-05,
+      "loss": 0.7105,
+      "step": 116320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1884234426930798e-05,
+      "loss": 0.5969,
+      "step": 116330
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1880943762545659e-05,
+      "loss": 0.9642,
+      "step": 116340
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.187765309816052e-05,
+      "loss": 0.5978,
+      "step": 116350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.187436243377538e-05,
+      "loss": 0.6472,
+      "step": 116360
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1871071769390241e-05,
+      "loss": 0.57,
+      "step": 116370
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.18677811050051e-05,
+      "loss": 0.6403,
+      "step": 116380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1864490440619962e-05,
+      "loss": 0.5736,
+      "step": 116390
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1861199776234823e-05,
+      "loss": 0.6421,
+      "step": 116400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1857909111849683e-05,
+      "loss": 0.6247,
+      "step": 116410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1854618447464544e-05,
+      "loss": 0.6332,
+      "step": 116420
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1851327783079403e-05,
+      "loss": 0.5331,
+      "step": 116430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1848037118694265e-05,
+      "loss": 0.5796,
+      "step": 116440
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1844746454309126e-05,
+      "loss": 0.4884,
+      "step": 116450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1841455789923986e-05,
+      "loss": 0.7359,
+      "step": 116460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1838165125538847e-05,
+      "loss": 0.6516,
+      "step": 116470
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1834874461153706e-05,
+      "loss": 0.6483,
+      "step": 116480
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1831583796768568e-05,
+      "loss": 0.5392,
+      "step": 116490
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1828293132383429e-05,
+      "loss": 0.6693,
+      "step": 116500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1825002467998288e-05,
+      "loss": 0.653,
+      "step": 116510
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182171180361315e-05,
+      "loss": 0.53,
+      "step": 116520
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.181842113922801e-05,
+      "loss": 0.6332,
+      "step": 116530
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.181513047484287e-05,
+      "loss": 0.6184,
+      "step": 116540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1811839810457732e-05,
+      "loss": 0.6395,
+      "step": 116550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1808549146072591e-05,
+      "loss": 0.6135,
+      "step": 116560
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1805258481687453e-05,
+      "loss": 0.6255,
+      "step": 116570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1801967817302314e-05,
+      "loss": 0.6296,
+      "step": 116580
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1798677152917175e-05,
+      "loss": 0.5413,
+      "step": 116590
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1795386488532035e-05,
+      "loss": 0.6108,
+      "step": 116600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1792095824146896e-05,
+      "loss": 0.6195,
+      "step": 116610
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1788805159761757e-05,
+      "loss": 0.578,
+      "step": 116620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1785514495376619e-05,
+      "loss": 0.6408,
+      "step": 116630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1782223830991478e-05,
+      "loss": 0.6218,
+      "step": 116640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177893316660634e-05,
+      "loss": 0.4832,
+      "step": 116650
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1775642502221199e-05,
+      "loss": 0.5587,
+      "step": 116660
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177235183783606e-05,
+      "loss": 0.7278,
+      "step": 116670
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1769061173450922e-05,
+      "loss": 0.5906,
+      "step": 116680
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1765770509065781e-05,
+      "loss": 0.6806,
+      "step": 116690
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1762479844680642e-05,
+      "loss": 0.5841,
+      "step": 116700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1759189180295502e-05,
+      "loss": 0.6837,
+      "step": 116710
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1755898515910363e-05,
+      "loss": 0.5533,
+      "step": 116720
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1752607851525225e-05,
+      "loss": 0.5453,
+      "step": 116730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1749317187140084e-05,
+      "loss": 0.6683,
+      "step": 116740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1746026522754945e-05,
+      "loss": 0.5644,
+      "step": 116750
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1742735858369805e-05,
+      "loss": 0.6689,
+      "step": 116760
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1739445193984666e-05,
+      "loss": 0.5871,
+      "step": 116770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1736154529599527e-05,
+      "loss": 0.493,
+      "step": 116780
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1732863865214387e-05,
+      "loss": 0.6445,
+      "step": 116790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1729573200829248e-05,
+      "loss": 0.6242,
+      "step": 116800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1726282536444108e-05,
+      "loss": 0.553,
+      "step": 116810
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1722991872058969e-05,
+      "loss": 0.5918,
+      "step": 116820
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.171970120767383e-05,
+      "loss": 0.573,
+      "step": 116830
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.171641054328869e-05,
+      "loss": 0.5318,
+      "step": 116840
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1713119878903551e-05,
+      "loss": 0.5797,
+      "step": 116850
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.170982921451841e-05,
+      "loss": 0.6711,
+      "step": 116860
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1706538550133272e-05,
+      "loss": 0.6587,
+      "step": 116870
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1703247885748133e-05,
+      "loss": 0.5411,
+      "step": 116880
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1699957221362993e-05,
+      "loss": 0.4872,
+      "step": 116890
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1696666556977854e-05,
+      "loss": 0.5846,
+      "step": 116900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1693375892592714e-05,
+      "loss": 0.5656,
+      "step": 116910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1690085228207575e-05,
+      "loss": 0.6077,
+      "step": 116920
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1686794563822436e-05,
+      "loss": 0.6768,
+      "step": 116930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1683503899437296e-05,
+      "loss": 0.6364,
+      "step": 116940
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1680213235052157e-05,
+      "loss": 0.5725,
+      "step": 116950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1676922570667018e-05,
+      "loss": 0.7622,
+      "step": 116960
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.167363190628188e-05,
+      "loss": 0.6066,
+      "step": 116970
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.167034124189674e-05,
+      "loss": 0.636,
+      "step": 116980
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.16670505775116e-05,
+      "loss": 0.5507,
+      "step": 116990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1663759913126462e-05,
+      "loss": 0.5892,
+      "step": 117000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1660469248741321e-05,
+      "loss": 0.6709,
+      "step": 117010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1657178584356183e-05,
+      "loss": 0.6139,
+      "step": 117020
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1653887919971044e-05,
+      "loss": 0.569,
+      "step": 117030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1650597255585903e-05,
+      "loss": 0.4829,
+      "step": 117040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1647306591200765e-05,
+      "loss": 0.6047,
+      "step": 117050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1644015926815624e-05,
+      "loss": 0.4686,
+      "step": 117060
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1640725262430486e-05,
+      "loss": 0.5344,
+      "step": 117070
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1637434598045347e-05,
+      "loss": 0.899,
+      "step": 117080
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1634143933660206e-05,
+      "loss": 0.5991,
+      "step": 117090
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1630853269275068e-05,
+      "loss": 0.5703,
+      "step": 117100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1627562604889927e-05,
+      "loss": 0.7078,
+      "step": 117110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1624271940504789e-05,
+      "loss": 0.573,
+      "step": 117120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.162098127611965e-05,
+      "loss": 0.6893,
+      "step": 117130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.161769061173451e-05,
+      "loss": 0.4782,
+      "step": 117140
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.161439994734937e-05,
+      "loss": 0.6104,
+      "step": 117150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.161110928296423e-05,
+      "loss": 0.5237,
+      "step": 117160
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1607818618579091e-05,
+      "loss": 0.5546,
+      "step": 117170
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1604527954193953e-05,
+      "loss": 0.5944,
+      "step": 117180
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1601237289808812e-05,
+      "loss": 0.6084,
+      "step": 117190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1597946625423674e-05,
+      "loss": 0.5453,
+      "step": 117200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1594655961038533e-05,
+      "loss": 0.6253,
+      "step": 117210
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1591365296653394e-05,
+      "loss": 0.7338,
+      "step": 117220
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1588074632268256e-05,
+      "loss": 0.6176,
+      "step": 117230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1584783967883115e-05,
+      "loss": 0.584,
+      "step": 117240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1581493303497977e-05,
+      "loss": 0.5301,
+      "step": 117250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1578202639112836e-05,
+      "loss": 0.5529,
+      "step": 117260
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1574911974727697e-05,
+      "loss": 0.5733,
+      "step": 117270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1571621310342559e-05,
+      "loss": 0.6234,
+      "step": 117280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1568330645957418e-05,
+      "loss": 0.6555,
+      "step": 117290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.156503998157228e-05,
+      "loss": 0.7439,
+      "step": 117300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.156174931718714e-05,
+      "loss": 0.5179,
+      "step": 117310
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1558458652802e-05,
+      "loss": 0.5715,
+      "step": 117320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1555167988416862e-05,
+      "loss": 0.5386,
+      "step": 117330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1551877324031723e-05,
+      "loss": 0.4507,
+      "step": 117340
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1548586659646584e-05,
+      "loss": 0.5949,
+      "step": 117350
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1545295995261444e-05,
+      "loss": 0.5988,
+      "step": 117360
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1542005330876305e-05,
+      "loss": 0.6496,
+      "step": 117370
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1538714666491166e-05,
+      "loss": 0.6236,
+      "step": 117380
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1535424002106026e-05,
+      "loss": 0.7887,
+      "step": 117390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1532133337720887e-05,
+      "loss": 0.6449,
+      "step": 117400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1528842673335748e-05,
+      "loss": 0.5976,
+      "step": 117410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1525552008950608e-05,
+      "loss": 0.5878,
+      "step": 117420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.152226134456547e-05,
+      "loss": 0.6071,
+      "step": 117430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1518970680180329e-05,
+      "loss": 0.5581,
+      "step": 117440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.151568001579519e-05,
+      "loss": 0.6254,
+      "step": 117450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1512389351410051e-05,
+      "loss": 0.6615,
+      "step": 117460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1509098687024911e-05,
+      "loss": 0.5383,
+      "step": 117470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1505808022639772e-05,
+      "loss": 0.5451,
+      "step": 117480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1502517358254632e-05,
+      "loss": 0.6603,
+      "step": 117490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1499226693869493e-05,
+      "loss": 0.7068,
+      "step": 117500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1495936029484354e-05,
+      "loss": 0.5187,
+      "step": 117510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1492645365099214e-05,
+      "loss": 0.562,
+      "step": 117520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1489354700714075e-05,
+      "loss": 0.5996,
+      "step": 117530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1486064036328935e-05,
+      "loss": 0.5236,
+      "step": 117540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1482773371943796e-05,
+      "loss": 0.589,
+      "step": 117550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1479482707558657e-05,
+      "loss": 0.586,
+      "step": 117560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1476192043173517e-05,
+      "loss": 0.6239,
+      "step": 117570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1472901378788378e-05,
+      "loss": 0.5948,
+      "step": 117580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1469610714403238e-05,
+      "loss": 0.5459,
+      "step": 117590
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1466320050018099e-05,
+      "loss": 0.5932,
+      "step": 117600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.146302938563296e-05,
+      "loss": 0.6286,
+      "step": 117610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.145973872124782e-05,
+      "loss": 0.6852,
+      "step": 117620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1456448056862681e-05,
+      "loss": 0.5384,
+      "step": 117630
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.145315739247754e-05,
+      "loss": 0.5959,
+      "step": 117640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1449866728092402e-05,
+      "loss": 0.5586,
+      "step": 117650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1446576063707263e-05,
+      "loss": 0.7504,
+      "step": 117660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1443285399322123e-05,
+      "loss": 0.5893,
+      "step": 117670
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1439994734936984e-05,
+      "loss": 0.4927,
+      "step": 117680
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1436704070551845e-05,
+      "loss": 0.5794,
+      "step": 117690
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1433413406166705e-05,
+      "loss": 0.6106,
+      "step": 117700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1430122741781566e-05,
+      "loss": 0.5522,
+      "step": 117710
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1426832077396427e-05,
+      "loss": 0.6552,
+      "step": 117720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1423541413011289e-05,
+      "loss": 0.6369,
+      "step": 117730
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1420250748626148e-05,
+      "loss": 0.5892,
+      "step": 117740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.141696008424101e-05,
+      "loss": 0.6543,
+      "step": 117750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.141366941985587e-05,
+      "loss": 0.6738,
+      "step": 117760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.141037875547073e-05,
+      "loss": 0.6051,
+      "step": 117770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1407088091085592e-05,
+      "loss": 0.667,
+      "step": 117780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1403797426700451e-05,
+      "loss": 0.619,
+      "step": 117790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1400506762315312e-05,
+      "loss": 0.6559,
+      "step": 117800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1397216097930174e-05,
+      "loss": 0.5888,
+      "step": 117810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1393925433545033e-05,
+      "loss": 0.7446,
+      "step": 117820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1390634769159894e-05,
+      "loss": 0.5902,
+      "step": 117830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1387344104774754e-05,
+      "loss": 0.5767,
+      "step": 117840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1384053440389615e-05,
+      "loss": 0.8006,
+      "step": 117850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1380762776004477e-05,
+      "loss": 0.5492,
+      "step": 117860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1377472111619336e-05,
+      "loss": 0.4942,
+      "step": 117870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1374181447234197e-05,
+      "loss": 0.6543,
+      "step": 117880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1370890782849057e-05,
+      "loss": 0.7895,
+      "step": 117890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1367600118463918e-05,
+      "loss": 0.624,
+      "step": 117900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.136430945407878e-05,
+      "loss": 0.5608,
+      "step": 117910
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1361018789693639e-05,
+      "loss": 0.6453,
+      "step": 117920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.13577281253085e-05,
+      "loss": 0.5624,
+      "step": 117930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.135443746092336e-05,
+      "loss": 0.7844,
+      "step": 117940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1351146796538221e-05,
+      "loss": 0.691,
+      "step": 117950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1347856132153082e-05,
+      "loss": 0.5917,
+      "step": 117960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1344565467767942e-05,
+      "loss": 0.3789,
+      "step": 117970
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1341274803382803e-05,
+      "loss": 0.5019,
+      "step": 117980
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1337984138997663e-05,
+      "loss": 0.5829,
+      "step": 117990
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1334693474612524e-05,
+      "loss": 0.6186,
+      "step": 118000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1331402810227385e-05,
+      "loss": 0.6323,
+      "step": 118010
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1328112145842245e-05,
+      "loss": 0.617,
+      "step": 118020
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1324821481457106e-05,
+      "loss": 0.7437,
+      "step": 118030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1321530817071968e-05,
+      "loss": 0.6808,
+      "step": 118040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1318240152686827e-05,
+      "loss": 0.5892,
+      "step": 118050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1314949488301688e-05,
+      "loss": 0.6148,
+      "step": 118060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.131165882391655e-05,
+      "loss": 0.6567,
+      "step": 118070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.130836815953141e-05,
+      "loss": 0.6262,
+      "step": 118080
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.130507749514627e-05,
+      "loss": 0.6824,
+      "step": 118090
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1301786830761132e-05,
+      "loss": 0.6684,
+      "step": 118100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1298496166375993e-05,
+      "loss": 0.5922,
+      "step": 118110
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1295205501990853e-05,
+      "loss": 0.6288,
+      "step": 118120
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1291914837605714e-05,
+      "loss": 0.6613,
+      "step": 118130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1288624173220575e-05,
+      "loss": 0.6495,
+      "step": 118140
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1285333508835435e-05,
+      "loss": 0.5131,
+      "step": 118150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1282042844450296e-05,
+      "loss": 0.5485,
+      "step": 118160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1278752180065156e-05,
+      "loss": 0.5865,
+      "step": 118170
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1275461515680017e-05,
+      "loss": 0.6373,
+      "step": 118180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1272170851294878e-05,
+      "loss": 0.5443,
+      "step": 118190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1268880186909738e-05,
+      "loss": 0.5289,
+      "step": 118200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1265589522524599e-05,
+      "loss": 0.6215,
+      "step": 118210
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1262298858139459e-05,
+      "loss": 0.5672,
+      "step": 118220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.125900819375432e-05,
+      "loss": 0.5938,
+      "step": 118230
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1255717529369181e-05,
+      "loss": 0.549,
+      "step": 118240
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.125242686498404e-05,
+      "loss": 0.5876,
+      "step": 118250
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1249136200598902e-05,
+      "loss": 0.5832,
+      "step": 118260
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1245845536213761e-05,
+      "loss": 0.6397,
+      "step": 118270
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1242554871828623e-05,
+      "loss": 0.673,
+      "step": 118280
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1239264207443484e-05,
+      "loss": 0.589,
+      "step": 118290
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1235973543058344e-05,
+      "loss": 0.6003,
+      "step": 118300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1232682878673205e-05,
+      "loss": 0.6048,
+      "step": 118310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1229392214288064e-05,
+      "loss": 0.5984,
+      "step": 118320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1226101549902926e-05,
+      "loss": 0.5568,
+      "step": 118330
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1222810885517787e-05,
+      "loss": 0.6496,
+      "step": 118340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1219520221132647e-05,
+      "loss": 0.578,
+      "step": 118350
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1216229556747508e-05,
+      "loss": 0.5944,
+      "step": 118360
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1212938892362367e-05,
+      "loss": 0.5622,
+      "step": 118370
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1209648227977229e-05,
+      "loss": 0.6127,
+      "step": 118380
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.120635756359209e-05,
+      "loss": 0.6141,
+      "step": 118390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.120306689920695e-05,
+      "loss": 0.6062,
+      "step": 118400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.119977623482181e-05,
+      "loss": 0.6568,
+      "step": 118410
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.119648557043667e-05,
+      "loss": 0.5259,
+      "step": 118420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1193194906051532e-05,
+      "loss": 0.678,
+      "step": 118430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1189904241666393e-05,
+      "loss": 0.6216,
+      "step": 118440
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1186613577281254e-05,
+      "loss": 0.6535,
+      "step": 118450
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1183322912896114e-05,
+      "loss": 0.6196,
+      "step": 118460
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1180032248510975e-05,
+      "loss": 0.6106,
+      "step": 118470
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1176741584125836e-05,
+      "loss": 0.6379,
+      "step": 118480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1173450919740697e-05,
+      "loss": 0.6256,
+      "step": 118490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1170160255355557e-05,
+      "loss": 0.6072,
+      "step": 118500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1166869590970418e-05,
+      "loss": 0.6034,
+      "step": 118510
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1163578926585278e-05,
+      "loss": 0.4409,
+      "step": 118520
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.116028826220014e-05,
+      "loss": 0.6471,
+      "step": 118530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1156997597815e-05,
+      "loss": 0.6363,
+      "step": 118540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.115370693342986e-05,
+      "loss": 0.5305,
+      "step": 118550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1150416269044721e-05,
+      "loss": 0.5343,
+      "step": 118560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.114712560465958e-05,
+      "loss": 0.7019,
+      "step": 118570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1143834940274442e-05,
+      "loss": 0.4887,
+      "step": 118580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1140544275889303e-05,
+      "loss": 0.599,
+      "step": 118590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1137253611504163e-05,
+      "loss": 0.5776,
+      "step": 118600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1133962947119024e-05,
+      "loss": 0.534,
+      "step": 118610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1130672282733884e-05,
+      "loss": 0.5487,
+      "step": 118620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1127381618348745e-05,
+      "loss": 0.582,
+      "step": 118630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1124090953963606e-05,
+      "loss": 0.5923,
+      "step": 118640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1120800289578466e-05,
+      "loss": 0.5884,
+      "step": 118650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1117509625193327e-05,
+      "loss": 0.5156,
+      "step": 118660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1114218960808187e-05,
+      "loss": 0.5397,
+      "step": 118670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1110928296423048e-05,
+      "loss": 0.691,
+      "step": 118680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.110763763203791e-05,
+      "loss": 0.6514,
+      "step": 118690
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1104346967652769e-05,
+      "loss": 0.6302,
+      "step": 118700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.110105630326763e-05,
+      "loss": 0.6124,
+      "step": 118710
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1097765638882491e-05,
+      "loss": 0.5559,
+      "step": 118720
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1094474974497351e-05,
+      "loss": 0.5247,
+      "step": 118730
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1091184310112212e-05,
+      "loss": 0.5957,
+      "step": 118740
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1087893645727072e-05,
+      "loss": 0.6294,
+      "step": 118750
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1084602981341933e-05,
+      "loss": 0.7614,
+      "step": 118760
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1081312316956794e-05,
+      "loss": 0.6801,
+      "step": 118770
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1078021652571654e-05,
+      "loss": 0.6699,
+      "step": 118780
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1074730988186515e-05,
+      "loss": 0.559,
+      "step": 118790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1071440323801375e-05,
+      "loss": 0.616,
+      "step": 118800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1068149659416236e-05,
+      "loss": 0.6336,
+      "step": 118810
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1064858995031097e-05,
+      "loss": 0.6114,
+      "step": 118820
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1061568330645959e-05,
+      "loss": 0.5541,
+      "step": 118830
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1058277666260818e-05,
+      "loss": 0.5963,
+      "step": 118840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.105498700187568e-05,
+      "loss": 0.7032,
+      "step": 118850
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.105169633749054e-05,
+      "loss": 0.5863,
+      "step": 118860
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1048405673105402e-05,
+      "loss": 0.625,
+      "step": 118870
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1045115008720261e-05,
+      "loss": 0.6398,
+      "step": 118880
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1041824344335123e-05,
+      "loss": 0.6256,
+      "step": 118890
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1038533679949982e-05,
+      "loss": 0.6898,
+      "step": 118900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1035243015564844e-05,
+      "loss": 0.6176,
+      "step": 118910
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1031952351179705e-05,
+      "loss": 0.6284,
+      "step": 118920
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1028661686794564e-05,
+      "loss": 0.7318,
+      "step": 118930
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1025371022409426e-05,
+      "loss": 0.6573,
+      "step": 118940
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1022080358024285e-05,
+      "loss": 0.7417,
+      "step": 118950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1018789693639147e-05,
+      "loss": 0.5518,
+      "step": 118960
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1015499029254008e-05,
+      "loss": 0.7118,
+      "step": 118970
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1012208364868867e-05,
+      "loss": 0.5457,
+      "step": 118980
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1008917700483729e-05,
+      "loss": 0.5301,
+      "step": 118990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1005627036098588e-05,
+      "loss": 0.6763,
+      "step": 119000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.100233637171345e-05,
+      "loss": 0.5554,
+      "step": 119010
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.099904570732831e-05,
+      "loss": 0.6726,
+      "step": 119020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.099575504294317e-05,
+      "loss": 0.6711,
+      "step": 119030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0992464378558032e-05,
+      "loss": 0.5396,
+      "step": 119040
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0989173714172891e-05,
+      "loss": 0.6705,
+      "step": 119050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0985883049787752e-05,
+      "loss": 0.5634,
+      "step": 119060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0982592385402614e-05,
+      "loss": 0.5228,
+      "step": 119070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0979301721017473e-05,
+      "loss": 0.6482,
+      "step": 119080
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0976011056632335e-05,
+      "loss": 0.6255,
+      "step": 119090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0972720392247194e-05,
+      "loss": 0.614,
+      "step": 119100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0969429727862055e-05,
+      "loss": 0.6039,
+      "step": 119110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0966139063476917e-05,
+      "loss": 0.6029,
+      "step": 119120
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0962848399091776e-05,
+      "loss": 0.6285,
+      "step": 119130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0959557734706638e-05,
+      "loss": 0.6345,
+      "step": 119140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0956267070321497e-05,
+      "loss": 0.6789,
+      "step": 119150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0952976405936358e-05,
+      "loss": 0.624,
+      "step": 119160
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.094968574155122e-05,
+      "loss": 0.7192,
+      "step": 119170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0946395077166081e-05,
+      "loss": 0.5806,
+      "step": 119180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.094310441278094e-05,
+      "loss": 0.5576,
+      "step": 119190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0939813748395802e-05,
+      "loss": 0.5885,
+      "step": 119200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0936523084010663e-05,
+      "loss": 0.6885,
+      "step": 119210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0933232419625524e-05,
+      "loss": 0.7132,
+      "step": 119220
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0929941755240384e-05,
+      "loss": 0.5485,
+      "step": 119230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0926651090855245e-05,
+      "loss": 0.6073,
+      "step": 119240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0923360426470105e-05,
+      "loss": 0.5896,
+      "step": 119250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0920069762084966e-05,
+      "loss": 0.5461,
+      "step": 119260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0916779097699827e-05,
+      "loss": 0.6488,
+      "step": 119270
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0913488433314687e-05,
+      "loss": 0.7364,
+      "step": 119280
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0910197768929548e-05,
+      "loss": 0.6554,
+      "step": 119290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0906907104544408e-05,
+      "loss": 0.647,
+      "step": 119300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0903616440159269e-05,
+      "loss": 0.6132,
+      "step": 119310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.090032577577413e-05,
+      "loss": 0.5196,
+      "step": 119320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.089703511138899e-05,
+      "loss": 0.5539,
+      "step": 119330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0893744447003851e-05,
+      "loss": 0.6752,
+      "step": 119340
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.089045378261871e-05,
+      "loss": 0.5753,
+      "step": 119350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0887163118233572e-05,
+      "loss": 0.5114,
+      "step": 119360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0883872453848433e-05,
+      "loss": 0.5271,
+      "step": 119370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0880581789463293e-05,
+      "loss": 0.5906,
+      "step": 119380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0877291125078154e-05,
+      "loss": 0.6047,
+      "step": 119390
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0874000460693014e-05,
+      "loss": 0.6885,
+      "step": 119400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0870709796307875e-05,
+      "loss": 0.6905,
+      "step": 119410
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0867419131922736e-05,
+      "loss": 0.6135,
+      "step": 119420
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0864128467537596e-05,
+      "loss": 0.8045,
+      "step": 119430
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0860837803152457e-05,
+      "loss": 0.6225,
+      "step": 119440
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0857547138767318e-05,
+      "loss": 0.6966,
+      "step": 119450
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0854256474382178e-05,
+      "loss": 0.6167,
+      "step": 119460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0850965809997039e-05,
+      "loss": 0.7304,
+      "step": 119470
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0847675145611899e-05,
+      "loss": 0.5355,
+      "step": 119480
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.084438448122676e-05,
+      "loss": 0.6132,
+      "step": 119490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0841093816841621e-05,
+      "loss": 0.5184,
+      "step": 119500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.083780315245648e-05,
+      "loss": 0.6015,
+      "step": 119510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0834512488071342e-05,
+      "loss": 0.5532,
+      "step": 119520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0831221823686202e-05,
+      "loss": 0.6459,
+      "step": 119530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0827931159301063e-05,
+      "loss": 0.6299,
+      "step": 119540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0824640494915924e-05,
+      "loss": 0.6221,
+      "step": 119550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0821349830530785e-05,
+      "loss": 0.6625,
+      "step": 119560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0818059166145645e-05,
+      "loss": 0.584,
+      "step": 119570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0814768501760506e-05,
+      "loss": 0.5587,
+      "step": 119580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0811477837375367e-05,
+      "loss": 0.6129,
+      "step": 119590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0808187172990229e-05,
+      "loss": 0.6196,
+      "step": 119600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0804896508605088e-05,
+      "loss": 0.6921,
+      "step": 119610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.080160584421995e-05,
+      "loss": 0.6487,
+      "step": 119620
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0798315179834809e-05,
+      "loss": 0.5611,
+      "step": 119630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.079502451544967e-05,
+      "loss": 0.6475,
+      "step": 119640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0791733851064532e-05,
+      "loss": 0.5998,
+      "step": 119650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0788443186679391e-05,
+      "loss": 0.6417,
+      "step": 119660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0785152522294252e-05,
+      "loss": 0.4492,
+      "step": 119670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0781861857909112e-05,
+      "loss": 0.6026,
+      "step": 119680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0778571193523973e-05,
+      "loss": 0.5638,
+      "step": 119690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0775280529138835e-05,
+      "loss": 0.5668,
+      "step": 119700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0771989864753694e-05,
+      "loss": 0.5282,
+      "step": 119710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0768699200368555e-05,
+      "loss": 0.5997,
+      "step": 119720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0765408535983415e-05,
+      "loss": 0.6006,
+      "step": 119730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0762117871598276e-05,
+      "loss": 0.5736,
+      "step": 119740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0758827207213138e-05,
+      "loss": 0.6998,
+      "step": 119750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0755536542827997e-05,
+      "loss": 0.5269,
+      "step": 119760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0752245878442858e-05,
+      "loss": 0.5857,
+      "step": 119770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0748955214057718e-05,
+      "loss": 0.7132,
+      "step": 119780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.074566454967258e-05,
+      "loss": 0.6687,
+      "step": 119790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.074237388528744e-05,
+      "loss": 0.6102,
+      "step": 119800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.07390832209023e-05,
+      "loss": 0.5961,
+      "step": 119810
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0735792556517161e-05,
+      "loss": 0.6553,
+      "step": 119820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0732501892132021e-05,
+      "loss": 0.6468,
+      "step": 119830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0729211227746882e-05,
+      "loss": 0.6856,
+      "step": 119840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0725920563361743e-05,
+      "loss": 0.6171,
+      "step": 119850
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0722629898976603e-05,
+      "loss": 0.6639,
+      "step": 119860
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0719339234591464e-05,
+      "loss": 0.5754,
+      "step": 119870
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0716048570206324e-05,
+      "loss": 0.6468,
+      "step": 119880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0712757905821185e-05,
+      "loss": 0.6428,
+      "step": 119890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0709467241436046e-05,
+      "loss": 0.6389,
+      "step": 119900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0706176577050906e-05,
+      "loss": 0.5816,
+      "step": 119910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0702885912665767e-05,
+      "loss": 0.6419,
+      "step": 119920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0699595248280629e-05,
+      "loss": 0.5782,
+      "step": 119930
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.069630458389549e-05,
+      "loss": 0.6914,
+      "step": 119940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.069301391951035e-05,
+      "loss": 0.5147,
+      "step": 119950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.068972325512521e-05,
+      "loss": 0.6106,
+      "step": 119960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0686432590740072e-05,
+      "loss": 0.5749,
+      "step": 119970
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0683141926354931e-05,
+      "loss": 0.8536,
+      "step": 119980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0679851261969793e-05,
+      "loss": 0.6118,
+      "step": 119990
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0676560597584654e-05,
+      "loss": 0.6397,
+      "step": 120000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0673269933199514e-05,
+      "loss": 0.484,
+      "step": 120010
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0669979268814375e-05,
+      "loss": 0.5758,
+      "step": 120020
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0666688604429234e-05,
+      "loss": 0.5668,
+      "step": 120030
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0663397940044096e-05,
+      "loss": 0.6438,
+      "step": 120040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0660107275658957e-05,
+      "loss": 0.5998,
+      "step": 120050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0656816611273817e-05,
+      "loss": 0.428,
+      "step": 120060
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0653525946888678e-05,
+      "loss": 0.5967,
+      "step": 120070
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0650235282503537e-05,
+      "loss": 0.7041,
+      "step": 120080
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0646944618118399e-05,
+      "loss": 0.5956,
+      "step": 120090
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.064365395373326e-05,
+      "loss": 0.5947,
+      "step": 120100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.064036328934812e-05,
+      "loss": 0.6227,
+      "step": 120110
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.063707262496298e-05,
+      "loss": 0.6611,
+      "step": 120120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.063378196057784e-05,
+      "loss": 0.5981,
+      "step": 120130
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0630491296192702e-05,
+      "loss": 0.5104,
+      "step": 120140
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0627200631807563e-05,
+      "loss": 0.5661,
+      "step": 120150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0623909967422422e-05,
+      "loss": 0.5865,
+      "step": 120160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0620619303037284e-05,
+      "loss": 0.6453,
+      "step": 120170
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0617328638652145e-05,
+      "loss": 0.6026,
+      "step": 120180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0614037974267005e-05,
+      "loss": 0.5787,
+      "step": 120190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0610747309881866e-05,
+      "loss": 0.5407,
+      "step": 120200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0607456645496725e-05,
+      "loss": 0.5532,
+      "step": 120210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0604165981111587e-05,
+      "loss": 0.7068,
+      "step": 120220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0600875316726448e-05,
+      "loss": 0.559,
+      "step": 120230
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0597584652341307e-05,
+      "loss": 0.6449,
+      "step": 120240
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0594293987956169e-05,
+      "loss": 0.6656,
+      "step": 120250
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0591003323571028e-05,
+      "loss": 0.7098,
+      "step": 120260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.058771265918589e-05,
+      "loss": 0.6061,
+      "step": 120270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0584421994800751e-05,
+      "loss": 0.6096,
+      "step": 120280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.058113133041561e-05,
+      "loss": 0.6048,
+      "step": 120290
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0577840666030472e-05,
+      "loss": 0.5919,
+      "step": 120300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0574550001645333e-05,
+      "loss": 0.6687,
+      "step": 120310
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0571259337260194e-05,
+      "loss": 0.6147,
+      "step": 120320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0567968672875054e-05,
+      "loss": 0.645,
+      "step": 120330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0564678008489915e-05,
+      "loss": 0.5795,
+      "step": 120340
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0561387344104776e-05,
+      "loss": 0.6107,
+      "step": 120350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0558096679719636e-05,
+      "loss": 0.6537,
+      "step": 120360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0554806015334497e-05,
+      "loss": 0.5721,
+      "step": 120370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0551515350949358e-05,
+      "loss": 0.6861,
+      "step": 120380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0548224686564218e-05,
+      "loss": 0.579,
+      "step": 120390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.054493402217908e-05,
+      "loss": 0.5121,
+      "step": 120400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0541643357793939e-05,
+      "loss": 0.625,
+      "step": 120410
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.05383526934088e-05,
+      "loss": 0.7379,
+      "step": 120420
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0535062029023661e-05,
+      "loss": 0.6313,
+      "step": 120430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0531771364638521e-05,
+      "loss": 0.6507,
+      "step": 120440
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0528480700253382e-05,
+      "loss": 0.5301,
+      "step": 120450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0525190035868242e-05,
+      "loss": 0.6553,
+      "step": 120460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0521899371483103e-05,
+      "loss": 0.5713,
+      "step": 120470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0518608707097964e-05,
+      "loss": 0.5076,
+      "step": 120480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0515318042712824e-05,
+      "loss": 0.7031,
+      "step": 120490
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0512027378327685e-05,
+      "loss": 0.5365,
+      "step": 120500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0508736713942545e-05,
+      "loss": 0.6614,
+      "step": 120510
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0505446049557406e-05,
+      "loss": 0.7316,
+      "step": 120520
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0502155385172267e-05,
+      "loss": 0.5285,
+      "step": 120530
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0498864720787127e-05,
+      "loss": 0.5061,
+      "step": 120540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0495574056401988e-05,
+      "loss": 0.557,
+      "step": 120550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0492283392016848e-05,
+      "loss": 0.5381,
+      "step": 120560
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0488992727631709e-05,
+      "loss": 0.582,
+      "step": 120570
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.048570206324657e-05,
+      "loss": 0.8746,
+      "step": 120580
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.048241139886143e-05,
+      "loss": 0.7158,
+      "step": 120590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0479120734476291e-05,
+      "loss": 0.6509,
+      "step": 120600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.047583007009115e-05,
+      "loss": 0.6983,
+      "step": 120610
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0472539405706012e-05,
+      "loss": 0.5793,
+      "step": 120620
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0469248741320873e-05,
+      "loss": 0.554,
+      "step": 120630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0465958076935733e-05,
+      "loss": 0.6106,
+      "step": 120640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0462667412550594e-05,
+      "loss": 0.6254,
+      "step": 120650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0459376748165455e-05,
+      "loss": 0.5861,
+      "step": 120660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0456086083780315e-05,
+      "loss": 0.8653,
+      "step": 120670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0452795419395176e-05,
+      "loss": 0.6228,
+      "step": 120680
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0449504755010037e-05,
+      "loss": 0.6522,
+      "step": 120690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0446214090624899e-05,
+      "loss": 0.5738,
+      "step": 120700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0442923426239758e-05,
+      "loss": 0.6979,
+      "step": 120710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.043963276185462e-05,
+      "loss": 0.7468,
+      "step": 120720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.043634209746948e-05,
+      "loss": 0.5451,
+      "step": 120730
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.043305143308434e-05,
+      "loss": 0.5827,
+      "step": 120740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0429760768699202e-05,
+      "loss": 0.6645,
+      "step": 120750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0426470104314061e-05,
+      "loss": 0.5723,
+      "step": 120760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0423179439928922e-05,
+      "loss": 0.5942,
+      "step": 120770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0419888775543784e-05,
+      "loss": 0.5899,
+      "step": 120780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0416598111158643e-05,
+      "loss": 0.5423,
+      "step": 120790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0413307446773505e-05,
+      "loss": 0.5632,
+      "step": 120800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0410016782388364e-05,
+      "loss": 0.6107,
+      "step": 120810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0406726118003225e-05,
+      "loss": 0.5684,
+      "step": 120820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0403435453618087e-05,
+      "loss": 0.4821,
+      "step": 120830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0400144789232946e-05,
+      "loss": 0.5414,
+      "step": 120840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0396854124847808e-05,
+      "loss": 0.6467,
+      "step": 120850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0393563460462669e-05,
+      "loss": 0.5253,
+      "step": 120860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0390272796077528e-05,
+      "loss": 0.5098,
+      "step": 120870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.038698213169239e-05,
+      "loss": 0.6061,
+      "step": 120880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.038369146730725e-05,
+      "loss": 0.5198,
+      "step": 120890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.038040080292211e-05,
+      "loss": 0.592,
+      "step": 120900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0377110138536972e-05,
+      "loss": 0.6171,
+      "step": 120910
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0373819474151831e-05,
+      "loss": 0.6068,
+      "step": 120920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0370528809766693e-05,
+      "loss": 0.6461,
+      "step": 120930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0367238145381552e-05,
+      "loss": 0.6038,
+      "step": 120940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0363947480996413e-05,
+      "loss": 0.5518,
+      "step": 120950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0360656816611275e-05,
+      "loss": 0.6552,
+      "step": 120960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0357366152226134e-05,
+      "loss": 0.5893,
+      "step": 120970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0354075487840996e-05,
+      "loss": 0.6898,
+      "step": 120980
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0350784823455855e-05,
+      "loss": 0.6069,
+      "step": 120990
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0347494159070716e-05,
+      "loss": 1.0014,
+      "step": 121000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0344203494685578e-05,
+      "loss": 0.5503,
+      "step": 121010
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0340912830300437e-05,
+      "loss": 0.5457,
+      "step": 121020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0337622165915298e-05,
+      "loss": 0.5838,
+      "step": 121030
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.033433150153016e-05,
+      "loss": 0.5976,
+      "step": 121040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.033104083714502e-05,
+      "loss": 0.5769,
+      "step": 121050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.032775017275988e-05,
+      "loss": 0.5676,
+      "step": 121060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0324459508374742e-05,
+      "loss": 0.5832,
+      "step": 121070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0321168843989603e-05,
+      "loss": 0.6166,
+      "step": 121080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0317878179604463e-05,
+      "loss": 0.6343,
+      "step": 121090
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0314587515219324e-05,
+      "loss": 0.5975,
+      "step": 121100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0311296850834185e-05,
+      "loss": 0.5113,
+      "step": 121110
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0308006186449045e-05,
+      "loss": 0.5833,
+      "step": 121120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0304715522063906e-05,
+      "loss": 0.495,
+      "step": 121130
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0301424857678766e-05,
+      "loss": 0.626,
+      "step": 121140
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0298134193293627e-05,
+      "loss": 0.6619,
+      "step": 121150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0294843528908488e-05,
+      "loss": 0.5314,
+      "step": 121160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0291552864523348e-05,
+      "loss": 0.5493,
+      "step": 121170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0288262200138209e-05,
+      "loss": 0.6391,
+      "step": 121180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0284971535753069e-05,
+      "loss": 0.5012,
+      "step": 121190
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.028168087136793e-05,
+      "loss": 0.5759,
+      "step": 121200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0278390206982791e-05,
+      "loss": 0.6447,
+      "step": 121210
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.027509954259765e-05,
+      "loss": 0.5232,
+      "step": 121220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0271808878212512e-05,
+      "loss": 0.5421,
+      "step": 121230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0268518213827372e-05,
+      "loss": 0.7304,
+      "step": 121240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0265227549442233e-05,
+      "loss": 0.5663,
+      "step": 121250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0261936885057094e-05,
+      "loss": 0.5302,
+      "step": 121260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0258646220671954e-05,
+      "loss": 0.6563,
+      "step": 121270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0255355556286815e-05,
+      "loss": 0.5762,
+      "step": 121280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0252064891901674e-05,
+      "loss": 0.6865,
+      "step": 121290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0248774227516536e-05,
+      "loss": 0.5805,
+      "step": 121300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0245483563131397e-05,
+      "loss": 0.5381,
+      "step": 121310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0242192898746257e-05,
+      "loss": 0.6302,
+      "step": 121320
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0238902234361118e-05,
+      "loss": 0.5333,
+      "step": 121330
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0235611569975977e-05,
+      "loss": 0.6256,
+      "step": 121340
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0232320905590839e-05,
+      "loss": 0.6563,
+      "step": 121350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.02290302412057e-05,
+      "loss": 0.625,
+      "step": 121360
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.022573957682056e-05,
+      "loss": 0.6136,
+      "step": 121370
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.022244891243542e-05,
+      "loss": 0.7177,
+      "step": 121380
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.021915824805028e-05,
+      "loss": 0.5741,
+      "step": 121390
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0215867583665142e-05,
+      "loss": 0.5678,
+      "step": 121400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0212576919280003e-05,
+      "loss": 0.6314,
+      "step": 121410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0209286254894864e-05,
+      "loss": 0.6723,
+      "step": 121420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0205995590509724e-05,
+      "loss": 0.6314,
+      "step": 121430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0202704926124585e-05,
+      "loss": 0.7002,
+      "step": 121440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0199414261739446e-05,
+      "loss": 0.5789,
+      "step": 121450
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0196123597354308e-05,
+      "loss": 0.5299,
+      "step": 121460
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0192832932969167e-05,
+      "loss": 0.7306,
+      "step": 121470
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0189542268584028e-05,
+      "loss": 0.6521,
+      "step": 121480
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0186251604198888e-05,
+      "loss": 0.6158,
+      "step": 121490
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.018296093981375e-05,
+      "loss": 0.7019,
+      "step": 121500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.017967027542861e-05,
+      "loss": 0.6531,
+      "step": 121510
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.017637961104347e-05,
+      "loss": 0.4534,
+      "step": 121520
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0173088946658331e-05,
+      "loss": 0.5035,
+      "step": 121530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0169798282273191e-05,
+      "loss": 0.5801,
+      "step": 121540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0166507617888052e-05,
+      "loss": 0.7178,
+      "step": 121550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0163216953502913e-05,
+      "loss": 0.6714,
+      "step": 121560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0159926289117773e-05,
+      "loss": 0.5835,
+      "step": 121570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0156635624732634e-05,
+      "loss": 0.5846,
+      "step": 121580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0153344960347496e-05,
+      "loss": 0.66,
+      "step": 121590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0150054295962355e-05,
+      "loss": 0.6072,
+      "step": 121600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0146763631577216e-05,
+      "loss": 0.5233,
+      "step": 121610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0143472967192076e-05,
+      "loss": 0.6038,
+      "step": 121620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0140182302806937e-05,
+      "loss": 0.6381,
+      "step": 121630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0136891638421799e-05,
+      "loss": 0.5878,
+      "step": 121640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0133600974036658e-05,
+      "loss": 0.5729,
+      "step": 121650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.013031030965152e-05,
+      "loss": 0.5944,
+      "step": 121660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0127019645266379e-05,
+      "loss": 0.7232,
+      "step": 121670
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.012372898088124e-05,
+      "loss": 0.6919,
+      "step": 121680
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0120438316496101e-05,
+      "loss": 0.623,
+      "step": 121690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0117147652110961e-05,
+      "loss": 0.6796,
+      "step": 121700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0113856987725822e-05,
+      "loss": 0.5632,
+      "step": 121710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0110566323340682e-05,
+      "loss": 0.6478,
+      "step": 121720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0107275658955543e-05,
+      "loss": 0.6584,
+      "step": 121730
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0103984994570404e-05,
+      "loss": 0.6718,
+      "step": 121740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0100694330185264e-05,
+      "loss": 0.6938,
+      "step": 121750
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0097403665800125e-05,
+      "loss": 0.6933,
+      "step": 121760
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0094113001414985e-05,
+      "loss": 0.536,
+      "step": 121770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0090822337029846e-05,
+      "loss": 0.4892,
+      "step": 121780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0087531672644707e-05,
+      "loss": 0.6095,
+      "step": 121790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0084241008259569e-05,
+      "loss": 0.5755,
+      "step": 121800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0080950343874428e-05,
+      "loss": 0.5923,
+      "step": 121810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.007765967948929e-05,
+      "loss": 0.709,
+      "step": 121820
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.007436901510415e-05,
+      "loss": 0.582,
+      "step": 121830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0071078350719012e-05,
+      "loss": 0.5582,
+      "step": 121840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0067787686333872e-05,
+      "loss": 0.6758,
+      "step": 121850
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0064497021948733e-05,
+      "loss": 0.5609,
+      "step": 121860
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0061206357563592e-05,
+      "loss": 0.4674,
+      "step": 121870
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0057915693178454e-05,
+      "loss": 0.7239,
+      "step": 121880
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0054625028793315e-05,
+      "loss": 0.652,
+      "step": 121890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0051334364408175e-05,
+      "loss": 0.5786,
+      "step": 121900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0048043700023036e-05,
+      "loss": 0.6969,
+      "step": 121910
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0044753035637895e-05,
+      "loss": 0.594,
+      "step": 121920
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0041462371252757e-05,
+      "loss": 0.6491,
+      "step": 121930
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0038171706867618e-05,
+      "loss": 0.7185,
+      "step": 121940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0034881042482477e-05,
+      "loss": 0.654,
+      "step": 121950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0031590378097339e-05,
+      "loss": 0.6919,
+      "step": 121960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0028299713712198e-05,
+      "loss": 0.5604,
+      "step": 121970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.002500904932706e-05,
+      "loss": 0.5925,
+      "step": 121980
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0021718384941921e-05,
+      "loss": 0.5514,
+      "step": 121990
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.001842772055678e-05,
+      "loss": 0.5788,
+      "step": 122000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0015137056171642e-05,
+      "loss": 0.5817,
+      "step": 122010
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0011846391786501e-05,
+      "loss": 0.6056,
+      "step": 122020
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0008555727401363e-05,
+      "loss": 0.5729,
+      "step": 122030
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0005265063016224e-05,
+      "loss": 0.6181,
+      "step": 122040
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0001974398631083e-05,
+      "loss": 0.5746,
+      "step": 122050
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.998683734245945e-06,
+      "loss": 0.5693,
+      "step": 122060
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.995393069860804e-06,
+      "loss": 0.666,
+      "step": 122070
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.992102405475665e-06,
+      "loss": 0.6983,
+      "step": 122080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.988811741090527e-06,
+      "loss": 0.6086,
+      "step": 122090
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.985521076705386e-06,
+      "loss": 0.5019,
+      "step": 122100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.982230412320248e-06,
+      "loss": 0.6256,
+      "step": 122110
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.978939747935107e-06,
+      "loss": 0.6766,
+      "step": 122120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.975649083549968e-06,
+      "loss": 0.6079,
+      "step": 122130
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.97235841916483e-06,
+      "loss": 0.5021,
+      "step": 122140
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.969067754779691e-06,
+      "loss": 0.6562,
+      "step": 122150
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.96577709039455e-06,
+      "loss": 0.5169,
+      "step": 122160
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.962486426009412e-06,
+      "loss": 0.6125,
+      "step": 122170
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.959195761624273e-06,
+      "loss": 0.5178,
+      "step": 122180
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.955905097239134e-06,
+      "loss": 0.6136,
+      "step": 122190
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.952614432853994e-06,
+      "loss": 0.6489,
+      "step": 122200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.949323768468855e-06,
+      "loss": 0.6305,
+      "step": 122210
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.946033104083715e-06,
+      "loss": 0.6282,
+      "step": 122220
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.942742439698576e-06,
+      "loss": 0.6176,
+      "step": 122230
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.939451775313437e-06,
+      "loss": 0.6162,
+      "step": 122240
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.936161110928297e-06,
+      "loss": 0.6109,
+      "step": 122250
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.932870446543158e-06,
+      "loss": 0.5725,
+      "step": 122260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.92957978215802e-06,
+      "loss": 0.7117,
+      "step": 122270
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.926289117772879e-06,
+      "loss": 0.6072,
+      "step": 122280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.92299845338774e-06,
+      "loss": 0.5844,
+      "step": 122290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.9197077890026e-06,
+      "loss": 0.6128,
+      "step": 122300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.916417124617461e-06,
+      "loss": 0.5371,
+      "step": 122310
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.913126460232322e-06,
+      "loss": 0.7126,
+      "step": 122320
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.909835795847182e-06,
+      "loss": 0.5292,
+      "step": 122330
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.906545131462043e-06,
+      "loss": 0.6133,
+      "step": 122340
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.903254467076903e-06,
+      "loss": 0.6389,
+      "step": 122350
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.899963802691764e-06,
+      "loss": 0.6864,
+      "step": 122360
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.896673138306625e-06,
+      "loss": 0.6052,
+      "step": 122370
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.893382473921485e-06,
+      "loss": 0.6382,
+      "step": 122380
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.890091809536346e-06,
+      "loss": 0.5971,
+      "step": 122390
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.886801145151206e-06,
+      "loss": 0.547,
+      "step": 122400
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.883510480766067e-06,
+      "loss": 0.564,
+      "step": 122410
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.880219816380928e-06,
+      "loss": 0.7157,
+      "step": 122420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.876929151995788e-06,
+      "loss": 0.6104,
+      "step": 122430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.873638487610649e-06,
+      "loss": 0.5487,
+      "step": 122440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.870347823225509e-06,
+      "loss": 0.4446,
+      "step": 122450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.86705715884037e-06,
+      "loss": 0.5988,
+      "step": 122460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.863766494455231e-06,
+      "loss": 0.6153,
+      "step": 122470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.86047583007009e-06,
+      "loss": 0.6453,
+      "step": 122480
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.857185165684952e-06,
+      "loss": 0.5542,
+      "step": 122490
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.853894501299812e-06,
+      "loss": 0.6672,
+      "step": 122500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.850603836914673e-06,
+      "loss": 0.6001,
+      "step": 122510
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.847313172529534e-06,
+      "loss": 0.8895,
+      "step": 122520
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.844022508144395e-06,
+      "loss": 0.6214,
+      "step": 122530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.840731843759255e-06,
+      "loss": 0.6303,
+      "step": 122540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.837441179374116e-06,
+      "loss": 0.6018,
+      "step": 122550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.834150514988978e-06,
+      "loss": 0.7543,
+      "step": 122560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.830859850603839e-06,
+      "loss": 0.5531,
+      "step": 122570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.827569186218698e-06,
+      "loss": 0.6509,
+      "step": 122580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.82427852183356e-06,
+      "loss": 0.521,
+      "step": 122590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.82098785744842e-06,
+      "loss": 0.5237,
+      "step": 122600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.81769719306328e-06,
+      "loss": 0.55,
+      "step": 122610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.814406528678142e-06,
+      "loss": 0.6082,
+      "step": 122620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.811115864293001e-06,
+      "loss": 0.6211,
+      "step": 122630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.807825199907863e-06,
+      "loss": 0.6173,
+      "step": 122640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.804534535522722e-06,
+      "loss": 0.6365,
+      "step": 122650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.801243871137583e-06,
+      "loss": 0.4964,
+      "step": 122660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.797953206752445e-06,
+      "loss": 0.6362,
+      "step": 122670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.794662542367304e-06,
+      "loss": 0.5827,
+      "step": 122680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.791371877982166e-06,
+      "loss": 0.637,
+      "step": 122690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.788081213597025e-06,
+      "loss": 0.581,
+      "step": 122700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.784790549211886e-06,
+      "loss": 0.6287,
+      "step": 122710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.781499884826748e-06,
+      "loss": 0.5417,
+      "step": 122720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.778209220441607e-06,
+      "loss": 0.582,
+      "step": 122730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.774918556056468e-06,
+      "loss": 0.5689,
+      "step": 122740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.771627891671328e-06,
+      "loss": 0.6408,
+      "step": 122750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.76833722728619e-06,
+      "loss": 0.6841,
+      "step": 122760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.76504656290105e-06,
+      "loss": 0.5235,
+      "step": 122770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.76175589851591e-06,
+      "loss": 0.5516,
+      "step": 122780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.758465234130771e-06,
+      "loss": 0.554,
+      "step": 122790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.755174569745631e-06,
+      "loss": 0.6712,
+      "step": 122800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.751883905360492e-06,
+      "loss": 0.5597,
+      "step": 122810
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.748593240975354e-06,
+      "loss": 0.5856,
+      "step": 122820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.745302576590213e-06,
+      "loss": 0.5834,
+      "step": 122830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.742011912205074e-06,
+      "loss": 0.6174,
+      "step": 122840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.738721247819934e-06,
+      "loss": 0.581,
+      "step": 122850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.735430583434795e-06,
+      "loss": 0.5586,
+      "step": 122860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.732139919049656e-06,
+      "loss": 0.6782,
+      "step": 122870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.728849254664516e-06,
+      "loss": 0.6864,
+      "step": 122880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.725558590279377e-06,
+      "loss": 0.5201,
+      "step": 122890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.722267925894239e-06,
+      "loss": 0.5977,
+      "step": 122900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.7189772615091e-06,
+      "loss": 0.5507,
+      "step": 122910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.71568659712396e-06,
+      "loss": 0.584,
+      "step": 122920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.71239593273882e-06,
+      "loss": 0.5282,
+      "step": 122930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.709105268353682e-06,
+      "loss": 0.6173,
+      "step": 122940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.705814603968542e-06,
+      "loss": 0.5429,
+      "step": 122950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.702523939583403e-06,
+      "loss": 0.5778,
+      "step": 122960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.699233275198264e-06,
+      "loss": 0.6557,
+      "step": 122970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.695942610813124e-06,
+      "loss": 0.6648,
+      "step": 122980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.692651946427985e-06,
+      "loss": 0.4806,
+      "step": 122990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.689361282042846e-06,
+      "loss": 0.5286,
+      "step": 123000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.686070617657706e-06,
+      "loss": 0.6142,
+      "step": 123010
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.682779953272567e-06,
+      "loss": 0.6209,
+      "step": 123020
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.679489288887427e-06,
+      "loss": 0.5595,
+      "step": 123030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.676198624502288e-06,
+      "loss": 0.4683,
+      "step": 123040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.672907960117149e-06,
+      "loss": 0.4965,
+      "step": 123050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.669617295732009e-06,
+      "loss": 0.5852,
+      "step": 123060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.66632663134687e-06,
+      "loss": 0.7289,
+      "step": 123070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.66303596696173e-06,
+      "loss": 0.6778,
+      "step": 123080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.65974530257659e-06,
+      "loss": 0.6465,
+      "step": 123090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.656454638191452e-06,
+      "loss": 0.6526,
+      "step": 123100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.653163973806312e-06,
+      "loss": 0.4901,
+      "step": 123110
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.649873309421173e-06,
+      "loss": 0.5548,
+      "step": 123120
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.646582645036033e-06,
+      "loss": 0.6752,
+      "step": 123130
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.643291980650894e-06,
+      "loss": 0.6288,
+      "step": 123140
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.640001316265755e-06,
+      "loss": 0.6711,
+      "step": 123150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.636710651880615e-06,
+      "loss": 0.6169,
+      "step": 123160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.633419987495476e-06,
+      "loss": 0.6184,
+      "step": 123170
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.630129323110335e-06,
+      "loss": 0.5917,
+      "step": 123180
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.626838658725197e-06,
+      "loss": 0.5746,
+      "step": 123190
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.623547994340058e-06,
+      "loss": 0.6026,
+      "step": 123200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.620257329954918e-06,
+      "loss": 0.5707,
+      "step": 123210
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.616966665569779e-06,
+      "loss": 0.6495,
+      "step": 123220
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.613676001184638e-06,
+      "loss": 0.6259,
+      "step": 123230
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.6103853367995e-06,
+      "loss": 0.5132,
+      "step": 123240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.607094672414361e-06,
+      "loss": 0.5353,
+      "step": 123250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.60380400802922e-06,
+      "loss": 0.588,
+      "step": 123260
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.600513343644082e-06,
+      "loss": 0.6369,
+      "step": 123270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.597222679258943e-06,
+      "loss": 0.5949,
+      "step": 123280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.593932014873804e-06,
+      "loss": 0.6119,
+      "step": 123290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.590641350488664e-06,
+      "loss": 0.583,
+      "step": 123300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.587350686103525e-06,
+      "loss": 0.6137,
+      "step": 123310
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.584060021718386e-06,
+      "loss": 0.5157,
+      "step": 123320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.580769357333246e-06,
+      "loss": 0.5544,
+      "step": 123330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.577478692948107e-06,
+      "loss": 0.5014,
+      "step": 123340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.574188028562969e-06,
+      "loss": 0.5286,
+      "step": 123350
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.570897364177828e-06,
+      "loss": 0.584,
+      "step": 123360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.56760669979269e-06,
+      "loss": 0.6227,
+      "step": 123370
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.564316035407549e-06,
+      "loss": 0.6584,
+      "step": 123380
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.56102537102241e-06,
+      "loss": 0.6745,
+      "step": 123390
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.557734706637271e-06,
+      "loss": 0.5759,
+      "step": 123400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.554444042252131e-06,
+      "loss": 0.6979,
+      "step": 123410
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.551153377866992e-06,
+      "loss": 0.5339,
+      "step": 123420
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.547862713481852e-06,
+      "loss": 0.581,
+      "step": 123430
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.544572049096713e-06,
+      "loss": 0.6368,
+      "step": 123440
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.541281384711574e-06,
+      "loss": 0.6121,
+      "step": 123450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.537990720326434e-06,
+      "loss": 0.6317,
+      "step": 123460
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.534700055941295e-06,
+      "loss": 0.5163,
+      "step": 123470
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.531409391556155e-06,
+      "loss": 0.6083,
+      "step": 123480
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.528118727171016e-06,
+      "loss": 0.6445,
+      "step": 123490
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.524828062785877e-06,
+      "loss": 0.7251,
+      "step": 123500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.521537398400737e-06,
+      "loss": 0.5073,
+      "step": 123510
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.518246734015598e-06,
+      "loss": 0.6162,
+      "step": 123520
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514956069630458e-06,
+      "loss": 0.6317,
+      "step": 123530
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.511665405245319e-06,
+      "loss": 0.629,
+      "step": 123540
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.50837474086018e-06,
+      "loss": 0.7137,
+      "step": 123550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.50508407647504e-06,
+      "loss": 0.5943,
+      "step": 123560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.501793412089901e-06,
+      "loss": 0.54,
+      "step": 123570
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.49850274770476e-06,
+      "loss": 0.6832,
+      "step": 123580
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.495212083319622e-06,
+      "loss": 0.6143,
+      "step": 123590
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.491921418934483e-06,
+      "loss": 0.6296,
+      "step": 123600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.488630754549343e-06,
+      "loss": 0.6527,
+      "step": 123610
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.485340090164204e-06,
+      "loss": 0.6613,
+      "step": 123620
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.482049425779065e-06,
+      "loss": 0.6234,
+      "step": 123630
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.478758761393925e-06,
+      "loss": 0.6097,
+      "step": 123640
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.475468097008786e-06,
+      "loss": 0.6146,
+      "step": 123650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.472177432623647e-06,
+      "loss": 0.5814,
+      "step": 123660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.468886768238509e-06,
+      "loss": 0.5251,
+      "step": 123670
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.465596103853368e-06,
+      "loss": 0.4987,
+      "step": 123680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.46230543946823e-06,
+      "loss": 0.596,
+      "step": 123690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.459014775083091e-06,
+      "loss": 0.6738,
+      "step": 123700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.45572411069795e-06,
+      "loss": 0.5674,
+      "step": 123710
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.452433446312812e-06,
+      "loss": 0.5812,
+      "step": 123720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.449142781927673e-06,
+      "loss": 0.5795,
+      "step": 123730
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.445852117542533e-06,
+      "loss": 0.5087,
+      "step": 123740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.442561453157394e-06,
+      "loss": 0.6696,
+      "step": 123750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.439270788772253e-06,
+      "loss": 0.5786,
+      "step": 123760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.435980124387115e-06,
+      "loss": 0.5711,
+      "step": 123770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.432689460001976e-06,
+      "loss": 0.6276,
+      "step": 123780
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.429398795616835e-06,
+      "loss": 0.642,
+      "step": 123790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.426108131231697e-06,
+      "loss": 0.5783,
+      "step": 123800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.422817466846556e-06,
+      "loss": 0.5939,
+      "step": 123810
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.419526802461418e-06,
+      "loss": 0.7598,
+      "step": 123820
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.416236138076279e-06,
+      "loss": 0.5296,
+      "step": 123830
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.412945473691138e-06,
+      "loss": 0.8279,
+      "step": 123840
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.409654809306e-06,
+      "loss": 0.6068,
+      "step": 123850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.40636414492086e-06,
+      "loss": 0.6755,
+      "step": 123860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.40307348053572e-06,
+      "loss": 0.5128,
+      "step": 123870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.399782816150582e-06,
+      "loss": 0.5569,
+      "step": 123880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.396492151765441e-06,
+      "loss": 0.5647,
+      "step": 123890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.393201487380303e-06,
+      "loss": 0.6087,
+      "step": 123900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.389910822995162e-06,
+      "loss": 0.601,
+      "step": 123910
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.386620158610024e-06,
+      "loss": 0.5293,
+      "step": 123920
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.383329494224885e-06,
+      "loss": 0.5704,
+      "step": 123930
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.380038829839744e-06,
+      "loss": 0.5797,
+      "step": 123940
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.376748165454606e-06,
+      "loss": 0.5798,
+      "step": 123950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.373457501069465e-06,
+      "loss": 0.5817,
+      "step": 123960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.370166836684326e-06,
+      "loss": 0.6625,
+      "step": 123970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.366876172299188e-06,
+      "loss": 0.539,
+      "step": 123980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.363585507914047e-06,
+      "loss": 0.5487,
+      "step": 123990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.360294843528909e-06,
+      "loss": 0.6191,
+      "step": 124000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.35700417914377e-06,
+      "loss": 0.467,
+      "step": 124010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.35371351475863e-06,
+      "loss": 0.538,
+      "step": 124020
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.35042285037349e-06,
+      "loss": 0.6016,
+      "step": 124030
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.347132185988352e-06,
+      "loss": 0.5611,
+      "step": 124040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.343841521603213e-06,
+      "loss": 0.5658,
+      "step": 124050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.340550857218073e-06,
+      "loss": 0.5934,
+      "step": 124060
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.337260192832934e-06,
+      "loss": 0.5286,
+      "step": 124070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.333969528447795e-06,
+      "loss": 0.6156,
+      "step": 124080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.330678864062655e-06,
+      "loss": 0.6573,
+      "step": 124090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.327388199677516e-06,
+      "loss": 0.5868,
+      "step": 124100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.324097535292376e-06,
+      "loss": 0.6484,
+      "step": 124110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.320806870907237e-06,
+      "loss": 0.6185,
+      "step": 124120
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.317516206522098e-06,
+      "loss": 0.5228,
+      "step": 124130
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.314225542136958e-06,
+      "loss": 0.6331,
+      "step": 124140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.310934877751819e-06,
+      "loss": 0.6795,
+      "step": 124150
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.307644213366679e-06,
+      "loss": 0.7549,
+      "step": 124160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.30435354898154e-06,
+      "loss": 0.516,
+      "step": 124170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.301062884596401e-06,
+      "loss": 0.5966,
+      "step": 124180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.29777222021126e-06,
+      "loss": 0.5082,
+      "step": 124190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.294481555826122e-06,
+      "loss": 0.5902,
+      "step": 124200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.291190891440982e-06,
+      "loss": 0.5745,
+      "step": 124210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.287900227055843e-06,
+      "loss": 0.6503,
+      "step": 124220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.284609562670704e-06,
+      "loss": 0.5529,
+      "step": 124230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.281318898285564e-06,
+      "loss": 0.6042,
+      "step": 124240
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.278028233900425e-06,
+      "loss": 0.6004,
+      "step": 124250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.274737569515285e-06,
+      "loss": 0.5793,
+      "step": 124260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.271446905130146e-06,
+      "loss": 0.6964,
+      "step": 124270
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.268156240745007e-06,
+      "loss": 0.5567,
+      "step": 124280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.264865576359867e-06,
+      "loss": 0.531,
+      "step": 124290
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.261574911974728e-06,
+      "loss": 0.6914,
+      "step": 124300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.258284247589588e-06,
+      "loss": 0.6228,
+      "step": 124310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.254993583204449e-06,
+      "loss": 0.6878,
+      "step": 124320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.25170291881931e-06,
+      "loss": 0.6196,
+      "step": 124330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.24841225443417e-06,
+      "loss": 0.696,
+      "step": 124340
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.245121590049031e-06,
+      "loss": 0.6742,
+      "step": 124350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.24183092566389e-06,
+      "loss": 0.5853,
+      "step": 124360
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.238540261278752e-06,
+      "loss": 0.5796,
+      "step": 124370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.235249596893613e-06,
+      "loss": 0.6514,
+      "step": 124380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.231958932508474e-06,
+      "loss": 0.5854,
+      "step": 124390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.228668268123334e-06,
+      "loss": 0.543,
+      "step": 124400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.225377603738195e-06,
+      "loss": 0.5483,
+      "step": 124410
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.222086939353056e-06,
+      "loss": 0.6329,
+      "step": 124420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.218796274967918e-06,
+      "loss": 0.6602,
+      "step": 124430
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.215505610582777e-06,
+      "loss": 0.5806,
+      "step": 124440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.212214946197638e-06,
+      "loss": 0.5854,
+      "step": 124450
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.2089242818125e-06,
+      "loss": 0.5702,
+      "step": 124460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.20563361742736e-06,
+      "loss": 0.5244,
+      "step": 124470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.20234295304222e-06,
+      "loss": 0.5987,
+      "step": 124480
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.19905228865708e-06,
+      "loss": 0.6183,
+      "step": 124490
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.195761624271941e-06,
+      "loss": 0.5674,
+      "step": 124500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.192470959886803e-06,
+      "loss": 0.6923,
+      "step": 124510
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.189180295501662e-06,
+      "loss": 0.5963,
+      "step": 124520
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.185889631116524e-06,
+      "loss": 0.5315,
+      "step": 124530
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.182598966731383e-06,
+      "loss": 0.5493,
+      "step": 124540
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.179308302346244e-06,
+      "loss": 0.5794,
+      "step": 124550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.176017637961106e-06,
+      "loss": 0.5768,
+      "step": 124560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.172726973575965e-06,
+      "loss": 0.5626,
+      "step": 124570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.169436309190826e-06,
+      "loss": 0.5882,
+      "step": 124580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.166145644805686e-06,
+      "loss": 0.5624,
+      "step": 124590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.162854980420547e-06,
+      "loss": 0.6202,
+      "step": 124600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.159564316035409e-06,
+      "loss": 0.5475,
+      "step": 124610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.156273651650268e-06,
+      "loss": 0.619,
+      "step": 124620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.15298298726513e-06,
+      "loss": 0.6303,
+      "step": 124630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.149692322879989e-06,
+      "loss": 0.656,
+      "step": 124640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.14640165849485e-06,
+      "loss": 0.6048,
+      "step": 124650
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.143110994109712e-06,
+      "loss": 0.5749,
+      "step": 124660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.139820329724571e-06,
+      "loss": 0.5884,
+      "step": 124670
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.136529665339432e-06,
+      "loss": 0.5312,
+      "step": 124680
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.133239000954292e-06,
+      "loss": 0.6022,
+      "step": 124690
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.129948336569153e-06,
+      "loss": 0.5959,
+      "step": 124700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.126657672184015e-06,
+      "loss": 0.8263,
+      "step": 124710
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.123367007798874e-06,
+      "loss": 0.6393,
+      "step": 124720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.120076343413735e-06,
+      "loss": 0.5676,
+      "step": 124730
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.116785679028595e-06,
+      "loss": 0.602,
+      "step": 124740
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.113495014643456e-06,
+      "loss": 0.6079,
+      "step": 124750
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.110204350258317e-06,
+      "loss": 0.6015,
+      "step": 124760
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.106913685873179e-06,
+      "loss": 0.5945,
+      "step": 124770
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.103623021488038e-06,
+      "loss": 0.5549,
+      "step": 124780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.1003323571029e-06,
+      "loss": 0.6207,
+      "step": 124790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.09704169271776e-06,
+      "loss": 0.689,
+      "step": 124800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.093751028332622e-06,
+      "loss": 0.62,
+      "step": 124810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.090460363947482e-06,
+      "loss": 0.6212,
+      "step": 124820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.087169699562343e-06,
+      "loss": 0.5424,
+      "step": 124830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.083879035177203e-06,
+      "loss": 0.6253,
+      "step": 124840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.080588370792064e-06,
+      "loss": 0.5161,
+      "step": 124850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.077297706406925e-06,
+      "loss": 0.5367,
+      "step": 124860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.074007042021785e-06,
+      "loss": 0.7835,
+      "step": 124870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.070716377636646e-06,
+      "loss": 0.6075,
+      "step": 124880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.067425713251505e-06,
+      "loss": 0.6507,
+      "step": 124890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.064135048866367e-06,
+      "loss": 0.6187,
+      "step": 124900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.060844384481228e-06,
+      "loss": 0.7604,
+      "step": 124910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.057553720096088e-06,
+      "loss": 0.5122,
+      "step": 124920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.054263055710949e-06,
+      "loss": 0.6295,
+      "step": 124930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.050972391325808e-06,
+      "loss": 0.5578,
+      "step": 124940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.04768172694067e-06,
+      "loss": 0.6397,
+      "step": 124950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.044391062555531e-06,
+      "loss": 0.5315,
+      "step": 124960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.04110039817039e-06,
+      "loss": 0.7122,
+      "step": 124970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.037809733785252e-06,
+      "loss": 0.7773,
+      "step": 124980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.034519069400111e-06,
+      "loss": 0.6806,
+      "step": 124990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.031228405014973e-06,
+      "loss": 0.8611,
+      "step": 125000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.027937740629834e-06,
+      "loss": 0.6303,
+      "step": 125010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.024647076244693e-06,
+      "loss": 0.599,
+      "step": 125020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.021356411859555e-06,
+      "loss": 0.6183,
+      "step": 125030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.018065747474414e-06,
+      "loss": 0.5373,
+      "step": 125040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.014775083089276e-06,
+      "loss": 0.6228,
+      "step": 125050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.011484418704137e-06,
+      "loss": 0.6489,
+      "step": 125060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.008193754318996e-06,
+      "loss": 0.7283,
+      "step": 125070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.004903089933858e-06,
+      "loss": 0.6389,
+      "step": 125080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.001612425548717e-06,
+      "loss": 0.5278,
+      "step": 125090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.998321761163579e-06,
+      "loss": 0.5469,
+      "step": 125100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.99503109677844e-06,
+      "loss": 0.6026,
+      "step": 125110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.991740432393301e-06,
+      "loss": 0.6381,
+      "step": 125120
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.98844976800816e-06,
+      "loss": 0.5575,
+      "step": 125130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.985159103623022e-06,
+      "loss": 0.7242,
+      "step": 125140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.981868439237883e-06,
+      "loss": 0.5379,
+      "step": 125150
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.978577774852744e-06,
+      "loss": 0.5309,
+      "step": 125160
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.975287110467604e-06,
+      "loss": 0.7008,
+      "step": 125170
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.971996446082465e-06,
+      "loss": 0.6849,
+      "step": 125180
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.968705781697327e-06,
+      "loss": 0.6481,
+      "step": 125190
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.965415117312186e-06,
+      "loss": 0.635,
+      "step": 125200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.962124452927047e-06,
+      "loss": 0.5222,
+      "step": 125210
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.958833788541907e-06,
+      "loss": 0.5295,
+      "step": 125220
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.955543124156768e-06,
+      "loss": 0.597,
+      "step": 125230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.95225245977163e-06,
+      "loss": 0.6338,
+      "step": 125240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.948961795386489e-06,
+      "loss": 0.5981,
+      "step": 125250
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.94567113100135e-06,
+      "loss": 0.5406,
+      "step": 125260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.94238046661621e-06,
+      "loss": 0.6132,
+      "step": 125270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.939089802231071e-06,
+      "loss": 0.5293,
+      "step": 125280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.935799137845932e-06,
+      "loss": 0.5307,
+      "step": 125290
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.932508473460792e-06,
+      "loss": 0.636,
+      "step": 125300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.929217809075653e-06,
+      "loss": 0.6096,
+      "step": 125310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.925927144690513e-06,
+      "loss": 0.5642,
+      "step": 125320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.922636480305374e-06,
+      "loss": 0.6062,
+      "step": 125330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.919345815920235e-06,
+      "loss": 0.5758,
+      "step": 125340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.916055151535095e-06,
+      "loss": 0.587,
+      "step": 125350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.912764487149956e-06,
+      "loss": 0.6235,
+      "step": 125360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.909473822764816e-06,
+      "loss": 0.5968,
+      "step": 125370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.906183158379677e-06,
+      "loss": 0.5136,
+      "step": 125380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.902892493994538e-06,
+      "loss": 0.5472,
+      "step": 125390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.899601829609398e-06,
+      "loss": 0.5666,
+      "step": 125400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.89631116522426e-06,
+      "loss": 0.5709,
+      "step": 125410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.893020500839119e-06,
+      "loss": 0.6558,
+      "step": 125420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.88972983645398e-06,
+      "loss": 0.5834,
+      "step": 125430
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.886439172068841e-06,
+      "loss": 0.6734,
+      "step": 125440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.883148507683701e-06,
+      "loss": 0.5022,
+      "step": 125450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.879857843298562e-06,
+      "loss": 0.6113,
+      "step": 125460
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.876567178913422e-06,
+      "loss": 0.5691,
+      "step": 125470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.873276514528283e-06,
+      "loss": 0.5623,
+      "step": 125480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.869985850143144e-06,
+      "loss": 0.5561,
+      "step": 125490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.866695185758006e-06,
+      "loss": 0.5686,
+      "step": 125500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.863404521372865e-06,
+      "loss": 0.5235,
+      "step": 125510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.860113856987726e-06,
+      "loss": 0.6634,
+      "step": 125520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.856823192602588e-06,
+      "loss": 0.598,
+      "step": 125530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.853532528217449e-06,
+      "loss": 0.6112,
+      "step": 125540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.850241863832308e-06,
+      "loss": 0.5957,
+      "step": 125550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.84695119944717e-06,
+      "loss": 0.6508,
+      "step": 125560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.84366053506203e-06,
+      "loss": 0.6609,
+      "step": 125570
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.84036987067689e-06,
+      "loss": 0.5427,
+      "step": 125580
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.837079206291752e-06,
+      "loss": 0.5249,
+      "step": 125590
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.833788541906611e-06,
+      "loss": 0.5997,
+      "step": 125600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.830497877521473e-06,
+      "loss": 0.6309,
+      "step": 125610
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.827207213136332e-06,
+      "loss": 0.4987,
+      "step": 125620
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.823916548751194e-06,
+      "loss": 0.608,
+      "step": 125630
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.820625884366055e-06,
+      "loss": 0.6139,
+      "step": 125640
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.817335219980914e-06,
+      "loss": 0.7188,
+      "step": 125650
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.814044555595776e-06,
+      "loss": 0.5853,
+      "step": 125660
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.810753891210635e-06,
+      "loss": 0.6128,
+      "step": 125670
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.807463226825496e-06,
+      "loss": 0.7395,
+      "step": 125680
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.804172562440358e-06,
+      "loss": 0.5543,
+      "step": 125690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.800881898055217e-06,
+      "loss": 0.6958,
+      "step": 125700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.797591233670079e-06,
+      "loss": 0.7264,
+      "step": 125710
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.794300569284938e-06,
+      "loss": 0.5318,
+      "step": 125720
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.7910099048998e-06,
+      "loss": 0.5477,
+      "step": 125730
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.78771924051466e-06,
+      "loss": 0.621,
+      "step": 125740
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.78442857612952e-06,
+      "loss": 0.5993,
+      "step": 125750
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.781137911744382e-06,
+      "loss": 0.5704,
+      "step": 125760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.777847247359241e-06,
+      "loss": 0.7163,
+      "step": 125770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.774556582974102e-06,
+      "loss": 0.5596,
+      "step": 125780
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.771265918588964e-06,
+      "loss": 0.6057,
+      "step": 125790
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.767975254203823e-06,
+      "loss": 0.5848,
+      "step": 125800
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.764684589818684e-06,
+      "loss": 0.5743,
+      "step": 125810
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.761393925433546e-06,
+      "loss": 0.7732,
+      "step": 125820
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.758103261048405e-06,
+      "loss": 0.6617,
+      "step": 125830
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.754812596663267e-06,
+      "loss": 0.6138,
+      "step": 125840
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.751521932278126e-06,
+      "loss": 0.7226,
+      "step": 125850
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.748231267892987e-06,
+      "loss": 0.6285,
+      "step": 125860
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.744940603507849e-06,
+      "loss": 0.5899,
+      "step": 125870
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.74164993912271e-06,
+      "loss": 0.6819,
+      "step": 125880
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.73835927473757e-06,
+      "loss": 0.5975,
+      "step": 125890
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.73506861035243e-06,
+      "loss": 0.5804,
+      "step": 125900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.731777945967292e-06,
+      "loss": 0.4993,
+      "step": 125910
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.728487281582153e-06,
+      "loss": 0.4408,
+      "step": 125920
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.725196617197013e-06,
+      "loss": 0.5428,
+      "step": 125930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.721905952811874e-06,
+      "loss": 0.6409,
+      "step": 125940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.718615288426734e-06,
+      "loss": 0.5637,
+      "step": 125950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.715324624041595e-06,
+      "loss": 0.5214,
+      "step": 125960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.712033959656456e-06,
+      "loss": 0.7279,
+      "step": 125970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.708743295271316e-06,
+      "loss": 0.662,
+      "step": 125980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.705452630886177e-06,
+      "loss": 0.5616,
+      "step": 125990
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.702161966501037e-06,
+      "loss": 0.5153,
+      "step": 126000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.698871302115898e-06,
+      "loss": 0.6371,
+      "step": 126010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.69558063773076e-06,
+      "loss": 0.5737,
+      "step": 126020
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.692289973345619e-06,
+      "loss": 0.6193,
+      "step": 126030
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.68899930896048e-06,
+      "loss": 0.609,
+      "step": 126040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.68570864457534e-06,
+      "loss": 0.6097,
+      "step": 126050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.682417980190201e-06,
+      "loss": 0.6188,
+      "step": 126060
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.679127315805062e-06,
+      "loss": 0.5834,
+      "step": 126070
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.675836651419922e-06,
+      "loss": 0.5941,
+      "step": 126080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.672545987034783e-06,
+      "loss": 0.6486,
+      "step": 126090
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.669255322649643e-06,
+      "loss": 0.7572,
+      "step": 126100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.665964658264504e-06,
+      "loss": 0.5523,
+      "step": 126110
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.662673993879365e-06,
+      "loss": 0.4985,
+      "step": 126120
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.659383329494225e-06,
+      "loss": 0.6837,
+      "step": 126130
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.656092665109086e-06,
+      "loss": 0.5852,
+      "step": 126140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.652802000723946e-06,
+      "loss": 0.7095,
+      "step": 126150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.649511336338807e-06,
+      "loss": 0.6037,
+      "step": 126160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.646220671953668e-06,
+      "loss": 0.5887,
+      "step": 126170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.642930007568528e-06,
+      "loss": 0.5944,
+      "step": 126180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.639639343183389e-06,
+      "loss": 0.623,
+      "step": 126190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.636348678798248e-06,
+      "loss": 0.6976,
+      "step": 126200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.63305801441311e-06,
+      "loss": 0.5262,
+      "step": 126210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.629767350027971e-06,
+      "loss": 0.6088,
+      "step": 126220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.62647668564283e-06,
+      "loss": 0.6511,
+      "step": 126230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.623186021257692e-06,
+      "loss": 0.4632,
+      "step": 126240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.619895356872553e-06,
+      "loss": 0.5604,
+      "step": 126250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.616604692487414e-06,
+      "loss": 0.6812,
+      "step": 126260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.613314028102274e-06,
+      "loss": 0.5366,
+      "step": 126270
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.610023363717135e-06,
+      "loss": 0.609,
+      "step": 126280
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.606732699331997e-06,
+      "loss": 0.5343,
+      "step": 126290
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.603442034946856e-06,
+      "loss": 0.5686,
+      "step": 126300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.600151370561717e-06,
+      "loss": 0.4777,
+      "step": 126310
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.596860706176579e-06,
+      "loss": 0.4756,
+      "step": 126320
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.593570041791438e-06,
+      "loss": 0.6455,
+      "step": 126330
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.5902793774063e-06,
+      "loss": 0.6274,
+      "step": 126340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.586988713021159e-06,
+      "loss": 0.6631,
+      "step": 126350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.58369804863602e-06,
+      "loss": 0.5484,
+      "step": 126360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.580407384250882e-06,
+      "loss": 0.7247,
+      "step": 126370
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.577116719865741e-06,
+      "loss": 0.6474,
+      "step": 126380
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.573826055480602e-06,
+      "loss": 0.5372,
+      "step": 126390
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.570535391095462e-06,
+      "loss": 0.7356,
+      "step": 126400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.567244726710323e-06,
+      "loss": 0.6119,
+      "step": 126410
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.563954062325185e-06,
+      "loss": 0.5854,
+      "step": 126420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.560663397940044e-06,
+      "loss": 0.5833,
+      "step": 126430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.557372733554905e-06,
+      "loss": 0.5677,
+      "step": 126440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.554082069169765e-06,
+      "loss": 0.61,
+      "step": 126450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.550791404784626e-06,
+      "loss": 0.5923,
+      "step": 126460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.547500740399487e-06,
+      "loss": 0.6185,
+      "step": 126470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.544210076014347e-06,
+      "loss": 0.6121,
+      "step": 126480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.540919411629208e-06,
+      "loss": 0.7895,
+      "step": 126490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.537628747244068e-06,
+      "loss": 0.5253,
+      "step": 126500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.534338082858929e-06,
+      "loss": 0.5536,
+      "step": 126510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.53104741847379e-06,
+      "loss": 0.6434,
+      "step": 126520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.52775675408865e-06,
+      "loss": 0.6507,
+      "step": 126530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.524466089703511e-06,
+      "loss": 0.6301,
+      "step": 126540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.521175425318373e-06,
+      "loss": 0.4957,
+      "step": 126550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.517884760933232e-06,
+      "loss": 0.6091,
+      "step": 126560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.514594096548093e-06,
+      "loss": 0.5869,
+      "step": 126570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.511303432162953e-06,
+      "loss": 0.6022,
+      "step": 126580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.508012767777814e-06,
+      "loss": 0.6499,
+      "step": 126590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.504722103392675e-06,
+      "loss": 0.7083,
+      "step": 126600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.501431439007535e-06,
+      "loss": 0.6739,
+      "step": 126610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.498140774622396e-06,
+      "loss": 0.6331,
+      "step": 126620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.494850110237258e-06,
+      "loss": 0.4596,
+      "step": 126630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.491559445852119e-06,
+      "loss": 0.7514,
+      "step": 126640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.488268781466978e-06,
+      "loss": 0.5939,
+      "step": 126650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.48497811708184e-06,
+      "loss": 0.5483,
+      "step": 126660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.481687452696701e-06,
+      "loss": 0.5922,
+      "step": 126670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.47839678831156e-06,
+      "loss": 0.6375,
+      "step": 126680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.475106123926422e-06,
+      "loss": 0.5088,
+      "step": 126690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.471815459541283e-06,
+      "loss": 0.5689,
+      "step": 126700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.468524795156143e-06,
+      "loss": 0.6712,
+      "step": 126710
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.465234130771004e-06,
+      "loss": 0.5734,
+      "step": 126720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.461943466385863e-06,
+      "loss": 0.5892,
+      "step": 126730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.458652802000725e-06,
+      "loss": 0.542,
+      "step": 126740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.455362137615586e-06,
+      "loss": 0.5587,
+      "step": 126750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.452071473230446e-06,
+      "loss": 0.6014,
+      "step": 126760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.448780808845307e-06,
+      "loss": 0.6158,
+      "step": 126770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.445490144460166e-06,
+      "loss": 0.5742,
+      "step": 126780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.442199480075028e-06,
+      "loss": 0.6745,
+      "step": 126790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.438908815689889e-06,
+      "loss": 0.5683,
+      "step": 126800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.435618151304749e-06,
+      "loss": 0.6291,
+      "step": 126810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.43232748691961e-06,
+      "loss": 0.571,
+      "step": 126820
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.42903682253447e-06,
+      "loss": 0.5753,
+      "step": 126830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.42574615814933e-06,
+      "loss": 0.5878,
+      "step": 126840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.422455493764192e-06,
+      "loss": 0.643,
+      "step": 126850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.419164829379051e-06,
+      "loss": 0.6711,
+      "step": 126860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.415874164993913e-06,
+      "loss": 1.0219,
+      "step": 126870
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.412583500608772e-06,
+      "loss": 0.6576,
+      "step": 126880
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.409292836223634e-06,
+      "loss": 0.682,
+      "step": 126890
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.406002171838495e-06,
+      "loss": 0.5556,
+      "step": 126900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.402711507453354e-06,
+      "loss": 0.5767,
+      "step": 126910
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.399420843068216e-06,
+      "loss": 0.5846,
+      "step": 126920
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.396130178683075e-06,
+      "loss": 0.4963,
+      "step": 126930
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.392839514297937e-06,
+      "loss": 0.6238,
+      "step": 126940
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.389548849912798e-06,
+      "loss": 0.5453,
+      "step": 126950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.386258185527657e-06,
+      "loss": 0.6069,
+      "step": 126960
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.382967521142519e-06,
+      "loss": 0.6281,
+      "step": 126970
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.37967685675738e-06,
+      "loss": 0.522,
+      "step": 126980
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.37638619237224e-06,
+      "loss": 0.6256,
+      "step": 126990
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.3730955279871e-06,
+      "loss": 0.6294,
+      "step": 127000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.369804863601962e-06,
+      "loss": 0.6779,
+      "step": 127010
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.366514199216823e-06,
+      "loss": 0.6569,
+      "step": 127020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.363223534831683e-06,
+      "loss": 0.5906,
+      "step": 127030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.359932870446544e-06,
+      "loss": 0.6528,
+      "step": 127040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.356642206061405e-06,
+      "loss": 0.568,
+      "step": 127050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.353351541676265e-06,
+      "loss": 0.631,
+      "step": 127060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.350060877291126e-06,
+      "loss": 0.6178,
+      "step": 127070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.346770212905986e-06,
+      "loss": 0.5772,
+      "step": 127080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.343479548520847e-06,
+      "loss": 0.5591,
+      "step": 127090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.340188884135708e-06,
+      "loss": 0.614,
+      "step": 127100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.336898219750568e-06,
+      "loss": 0.6172,
+      "step": 127110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.33360755536543e-06,
+      "loss": 0.5166,
+      "step": 127120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.330316890980289e-06,
+      "loss": 0.6606,
+      "step": 127130
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.32702622659515e-06,
+      "loss": 0.5597,
+      "step": 127140
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.323735562210011e-06,
+      "loss": 0.6341,
+      "step": 127150
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.320444897824871e-06,
+      "loss": 0.6662,
+      "step": 127160
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.317154233439732e-06,
+      "loss": 0.5778,
+      "step": 127170
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.313863569054592e-06,
+      "loss": 0.5468,
+      "step": 127180
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.310572904669453e-06,
+      "loss": 0.5417,
+      "step": 127190
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.307282240284314e-06,
+      "loss": 0.611,
+      "step": 127200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.303991575899174e-06,
+      "loss": 0.5211,
+      "step": 127210
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.300700911514035e-06,
+      "loss": 0.6387,
+      "step": 127220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.297410247128895e-06,
+      "loss": 0.6065,
+      "step": 127230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.294119582743756e-06,
+      "loss": 0.6853,
+      "step": 127240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.290828918358617e-06,
+      "loss": 0.6359,
+      "step": 127250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.287538253973477e-06,
+      "loss": 0.5213,
+      "step": 127260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.284247589588338e-06,
+      "loss": 0.5691,
+      "step": 127270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.2809569252032e-06,
+      "loss": 0.6767,
+      "step": 127280
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.277666260818059e-06,
+      "loss": 0.5804,
+      "step": 127290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.27437559643292e-06,
+      "loss": 0.578,
+      "step": 127300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.27108493204778e-06,
+      "loss": 0.6094,
+      "step": 127310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.267794267662641e-06,
+      "loss": 0.5909,
+      "step": 127320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.264503603277502e-06,
+      "loss": 0.6249,
+      "step": 127330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.261212938892362e-06,
+      "loss": 0.7189,
+      "step": 127340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.257922274507223e-06,
+      "loss": 0.5881,
+      "step": 127350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.254631610122084e-06,
+      "loss": 0.6012,
+      "step": 127360
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.251340945736944e-06,
+      "loss": 0.7016,
+      "step": 127370
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.248050281351805e-06,
+      "loss": 0.7234,
+      "step": 127380
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.244759616966666e-06,
+      "loss": 0.6728,
+      "step": 127390
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.241468952581528e-06,
+      "loss": 0.6138,
+      "step": 127400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.238178288196387e-06,
+      "loss": 0.5696,
+      "step": 127410
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.234887623811249e-06,
+      "loss": 0.6807,
+      "step": 127420
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.23159695942611e-06,
+      "loss": 0.4872,
+      "step": 127430
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.22830629504097e-06,
+      "loss": 0.6372,
+      "step": 127440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.22501563065583e-06,
+      "loss": 0.5721,
+      "step": 127450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.22172496627069e-06,
+      "loss": 0.5116,
+      "step": 127460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.218434301885552e-06,
+      "loss": 0.543,
+      "step": 127470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.215143637500413e-06,
+      "loss": 0.4997,
+      "step": 127480
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.211852973115272e-06,
+      "loss": 0.5754,
+      "step": 127490
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.208562308730134e-06,
+      "loss": 0.649,
+      "step": 127500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.205271644344993e-06,
+      "loss": 0.6839,
+      "step": 127510
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.201980979959854e-06,
+      "loss": 0.4933,
+      "step": 127520
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.198690315574716e-06,
+      "loss": 0.622,
+      "step": 127530
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.195399651189575e-06,
+      "loss": 0.5776,
+      "step": 127540
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.192108986804437e-06,
+      "loss": 0.6353,
+      "step": 127550
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.188818322419296e-06,
+      "loss": 0.5833,
+      "step": 127560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.185527658034157e-06,
+      "loss": 0.8376,
+      "step": 127570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.182236993649019e-06,
+      "loss": 0.5426,
+      "step": 127580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.178946329263878e-06,
+      "loss": 0.5491,
+      "step": 127590
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.17565566487874e-06,
+      "loss": 0.5774,
+      "step": 127600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.172365000493599e-06,
+      "loss": 0.7043,
+      "step": 127610
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.16907433610846e-06,
+      "loss": 0.6056,
+      "step": 127620
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.165783671723322e-06,
+      "loss": 0.6201,
+      "step": 127630
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.162493007338181e-06,
+      "loss": 0.5829,
+      "step": 127640
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.159202342953042e-06,
+      "loss": 0.575,
+      "step": 127650
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.155911678567902e-06,
+      "loss": 0.6478,
+      "step": 127660
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.152621014182763e-06,
+      "loss": 0.5435,
+      "step": 127670
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.149330349797625e-06,
+      "loss": 0.7218,
+      "step": 127680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.146039685412484e-06,
+      "loss": 0.5827,
+      "step": 127690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.142749021027345e-06,
+      "loss": 0.4946,
+      "step": 127700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.139458356642205e-06,
+      "loss": 0.5705,
+      "step": 127710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.136167692257066e-06,
+      "loss": 0.5236,
+      "step": 127720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.132877027871928e-06,
+      "loss": 0.5387,
+      "step": 127730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.129586363486789e-06,
+      "loss": 0.6133,
+      "step": 127740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.126295699101648e-06,
+      "loss": 0.602,
+      "step": 127750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.12300503471651e-06,
+      "loss": 0.5408,
+      "step": 127760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.119714370331371e-06,
+      "loss": 0.643,
+      "step": 127770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.116423705946232e-06,
+      "loss": 0.6871,
+      "step": 127780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.113133041561092e-06,
+      "loss": 0.5791,
+      "step": 127790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.109842377175953e-06,
+      "loss": 0.568,
+      "step": 127800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.106551712790813e-06,
+      "loss": 0.5734,
+      "step": 127810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.103261048405674e-06,
+      "loss": 0.5615,
+      "step": 127820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.099970384020535e-06,
+      "loss": 0.6503,
+      "step": 127830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.096679719635395e-06,
+      "loss": 0.5702,
+      "step": 127840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.093389055250256e-06,
+      "loss": 0.4711,
+      "step": 127850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.090098390865116e-06,
+      "loss": 0.6285,
+      "step": 127860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.086807726479977e-06,
+      "loss": 0.5891,
+      "step": 127870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.083517062094838e-06,
+      "loss": 0.6439,
+      "step": 127880
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.080226397709698e-06,
+      "loss": 0.7974,
+      "step": 127890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.076935733324559e-06,
+      "loss": 0.5807,
+      "step": 127900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.073645068939418e-06,
+      "loss": 0.5374,
+      "step": 127910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.07035440455428e-06,
+      "loss": 0.5055,
+      "step": 127920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.067063740169141e-06,
+      "loss": 0.616,
+      "step": 127930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.063773075784e-06,
+      "loss": 0.518,
+      "step": 127940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.060482411398862e-06,
+      "loss": 0.6094,
+      "step": 127950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.057191747013723e-06,
+      "loss": 0.6111,
+      "step": 127960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.053901082628583e-06,
+      "loss": 0.5863,
+      "step": 127970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.050610418243444e-06,
+      "loss": 0.5871,
+      "step": 127980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.047319753858304e-06,
+      "loss": 0.5591,
+      "step": 127990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.044029089473165e-06,
+      "loss": 0.8829,
+      "step": 128000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.040738425088026e-06,
+      "loss": 0.5448,
+      "step": 128010
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.037447760702886e-06,
+      "loss": 0.588,
+      "step": 128020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.034157096317747e-06,
+      "loss": 0.6643,
+      "step": 128030
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.030866431932607e-06,
+      "loss": 0.5559,
+      "step": 128040
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.027575767547468e-06,
+      "loss": 0.5904,
+      "step": 128050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.024285103162329e-06,
+      "loss": 0.7188,
+      "step": 128060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.020994438777189e-06,
+      "loss": 0.5733,
+      "step": 128070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.01770377439205e-06,
+      "loss": 0.5324,
+      "step": 128080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.014413110006911e-06,
+      "loss": 0.5696,
+      "step": 128090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.01112244562177e-06,
+      "loss": 0.5827,
+      "step": 128100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.007831781236632e-06,
+      "loss": 0.6412,
+      "step": 128110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.004541116851493e-06,
+      "loss": 0.6256,
+      "step": 128120
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.001250452466355e-06,
+      "loss": 0.5826,
+      "step": 128130
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.997959788081214e-06,
+      "loss": 0.5838,
+      "step": 128140
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.994669123696075e-06,
+      "loss": 0.5872,
+      "step": 128150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.991378459310937e-06,
+      "loss": 0.6291,
+      "step": 128160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.988087794925796e-06,
+      "loss": 0.6452,
+      "step": 128170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.984797130540657e-06,
+      "loss": 0.539,
+      "step": 128180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.981506466155517e-06,
+      "loss": 0.5612,
+      "step": 128190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.978215801770378e-06,
+      "loss": 0.589,
+      "step": 128200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.97492513738524e-06,
+      "loss": 0.8155,
+      "step": 128210
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.9716344730001e-06,
+      "loss": 0.5411,
+      "step": 128220
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.96834380861496e-06,
+      "loss": 0.7642,
+      "step": 128230
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.96505314422982e-06,
+      "loss": 0.5822,
+      "step": 128240
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.961762479844681e-06,
+      "loss": 0.7679,
+      "step": 128250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.958471815459543e-06,
+      "loss": 0.7235,
+      "step": 128260
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.955181151074402e-06,
+      "loss": 0.6413,
+      "step": 128270
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.951890486689263e-06,
+      "loss": 0.6054,
+      "step": 128280
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.948599822304123e-06,
+      "loss": 0.7131,
+      "step": 128290
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.945309157918984e-06,
+      "loss": 0.5514,
+      "step": 128300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.942018493533845e-06,
+      "loss": 0.6331,
+      "step": 128310
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.938727829148705e-06,
+      "loss": 0.684,
+      "step": 128320
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.935437164763566e-06,
+      "loss": 0.7056,
+      "step": 128330
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.932146500378426e-06,
+      "loss": 0.4993,
+      "step": 128340
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.928855835993287e-06,
+      "loss": 0.569,
+      "step": 128350
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.925565171608148e-06,
+      "loss": 0.684,
+      "step": 128360
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.922274507223008e-06,
+      "loss": 0.5709,
+      "step": 128370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.91898384283787e-06,
+      "loss": 0.615,
+      "step": 128380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.915693178452729e-06,
+      "loss": 0.5345,
+      "step": 128390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.91240251406759e-06,
+      "loss": 0.4512,
+      "step": 128400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.909111849682451e-06,
+      "loss": 0.5612,
+      "step": 128410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.905821185297311e-06,
+      "loss": 0.5116,
+      "step": 128420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.902530520912172e-06,
+      "loss": 0.6588,
+      "step": 128430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.899239856527032e-06,
+      "loss": 0.601,
+      "step": 128440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.895949192141893e-06,
+      "loss": 0.6846,
+      "step": 128450
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.892658527756754e-06,
+      "loss": 0.6721,
+      "step": 128460
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.889367863371616e-06,
+      "loss": 0.6296,
+      "step": 128470
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.886077198986475e-06,
+      "loss": 0.7389,
+      "step": 128480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.882786534601336e-06,
+      "loss": 0.6085,
+      "step": 128490
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.879495870216198e-06,
+      "loss": 0.847,
+      "step": 128500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.876205205831059e-06,
+      "loss": 0.5482,
+      "step": 128510
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.872914541445919e-06,
+      "loss": 0.6012,
+      "step": 128520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.86962387706078e-06,
+      "loss": 0.6169,
+      "step": 128530
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.86633321267564e-06,
+      "loss": 0.5873,
+      "step": 128540
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.8630425482905e-06,
+      "loss": 0.5206,
+      "step": 128550
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.859751883905362e-06,
+      "loss": 0.6562,
+      "step": 128560
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.856461219520221e-06,
+      "loss": 0.6762,
+      "step": 128570
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.853170555135083e-06,
+      "loss": 0.6276,
+      "step": 128580
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.849879890749942e-06,
+      "loss": 0.6758,
+      "step": 128590
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.846589226364804e-06,
+      "loss": 0.6296,
+      "step": 128600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.843298561979665e-06,
+      "loss": 0.6064,
+      "step": 128610
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.840007897594524e-06,
+      "loss": 0.5946,
+      "step": 128620
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.836717233209386e-06,
+      "loss": 0.5331,
+      "step": 128630
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.833426568824245e-06,
+      "loss": 0.6103,
+      "step": 128640
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.830135904439107e-06,
+      "loss": 0.5981,
+      "step": 128650
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.826845240053968e-06,
+      "loss": 0.5139,
+      "step": 128660
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.823554575668827e-06,
+      "loss": 0.6298,
+      "step": 128670
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.820263911283689e-06,
+      "loss": 0.6903,
+      "step": 128680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.81697324689855e-06,
+      "loss": 0.4862,
+      "step": 128690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.81368258251341e-06,
+      "loss": 0.6382,
+      "step": 128700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.81039191812827e-06,
+      "loss": 0.6356,
+      "step": 128710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.80710125374313e-06,
+      "loss": 0.6333,
+      "step": 128720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.803810589357992e-06,
+      "loss": 0.6073,
+      "step": 128730
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.800519924972853e-06,
+      "loss": 0.642,
+      "step": 128740
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.797229260587712e-06,
+      "loss": 0.5712,
+      "step": 128750
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.793938596202574e-06,
+      "loss": 0.5781,
+      "step": 128760
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.790647931817433e-06,
+      "loss": 0.6014,
+      "step": 128770
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.787357267432295e-06,
+      "loss": 0.5713,
+      "step": 128780
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.784066603047156e-06,
+      "loss": 0.6156,
+      "step": 128790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.780775938662015e-06,
+      "loss": 0.7266,
+      "step": 128800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.777485274276877e-06,
+      "loss": 0.5679,
+      "step": 128810
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.774194609891736e-06,
+      "loss": 0.753,
+      "step": 128820
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.770903945506598e-06,
+      "loss": 0.5193,
+      "step": 128830
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.767613281121459e-06,
+      "loss": 0.6512,
+      "step": 128840
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.76432261673632e-06,
+      "loss": 0.6827,
+      "step": 128850
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.76103195235118e-06,
+      "loss": 0.577,
+      "step": 128860
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.757741287966041e-06,
+      "loss": 0.6313,
+      "step": 128870
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.754450623580902e-06,
+      "loss": 0.5839,
+      "step": 128880
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.751159959195763e-06,
+      "loss": 0.7009,
+      "step": 128890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.747869294810623e-06,
+      "loss": 0.6557,
+      "step": 128900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.744578630425484e-06,
+      "loss": 0.7075,
+      "step": 128910
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.741287966040344e-06,
+      "loss": 0.5263,
+      "step": 128920
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.737997301655205e-06,
+      "loss": 0.5203,
+      "step": 128930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.734706637270066e-06,
+      "loss": 0.5074,
+      "step": 128940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.731415972884926e-06,
+      "loss": 0.6835,
+      "step": 128950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.728125308499787e-06,
+      "loss": 0.5655,
+      "step": 128960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.724834644114647e-06,
+      "loss": 0.5994,
+      "step": 128970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.721543979729508e-06,
+      "loss": 0.6855,
+      "step": 128980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.71825331534437e-06,
+      "loss": 0.4542,
+      "step": 128990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.714962650959229e-06,
+      "loss": 0.5127,
+      "step": 129000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.71167198657409e-06,
+      "loss": 0.5887,
+      "step": 129010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.70838132218895e-06,
+      "loss": 0.7229,
+      "step": 129020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.705090657803811e-06,
+      "loss": 0.5801,
+      "step": 129030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.701799993418672e-06,
+      "loss": 0.6172,
+      "step": 129040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.698509329033532e-06,
+      "loss": 0.5796,
+      "step": 129050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.695218664648393e-06,
+      "loss": 0.5829,
+      "step": 129060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.691928000263253e-06,
+      "loss": 0.5723,
+      "step": 129070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.688637335878114e-06,
+      "loss": 0.5334,
+      "step": 129080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.685346671492975e-06,
+      "loss": 0.556,
+      "step": 129090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.682056007107835e-06,
+      "loss": 0.5829,
+      "step": 129100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.678765342722696e-06,
+      "loss": 0.486,
+      "step": 129110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.675474678337556e-06,
+      "loss": 0.6291,
+      "step": 129120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.672184013952417e-06,
+      "loss": 0.644,
+      "step": 129130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.668893349567278e-06,
+      "loss": 0.5631,
+      "step": 129140
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.665602685182138e-06,
+      "loss": 0.5737,
+      "step": 129150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.662312020796999e-06,
+      "loss": 0.7134,
+      "step": 129160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.659021356411859e-06,
+      "loss": 0.547,
+      "step": 129170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.65573069202672e-06,
+      "loss": 0.6429,
+      "step": 129180
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.652440027641581e-06,
+      "loss": 0.5674,
+      "step": 129190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.64914936325644e-06,
+      "loss": 0.4968,
+      "step": 129200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.645858698871302e-06,
+      "loss": 0.5033,
+      "step": 129210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.642568034486163e-06,
+      "loss": 0.583,
+      "step": 129220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.639277370101024e-06,
+      "loss": 0.5888,
+      "step": 129230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.635986705715884e-06,
+      "loss": 0.5183,
+      "step": 129240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.632696041330745e-06,
+      "loss": 0.5957,
+      "step": 129250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.629405376945607e-06,
+      "loss": 0.5667,
+      "step": 129260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.626114712560465e-06,
+      "loss": 0.5604,
+      "step": 129270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.622824048175327e-06,
+      "loss": 0.6415,
+      "step": 129280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.619533383790188e-06,
+      "loss": 0.5947,
+      "step": 129290
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.6162427194050474e-06,
+      "loss": 0.5783,
+      "step": 129300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.612952055019909e-06,
+      "loss": 0.6337,
+      "step": 129310
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.609661390634769e-06,
+      "loss": 0.6203,
+      "step": 129320
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.60637072624963e-06,
+      "loss": 0.5683,
+      "step": 129330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.603080061864492e-06,
+      "loss": 0.6395,
+      "step": 129340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.599789397479351e-06,
+      "loss": 0.5407,
+      "step": 129350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5964987330942125e-06,
+      "loss": 0.667,
+      "step": 129360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.593208068709074e-06,
+      "loss": 0.58,
+      "step": 129370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.589917404323933e-06,
+      "loss": 0.5805,
+      "step": 129380
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.586626739938795e-06,
+      "loss": 0.5375,
+      "step": 129390
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.583336075553654e-06,
+      "loss": 0.6586,
+      "step": 129400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5800454111685154e-06,
+      "loss": 0.5389,
+      "step": 129410
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.576754746783377e-06,
+      "loss": 0.6351,
+      "step": 129420
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.573464082398236e-06,
+      "loss": 0.6189,
+      "step": 129430
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5701734180130976e-06,
+      "loss": 0.5528,
+      "step": 129440
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.566882753627957e-06,
+      "loss": 0.7455,
+      "step": 129450
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.563592089242818e-06,
+      "loss": 0.4978,
+      "step": 129460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.56030142485768e-06,
+      "loss": 0.5807,
+      "step": 129470
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.557010760472539e-06,
+      "loss": 0.5387,
+      "step": 129480
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5537200960874005e-06,
+      "loss": 0.5804,
+      "step": 129490
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.550429431702261e-06,
+      "loss": 0.7469,
+      "step": 129500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.547138767317122e-06,
+      "loss": 0.547,
+      "step": 129510
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.543848102931983e-06,
+      "loss": 0.6759,
+      "step": 129520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.540557438546843e-06,
+      "loss": 0.7848,
+      "step": 129530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.537266774161704e-06,
+      "loss": 0.5892,
+      "step": 129540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.533976109776564e-06,
+      "loss": 0.5247,
+      "step": 129550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.530685445391425e-06,
+      "loss": 0.6007,
+      "step": 129560
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.527394781006286e-06,
+      "loss": 0.5411,
+      "step": 129570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.524104116621146e-06,
+      "loss": 0.621,
+      "step": 129580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.520813452236007e-06,
+      "loss": 0.6102,
+      "step": 129590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.517522787850867e-06,
+      "loss": 0.5218,
+      "step": 129600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.514232123465728e-06,
+      "loss": 0.6481,
+      "step": 129610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.510941459080589e-06,
+      "loss": 0.5757,
+      "step": 129620
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.507650794695449e-06,
+      "loss": 0.5501,
+      "step": 129630
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.50436013031031e-06,
+      "loss": 0.5968,
+      "step": 129640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.50106946592517e-06,
+      "loss": 0.6337,
+      "step": 129650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.497778801540031e-06,
+      "loss": 0.6558,
+      "step": 129660
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.494488137154892e-06,
+      "loss": 0.6919,
+      "step": 129670
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.491197472769753e-06,
+      "loss": 0.5612,
+      "step": 129680
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.487906808384613e-06,
+      "loss": 0.5883,
+      "step": 129690
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4846161439994736e-06,
+      "loss": 0.6457,
+      "step": 129700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.481325479614335e-06,
+      "loss": 0.4895,
+      "step": 129710
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.478034815229196e-06,
+      "loss": 0.6315,
+      "step": 129720
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.474744150844056e-06,
+      "loss": 0.616,
+      "step": 129730
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.471453486458917e-06,
+      "loss": 0.5186,
+      "step": 129740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4681628220737765e-06,
+      "loss": 0.5709,
+      "step": 129750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.464872157688638e-06,
+      "loss": 0.5241,
+      "step": 129760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.461581493303499e-06,
+      "loss": 0.5042,
+      "step": 129770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.458290828918359e-06,
+      "loss": 0.5947,
+      "step": 129780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.45500016453322e-06,
+      "loss": 0.723,
+      "step": 129790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4517095001480795e-06,
+      "loss": 0.627,
+      "step": 129800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.448418835762941e-06,
+      "loss": 0.6652,
+      "step": 129810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.445128171377802e-06,
+      "loss": 0.6624,
+      "step": 129820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4418375069926616e-06,
+      "loss": 0.6085,
+      "step": 129830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.438546842607523e-06,
+      "loss": 0.6996,
+      "step": 129840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.435256178222383e-06,
+      "loss": 0.658,
+      "step": 129850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.431965513837244e-06,
+      "loss": 0.5474,
+      "step": 129860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.428674849452105e-06,
+      "loss": 0.5751,
+      "step": 129870
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.425384185066965e-06,
+      "loss": 0.5103,
+      "step": 129880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.422093520681827e-06,
+      "loss": 0.6205,
+      "step": 129890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.418802856296686e-06,
+      "loss": 0.6025,
+      "step": 129900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.4155121919115475e-06,
+      "loss": 0.6016,
+      "step": 129910
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.412221527526409e-06,
+      "loss": 0.5687,
+      "step": 129920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.408930863141268e-06,
+      "loss": 0.7657,
+      "step": 129930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.40564019875613e-06,
+      "loss": 0.537,
+      "step": 129940
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.402349534370989e-06,
+      "loss": 0.6073,
+      "step": 129950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3990588699858504e-06,
+      "loss": 0.5812,
+      "step": 129960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.395768205600712e-06,
+      "loss": 0.6573,
+      "step": 129970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.392477541215571e-06,
+      "loss": 0.5937,
+      "step": 129980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3891868768304325e-06,
+      "loss": 0.426,
+      "step": 129990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.385896212445292e-06,
+      "loss": 0.6237,
+      "step": 130000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.382605548060153e-06,
+      "loss": 0.6888,
+      "step": 130010
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.379314883675015e-06,
+      "loss": 0.6397,
+      "step": 130020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.376024219289874e-06,
+      "loss": 0.4985,
+      "step": 130030
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3727335549047355e-06,
+      "loss": 0.5234,
+      "step": 130040
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.369442890519596e-06,
+      "loss": 0.5864,
+      "step": 130050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.366152226134457e-06,
+      "loss": 0.6559,
+      "step": 130060
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.362861561749318e-06,
+      "loss": 0.6337,
+      "step": 130070
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.359570897364178e-06,
+      "loss": 0.6153,
+      "step": 130080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.356280232979039e-06,
+      "loss": 0.5889,
+      "step": 130090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3529895685939005e-06,
+      "loss": 0.463,
+      "step": 130100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.34969890420876e-06,
+      "loss": 0.5007,
+      "step": 130110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.346408239823621e-06,
+      "loss": 0.5282,
+      "step": 130120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.343117575438481e-06,
+      "loss": 0.5922,
+      "step": 130130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.339826911053342e-06,
+      "loss": 0.6672,
+      "step": 130140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3365362466682035e-06,
+      "loss": 0.5877,
+      "step": 130150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.333245582283063e-06,
+      "loss": 0.5738,
+      "step": 130160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.329954917897924e-06,
+      "loss": 0.5559,
+      "step": 130170
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.326664253512784e-06,
+      "loss": 0.9061,
+      "step": 130180
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.323373589127645e-06,
+      "loss": 0.5417,
+      "step": 130190
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3200829247425064e-06,
+      "loss": 0.5428,
+      "step": 130200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.316792260357366e-06,
+      "loss": 0.6008,
+      "step": 130210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.313501595972227e-06,
+      "loss": 0.676,
+      "step": 130220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.310210931587088e-06,
+      "loss": 0.5517,
+      "step": 130230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.306920267201948e-06,
+      "loss": 0.6103,
+      "step": 130240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.303629602816809e-06,
+      "loss": 0.6319,
+      "step": 130250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.30033893843167e-06,
+      "loss": 0.532,
+      "step": 130260
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.297048274046531e-06,
+      "loss": 0.6073,
+      "step": 130270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.293757609661391e-06,
+      "loss": 0.6396,
+      "step": 130280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.290466945276252e-06,
+      "loss": 0.5364,
+      "step": 130290
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.287176280891113e-06,
+      "loss": 0.5619,
+      "step": 130300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.283885616505973e-06,
+      "loss": 0.9174,
+      "step": 130310
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.280594952120834e-06,
+      "loss": 0.5891,
+      "step": 130320
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.277304287735694e-06,
+      "loss": 0.4835,
+      "step": 130330
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.274013623350555e-06,
+      "loss": 0.5828,
+      "step": 130340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.270722958965416e-06,
+      "loss": 0.5696,
+      "step": 130350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.267432294580276e-06,
+      "loss": 0.7059,
+      "step": 130360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.264141630195137e-06,
+      "loss": 0.6656,
+      "step": 130370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2608509658099966e-06,
+      "loss": 0.6028,
+      "step": 130380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.257560301424858e-06,
+      "loss": 0.5056,
+      "step": 130390
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.254269637039719e-06,
+      "loss": 0.4375,
+      "step": 130400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.250978972654579e-06,
+      "loss": 0.5611,
+      "step": 130410
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.24768830826944e-06,
+      "loss": 0.5687,
+      "step": 130420
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2443976438843e-06,
+      "loss": 0.478,
+      "step": 130430
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.241106979499162e-06,
+      "loss": 0.6267,
+      "step": 130440
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.237816315114022e-06,
+      "loss": 0.6098,
+      "step": 130450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2345256507288825e-06,
+      "loss": 0.666,
+      "step": 130460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.231234986343744e-06,
+      "loss": 0.5209,
+      "step": 130470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.227944321958603e-06,
+      "loss": 0.7667,
+      "step": 130480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2246536575734646e-06,
+      "loss": 0.6252,
+      "step": 130490
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.221362993188326e-06,
+      "loss": 0.5662,
+      "step": 130500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.218072328803185e-06,
+      "loss": 0.5978,
+      "step": 130510
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.214781664418047e-06,
+      "loss": 0.5439,
+      "step": 130520
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.211491000032906e-06,
+      "loss": 0.7099,
+      "step": 130530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2082003356477675e-06,
+      "loss": 0.6555,
+      "step": 130540
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.204909671262629e-06,
+      "loss": 0.5896,
+      "step": 130550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.201619006877488e-06,
+      "loss": 0.6166,
+      "step": 130560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.19832834249235e-06,
+      "loss": 0.5529,
+      "step": 130570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.195037678107209e-06,
+      "loss": 0.5603,
+      "step": 130580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1917470137220705e-06,
+      "loss": 0.5273,
+      "step": 130590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.188456349336932e-06,
+      "loss": 0.4742,
+      "step": 130600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.185165684951792e-06,
+      "loss": 0.5002,
+      "step": 130610
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1818750205666526e-06,
+      "loss": 0.5077,
+      "step": 130620
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.178584356181513e-06,
+      "loss": 0.6749,
+      "step": 130630
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.175293691796374e-06,
+      "loss": 0.6173,
+      "step": 130640
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1720030274112355e-06,
+      "loss": 0.6142,
+      "step": 130650
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.168712363026095e-06,
+      "loss": 0.6053,
+      "step": 130660
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.165421698640956e-06,
+      "loss": 0.6979,
+      "step": 130670
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.162131034255816e-06,
+      "loss": 0.5853,
+      "step": 130680
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.158840369870677e-06,
+      "loss": 0.6186,
+      "step": 130690
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1555497054855385e-06,
+      "loss": 0.6252,
+      "step": 130700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.152259041100398e-06,
+      "loss": 0.5632,
+      "step": 130710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.148968376715259e-06,
+      "loss": 0.5435,
+      "step": 130720
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.145677712330119e-06,
+      "loss": 0.6367,
+      "step": 130730
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.14238704794498e-06,
+      "loss": 0.6059,
+      "step": 130740
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1390963835598414e-06,
+      "loss": 0.5648,
+      "step": 130750
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.135805719174701e-06,
+      "loss": 0.6885,
+      "step": 130760
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.132515054789562e-06,
+      "loss": 0.6023,
+      "step": 130770
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.129224390404423e-06,
+      "loss": 0.5696,
+      "step": 130780
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.125933726019283e-06,
+      "loss": 0.6471,
+      "step": 130790
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.122643061634144e-06,
+      "loss": 0.6061,
+      "step": 130800
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.119352397249005e-06,
+      "loss": 0.5136,
+      "step": 130810
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.116061732863866e-06,
+      "loss": 0.6093,
+      "step": 130820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1127710684787265e-06,
+      "loss": 0.627,
+      "step": 130830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.109480404093587e-06,
+      "loss": 0.7154,
+      "step": 130840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.106189739708448e-06,
+      "loss": 0.736,
+      "step": 130850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.102899075323308e-06,
+      "loss": 0.4719,
+      "step": 130860
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.099608410938169e-06,
+      "loss": 0.6784,
+      "step": 130870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.09631774655303e-06,
+      "loss": 0.5779,
+      "step": 130880
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.09302708216789e-06,
+      "loss": 0.5717,
+      "step": 130890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.089736417782751e-06,
+      "loss": 0.514,
+      "step": 130900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.086445753397611e-06,
+      "loss": 0.6024,
+      "step": 130910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.083155089012472e-06,
+      "loss": 0.6215,
+      "step": 130920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.079864424627333e-06,
+      "loss": 0.5619,
+      "step": 130930
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.076573760242193e-06,
+      "loss": 0.5894,
+      "step": 130940
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.073283095857054e-06,
+      "loss": 0.4904,
+      "step": 130950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.069992431471914e-06,
+      "loss": 0.5138,
+      "step": 130960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.066701767086775e-06,
+      "loss": 0.5812,
+      "step": 130970
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.063411102701636e-06,
+      "loss": 0.5988,
+      "step": 130980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.060120438316497e-06,
+      "loss": 0.5554,
+      "step": 130990
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.056829773931357e-06,
+      "loss": 0.6302,
+      "step": 131000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0535391095462174e-06,
+      "loss": 0.6233,
+      "step": 131010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.050248445161079e-06,
+      "loss": 0.6063,
+      "step": 131020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.04695778077594e-06,
+      "loss": 0.5131,
+      "step": 131030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0436671163907995e-06,
+      "loss": 0.6529,
+      "step": 131040
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.040376452005661e-06,
+      "loss": 0.6651,
+      "step": 131050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.03708578762052e-06,
+      "loss": 0.6527,
+      "step": 131060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.033795123235382e-06,
+      "loss": 0.7298,
+      "step": 131070
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.030504458850243e-06,
+      "loss": 0.5785,
+      "step": 131080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0272137944651025e-06,
+      "loss": 0.5722,
+      "step": 131090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.023923130079964e-06,
+      "loss": 0.5439,
+      "step": 131100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.020632465694823e-06,
+      "loss": 0.706,
+      "step": 131110
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.017341801309685e-06,
+      "loss": 0.5624,
+      "step": 131120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.014051136924546e-06,
+      "loss": 0.5626,
+      "step": 131130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0107604725394054e-06,
+      "loss": 0.6495,
+      "step": 131140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.007469808154267e-06,
+      "loss": 0.5471,
+      "step": 131150
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.004179143769127e-06,
+      "loss": 0.6997,
+      "step": 131160
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0008884793839876e-06,
+      "loss": 0.5448,
+      "step": 131170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.997597814998849e-06,
+      "loss": 0.6994,
+      "step": 131180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.994307150613709e-06,
+      "loss": 0.6587,
+      "step": 131190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9910164862285705e-06,
+      "loss": 0.6031,
+      "step": 131200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.98772582184343e-06,
+      "loss": 0.6352,
+      "step": 131210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.984435157458291e-06,
+      "loss": 0.6773,
+      "step": 131220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.981144493073153e-06,
+      "loss": 0.606,
+      "step": 131230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.977853828688012e-06,
+      "loss": 0.5619,
+      "step": 131240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9745631643028735e-06,
+      "loss": 0.6717,
+      "step": 131250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.971272499917733e-06,
+      "loss": 0.6858,
+      "step": 131260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.967981835532594e-06,
+      "loss": 0.61,
+      "step": 131270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9646911711474556e-06,
+      "loss": 0.6698,
+      "step": 131280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.961400506762315e-06,
+      "loss": 0.4277,
+      "step": 131290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.958109842377176e-06,
+      "loss": 0.6142,
+      "step": 131300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.954819177992036e-06,
+      "loss": 0.5122,
+      "step": 131310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.951528513606897e-06,
+      "loss": 0.5722,
+      "step": 131320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9482378492217585e-06,
+      "loss": 0.507,
+      "step": 131330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.944947184836618e-06,
+      "loss": 0.6068,
+      "step": 131340
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.941656520451479e-06,
+      "loss": 0.611,
+      "step": 131350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.93836585606634e-06,
+      "loss": 0.564,
+      "step": 131360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.935075191681201e-06,
+      "loss": 0.5931,
+      "step": 131370
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.9317845272960615e-06,
+      "loss": 0.6123,
+      "step": 131380
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.928493862910922e-06,
+      "loss": 0.5193,
+      "step": 131390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.925203198525783e-06,
+      "loss": 0.6193,
+      "step": 131400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.921912534140643e-06,
+      "loss": 0.5449,
+      "step": 131410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.918621869755504e-06,
+      "loss": 0.5884,
+      "step": 131420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.915331205370365e-06,
+      "loss": 0.6023,
+      "step": 131430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.912040540985225e-06,
+      "loss": 0.5458,
+      "step": 131440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.908749876600086e-06,
+      "loss": 0.5883,
+      "step": 131450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.905459212214946e-06,
+      "loss": 0.6246,
+      "step": 131460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.902168547829807e-06,
+      "loss": 0.5613,
+      "step": 131470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.898877883444668e-06,
+      "loss": 0.6184,
+      "step": 131480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.895587219059528e-06,
+      "loss": 0.5203,
+      "step": 131490
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.892296554674389e-06,
+      "loss": 0.5999,
+      "step": 131500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.88900589028925e-06,
+      "loss": 0.4977,
+      "step": 131510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.88571522590411e-06,
+      "loss": 0.6511,
+      "step": 131520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.882424561518971e-06,
+      "loss": 0.552,
+      "step": 131530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.879133897133832e-06,
+      "loss": 0.4812,
+      "step": 131540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.875843232748692e-06,
+      "loss": 0.5702,
+      "step": 131550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.872552568363553e-06,
+      "loss": 0.545,
+      "step": 131560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.869261903978414e-06,
+      "loss": 0.9609,
+      "step": 131570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.865971239593275e-06,
+      "loss": 0.5712,
+      "step": 131580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8626805752081345e-06,
+      "loss": 0.6361,
+      "step": 131590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.859389910822996e-06,
+      "loss": 0.5782,
+      "step": 131600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.856099246437857e-06,
+      "loss": 0.5178,
+      "step": 131610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.852808582052717e-06,
+      "loss": 0.6227,
+      "step": 131620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.849517917667578e-06,
+      "loss": 0.5811,
+      "step": 131630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8462272532824375e-06,
+      "loss": 0.4726,
+      "step": 131640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.842936588897299e-06,
+      "loss": 0.6668,
+      "step": 131650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.83964592451216e-06,
+      "loss": 0.5909,
+      "step": 131660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.83635526012702e-06,
+      "loss": 0.5858,
+      "step": 131670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.833064595741881e-06,
+      "loss": 0.6134,
+      "step": 131680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8297739313567404e-06,
+      "loss": 0.6477,
+      "step": 131690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.826483266971602e-06,
+      "loss": 0.5403,
+      "step": 131700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.823192602586463e-06,
+      "loss": 0.5835,
+      "step": 131710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8199019382013225e-06,
+      "loss": 0.6587,
+      "step": 131720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.816611273816184e-06,
+      "loss": 0.4815,
+      "step": 131730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.813320609431044e-06,
+      "loss": 0.5951,
+      "step": 131740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8100299450459055e-06,
+      "loss": 0.5236,
+      "step": 131750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.806739280660766e-06,
+      "loss": 0.5499,
+      "step": 131760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.803448616275626e-06,
+      "loss": 0.5789,
+      "step": 131770
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.800157951890488e-06,
+      "loss": 0.681,
+      "step": 131780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.796867287505347e-06,
+      "loss": 0.6186,
+      "step": 131790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.7935766231202084e-06,
+      "loss": 0.7314,
+      "step": 131800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.79028595873507e-06,
+      "loss": 0.637,
+      "step": 131810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.786995294349929e-06,
+      "loss": 0.4495,
+      "step": 131820
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.7837046299647906e-06,
+      "loss": 0.5075,
+      "step": 131830
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.78041396557965e-06,
+      "loss": 0.5642,
+      "step": 131840
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.777123301194511e-06,
+      "loss": 0.6727,
+      "step": 131850
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.773832636809373e-06,
+      "loss": 0.5838,
+      "step": 131860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.770541972424232e-06,
+      "loss": 0.5934,
+      "step": 131870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7672513080390935e-06,
+      "loss": 0.597,
+      "step": 131880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.763960643653953e-06,
+      "loss": 0.6439,
+      "step": 131890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.760669979268814e-06,
+      "loss": 0.534,
+      "step": 131900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.757379314883676e-06,
+      "loss": 0.6803,
+      "step": 131910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.754088650498536e-06,
+      "loss": 0.5924,
+      "step": 131920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7507979861133964e-06,
+      "loss": 0.4999,
+      "step": 131930
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.747507321728257e-06,
+      "loss": 0.635,
+      "step": 131940
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.744216657343118e-06,
+      "loss": 0.5815,
+      "step": 131950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.740925992957979e-06,
+      "loss": 0.6295,
+      "step": 131960
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.737635328572839e-06,
+      "loss": 0.5653,
+      "step": 131970
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7343446641877e-06,
+      "loss": 0.5415,
+      "step": 131980
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.73105399980256e-06,
+      "loss": 0.6552,
+      "step": 131990
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.727763335417421e-06,
+      "loss": 0.6339,
+      "step": 132000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.724472671032282e-06,
+      "loss": 0.5577,
+      "step": 132010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.721182006647142e-06,
+      "loss": 0.5623,
+      "step": 132020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.717891342262003e-06,
+      "loss": 0.4866,
+      "step": 132030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.714600677876863e-06,
+      "loss": 0.5629,
+      "step": 132040
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.711310013491724e-06,
+      "loss": 0.6375,
+      "step": 132050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.708019349106585e-06,
+      "loss": 0.5301,
+      "step": 132060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.704728684721445e-06,
+      "loss": 0.5013,
+      "step": 132070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.701438020336306e-06,
+      "loss": 0.6782,
+      "step": 132080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6981473559511666e-06,
+      "loss": 0.618,
+      "step": 132090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.694856691566027e-06,
+      "loss": 0.4643,
+      "step": 132100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.691566027180888e-06,
+      "loss": 0.5821,
+      "step": 132110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.688275362795749e-06,
+      "loss": 0.5666,
+      "step": 132120
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.68498469841061e-06,
+      "loss": 0.5602,
+      "step": 132130
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6816940340254695e-06,
+      "loss": 0.5312,
+      "step": 132140
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.678403369640331e-06,
+      "loss": 0.6437,
+      "step": 132150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.675112705255192e-06,
+      "loss": 0.5266,
+      "step": 132160
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.671822040870052e-06,
+      "loss": 0.517,
+      "step": 132170
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.668531376484913e-06,
+      "loss": 0.7153,
+      "step": 132180
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6652407120997725e-06,
+      "loss": 0.6425,
+      "step": 132190
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.661950047714634e-06,
+      "loss": 0.6917,
+      "step": 132200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.658659383329495e-06,
+      "loss": 0.5963,
+      "step": 132210
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6553687189443546e-06,
+      "loss": 0.6407,
+      "step": 132220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.652078054559216e-06,
+      "loss": 0.7281,
+      "step": 132230
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.648787390174077e-06,
+      "loss": 0.7083,
+      "step": 132240
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.645496725788937e-06,
+      "loss": 0.6042,
+      "step": 132250
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.642206061403798e-06,
+      "loss": 0.5813,
+      "step": 132260
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.638915397018658e-06,
+      "loss": 0.56,
+      "step": 132270
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.635624732633519e-06,
+      "loss": 0.6501,
+      "step": 132280
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.63233406824838e-06,
+      "loss": 0.5291,
+      "step": 132290
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6290434038632405e-06,
+      "loss": 0.5946,
+      "step": 132300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.625752739478102e-06,
+      "loss": 0.6218,
+      "step": 132310
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.622462075092961e-06,
+      "loss": 0.6076,
+      "step": 132320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.619171410707823e-06,
+      "loss": 0.5647,
+      "step": 132330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.615880746322684e-06,
+      "loss": 0.5863,
+      "step": 132340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6125900819375434e-06,
+      "loss": 0.5472,
+      "step": 132350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.609299417552405e-06,
+      "loss": 0.5802,
+      "step": 132360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.606008753167264e-06,
+      "loss": 0.5508,
+      "step": 132370
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.6027180887821255e-06,
+      "loss": 0.4621,
+      "step": 132380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.599427424396987e-06,
+      "loss": 0.7392,
+      "step": 132390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.596136760011846e-06,
+      "loss": 0.6612,
+      "step": 132400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.592846095626708e-06,
+      "loss": 0.7042,
+      "step": 132410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.589555431241567e-06,
+      "loss": 0.5934,
+      "step": 132420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5862647668564285e-06,
+      "loss": 0.6302,
+      "step": 132430
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.58297410247129e-06,
+      "loss": 0.665,
+      "step": 132440
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.579683438086149e-06,
+      "loss": 0.6873,
+      "step": 132450
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.576392773701011e-06,
+      "loss": 0.6797,
+      "step": 132460
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.573102109315871e-06,
+      "loss": 0.6802,
+      "step": 132470
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.569811444930732e-06,
+      "loss": 0.6191,
+      "step": 132480
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.566520780545593e-06,
+      "loss": 0.5589,
+      "step": 132490
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.563230116160453e-06,
+      "loss": 0.5164,
+      "step": 132500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.559939451775314e-06,
+      "loss": 0.6492,
+      "step": 132510
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.556648787390174e-06,
+      "loss": 0.6199,
+      "step": 132520
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.553358123005035e-06,
+      "loss": 0.6524,
+      "step": 132530
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5500674586198965e-06,
+      "loss": 0.5492,
+      "step": 132540
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.546776794234756e-06,
+      "loss": 0.6089,
+      "step": 132550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.543486129849617e-06,
+      "loss": 0.5183,
+      "step": 132560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.540195465464477e-06,
+      "loss": 0.5896,
+      "step": 132570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.536904801079338e-06,
+      "loss": 0.5625,
+      "step": 132580
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5336141366941994e-06,
+      "loss": 0.6393,
+      "step": 132590
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.530323472309059e-06,
+      "loss": 0.609,
+      "step": 132600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.52703280792392e-06,
+      "loss": 0.5148,
+      "step": 132610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.52374214353878e-06,
+      "loss": 0.542,
+      "step": 132620
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.520451479153641e-06,
+      "loss": 0.7004,
+      "step": 132630
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.517160814768502e-06,
+      "loss": 0.4611,
+      "step": 132640
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.513870150383363e-06,
+      "loss": 0.5954,
+      "step": 132650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.510579485998223e-06,
+      "loss": 0.5193,
+      "step": 132660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.507288821613084e-06,
+      "loss": 0.5292,
+      "step": 132670
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.503998157227945e-06,
+      "loss": 0.5863,
+      "step": 132680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.500707492842806e-06,
+      "loss": 0.511,
+      "step": 132690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.497416828457666e-06,
+      "loss": 0.6544,
+      "step": 132700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.494126164072527e-06,
+      "loss": 0.5849,
+      "step": 132710
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.490835499687387e-06,
+      "loss": 0.6265,
+      "step": 132720
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.487544835302248e-06,
+      "loss": 0.5446,
+      "step": 132730
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.484254170917109e-06,
+      "loss": 0.5809,
+      "step": 132740
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.480963506531969e-06,
+      "loss": 0.5783,
+      "step": 132750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.47767284214683e-06,
+      "loss": 0.6554,
+      "step": 132760
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4743821777616896e-06,
+      "loss": 0.5174,
+      "step": 132770
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.471091513376551e-06,
+      "loss": 0.585,
+      "step": 132780
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.467800848991412e-06,
+      "loss": 0.6253,
+      "step": 132790
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.464510184606272e-06,
+      "loss": 0.6823,
+      "step": 132800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.461219520221133e-06,
+      "loss": 0.6291,
+      "step": 132810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.457928855835993e-06,
+      "loss": 0.5886,
+      "step": 132820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.454638191450854e-06,
+      "loss": 0.5885,
+      "step": 132830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.451347527065715e-06,
+      "loss": 0.7277,
+      "step": 132840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4480568626805755e-06,
+      "loss": 0.5698,
+      "step": 132850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.444766198295437e-06,
+      "loss": 0.5738,
+      "step": 132860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.441475533910296e-06,
+      "loss": 0.5806,
+      "step": 132870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4381848695251576e-06,
+      "loss": 0.654,
+      "step": 132880
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.434894205140019e-06,
+      "loss": 0.678,
+      "step": 132890
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.431603540754878e-06,
+      "loss": 0.5686,
+      "step": 132900
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.42831287636974e-06,
+      "loss": 0.5957,
+      "step": 132910
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.425022211984601e-06,
+      "loss": 0.7306,
+      "step": 132920
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4217315475994605e-06,
+      "loss": 0.5407,
+      "step": 132930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.418440883214322e-06,
+      "loss": 0.6396,
+      "step": 132940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.415150218829181e-06,
+      "loss": 0.5666,
+      "step": 132950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.411859554444043e-06,
+      "loss": 0.6088,
+      "step": 132960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.408568890058904e-06,
+      "loss": 0.5561,
+      "step": 132970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4052782256737635e-06,
+      "loss": 0.5553,
+      "step": 132980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.401987561288625e-06,
+      "loss": 0.6566,
+      "step": 132990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.398696896903484e-06,
+      "loss": 0.6855,
+      "step": 133000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3954062325183456e-06,
+      "loss": 0.5573,
+      "step": 133010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.392115568133207e-06,
+      "loss": 0.507,
+      "step": 133020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.388824903748067e-06,
+      "loss": 0.5015,
+      "step": 133030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.385534239362928e-06,
+      "loss": 0.7487,
+      "step": 133040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.382243574977788e-06,
+      "loss": 0.6221,
+      "step": 133050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.378952910592649e-06,
+      "loss": 0.5629,
+      "step": 133060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.375662246207511e-06,
+      "loss": 0.5893,
+      "step": 133070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.37237158182237e-06,
+      "loss": 0.5802,
+      "step": 133080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3690809174372315e-06,
+      "loss": 0.5855,
+      "step": 133090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.365790253052091e-06,
+      "loss": 0.5424,
+      "step": 133100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.362499588666952e-06,
+      "loss": 0.6112,
+      "step": 133110
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.359208924281814e-06,
+      "loss": 0.6428,
+      "step": 133120
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.355918259896673e-06,
+      "loss": 0.6086,
+      "step": 133130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3526275955115344e-06,
+      "loss": 0.5302,
+      "step": 133140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.349336931126394e-06,
+      "loss": 0.6642,
+      "step": 133150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.346046266741255e-06,
+      "loss": 0.4719,
+      "step": 133160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3427556023561165e-06,
+      "loss": 0.6313,
+      "step": 133170
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.339464937970976e-06,
+      "loss": 0.5729,
+      "step": 133180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.336174273585837e-06,
+      "loss": 0.6947,
+      "step": 133190
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.332883609200698e-06,
+      "loss": 0.5174,
+      "step": 133200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.329592944815558e-06,
+      "loss": 0.5737,
+      "step": 133210
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3263022804304195e-06,
+      "loss": 0.5595,
+      "step": 133220
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.32301161604528e-06,
+      "loss": 0.5769,
+      "step": 133230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.319720951660141e-06,
+      "loss": 0.6309,
+      "step": 133240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.316430287275001e-06,
+      "loss": 0.718,
+      "step": 133250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.313139622889862e-06,
+      "loss": 0.5032,
+      "step": 133260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.309848958504723e-06,
+      "loss": 0.7181,
+      "step": 133270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.306558294119583e-06,
+      "loss": 0.6889,
+      "step": 133280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.303267629734444e-06,
+      "loss": 0.5027,
+      "step": 133290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.299976965349304e-06,
+      "loss": 0.4777,
+      "step": 133300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.296686300964165e-06,
+      "loss": 0.5576,
+      "step": 133310
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.293395636579026e-06,
+      "loss": 0.5766,
+      "step": 133320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.290104972193886e-06,
+      "loss": 0.5466,
+      "step": 133330
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.286814307808747e-06,
+      "loss": 0.6406,
+      "step": 133340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.283523643423607e-06,
+      "loss": 0.6336,
+      "step": 133350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.280232979038468e-06,
+      "loss": 0.6211,
+      "step": 133360
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.276942314653329e-06,
+      "loss": 0.5855,
+      "step": 133370
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.273651650268189e-06,
+      "loss": 0.6214,
+      "step": 133380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.27036098588305e-06,
+      "loss": 0.5887,
+      "step": 133390
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2670703214979104e-06,
+      "loss": 0.5813,
+      "step": 133400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.263779657112772e-06,
+      "loss": 0.5353,
+      "step": 133410
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.260488992727632e-06,
+      "loss": 0.6255,
+      "step": 133420
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2571983283424925e-06,
+      "loss": 0.641,
+      "step": 133430
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.253907663957354e-06,
+      "loss": 0.7404,
+      "step": 133440
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.250616999572213e-06,
+      "loss": 0.5642,
+      "step": 133450
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.247326335187075e-06,
+      "loss": 0.5928,
+      "step": 133460
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.244035670801935e-06,
+      "loss": 0.5876,
+      "step": 133470
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.240745006416796e-06,
+      "loss": 0.6033,
+      "step": 133480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.237454342031657e-06,
+      "loss": 0.6096,
+      "step": 133490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.234163677646517e-06,
+      "loss": 0.602,
+      "step": 133500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.230873013261378e-06,
+      "loss": 0.6623,
+      "step": 133510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.227582348876238e-06,
+      "loss": 0.5373,
+      "step": 133520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.224291684491099e-06,
+      "loss": 0.5076,
+      "step": 133530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.22100102010596e-06,
+      "loss": 0.5423,
+      "step": 133540
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.21771035572082e-06,
+      "loss": 0.5586,
+      "step": 133550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2144196913356806e-06,
+      "loss": 0.5502,
+      "step": 133560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.211129026950541e-06,
+      "loss": 0.7349,
+      "step": 133570
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.207838362565402e-06,
+      "loss": 0.6255,
+      "step": 133580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.204547698180263e-06,
+      "loss": 0.5669,
+      "step": 133590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.201257033795124e-06,
+      "loss": 0.4846,
+      "step": 133600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.197966369409984e-06,
+      "loss": 0.4789,
+      "step": 133610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.194675705024845e-06,
+      "loss": 0.5856,
+      "step": 133620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.191385040639706e-06,
+      "loss": 0.5158,
+      "step": 133630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1880943762545665e-06,
+      "loss": 0.5624,
+      "step": 133640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.184803711869427e-06,
+      "loss": 0.4933,
+      "step": 133650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.181513047484287e-06,
+      "loss": 0.6342,
+      "step": 133660
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.178222383099148e-06,
+      "loss": 0.4862,
+      "step": 133670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.174931718714009e-06,
+      "loss": 0.6838,
+      "step": 133680
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.171641054328869e-06,
+      "loss": 0.5536,
+      "step": 133690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.16835038994373e-06,
+      "loss": 0.5615,
+      "step": 133700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.16505972555859e-06,
+      "loss": 0.5486,
+      "step": 133710
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.161769061173451e-06,
+      "loss": 0.5168,
+      "step": 133720
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.158478396788312e-06,
+      "loss": 0.6131,
+      "step": 133730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.155187732403172e-06,
+      "loss": 0.6066,
+      "step": 133740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.151897068018033e-06,
+      "loss": 0.6171,
+      "step": 133750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.148606403632893e-06,
+      "loss": 0.6436,
+      "step": 133760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1453157392477545e-06,
+      "loss": 0.6012,
+      "step": 133770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.142025074862615e-06,
+      "loss": 0.5986,
+      "step": 133780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.138734410477476e-06,
+      "loss": 0.5789,
+      "step": 133790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1354437460923366e-06,
+      "loss": 0.5381,
+      "step": 133800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.132153081707197e-06,
+      "loss": 0.6306,
+      "step": 133810
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.128862417322058e-06,
+      "loss": 0.654,
+      "step": 133820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.125571752936919e-06,
+      "loss": 0.5466,
+      "step": 133830
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.122281088551779e-06,
+      "loss": 0.5566,
+      "step": 133840
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1189904241666395e-06,
+      "loss": 0.5312,
+      "step": 133850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1156997597815e-06,
+      "loss": 0.6802,
+      "step": 133860
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.112409095396361e-06,
+      "loss": 0.6513,
+      "step": 133870
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.109118431011222e-06,
+      "loss": 0.5843,
+      "step": 133880
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.105827766626082e-06,
+      "loss": 0.6469,
+      "step": 133890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.1025371022409425e-06,
+      "loss": 0.5503,
+      "step": 133900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.099246437855803e-06,
+      "loss": 0.6325,
+      "step": 133910
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.095955773470664e-06,
+      "loss": 0.6236,
+      "step": 133920
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.092665109085525e-06,
+      "loss": 0.4996,
+      "step": 133930
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.089374444700385e-06,
+      "loss": 0.5109,
+      "step": 133940
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.086083780315245e-06,
+      "loss": 0.5424,
+      "step": 133950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.082793115930107e-06,
+      "loss": 0.6674,
+      "step": 133960
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.079502451544967e-06,
+      "loss": 0.576,
+      "step": 133970
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.076211787159828e-06,
+      "loss": 0.5519,
+      "step": 133980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.072921122774689e-06,
+      "loss": 0.5585,
+      "step": 133990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.069630458389549e-06,
+      "loss": 0.5948,
+      "step": 134000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.06633979400441e-06,
+      "loss": 0.5869,
+      "step": 134010
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.063049129619271e-06,
+      "loss": 0.637,
+      "step": 134020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.059758465234131e-06,
+      "loss": 0.5115,
+      "step": 134030
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.056467800848992e-06,
+      "loss": 0.6273,
+      "step": 134040
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.053177136463852e-06,
+      "loss": 0.7742,
+      "step": 134050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.049886472078713e-06,
+      "loss": 0.561,
+      "step": 134060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.046595807693574e-06,
+      "loss": 0.5359,
+      "step": 134070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.043305143308434e-06,
+      "loss": 0.5912,
+      "step": 134080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.040014478923295e-06,
+      "loss": 0.5958,
+      "step": 134090
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.036723814538155e-06,
+      "loss": 0.7198,
+      "step": 134100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0334331501530155e-06,
+      "loss": 0.5913,
+      "step": 134110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.030142485767877e-06,
+      "loss": 0.6868,
+      "step": 134120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.026851821382737e-06,
+      "loss": 0.7223,
+      "step": 134130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.023561156997598e-06,
+      "loss": 0.6133,
+      "step": 134140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.020270492612459e-06,
+      "loss": 0.5743,
+      "step": 134150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.016979828227319e-06,
+      "loss": 0.6363,
+      "step": 134160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.013689163842181e-06,
+      "loss": 0.6821,
+      "step": 134170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.010398499457041e-06,
+      "loss": 0.4829,
+      "step": 134180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0071078350719014e-06,
+      "loss": 0.6228,
+      "step": 134190
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.003817170686762e-06,
+      "loss": 0.4491,
+      "step": 134200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.000526506301623e-06,
+      "loss": 0.6225,
+      "step": 134210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9972358419164835e-06,
+      "loss": 0.6068,
+      "step": 134220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.993945177531344e-06,
+      "loss": 0.6193,
+      "step": 134230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.990654513146204e-06,
+      "loss": 0.6092,
+      "step": 134240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.987363848761065e-06,
+      "loss": 0.5089,
+      "step": 134250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.984073184375926e-06,
+      "loss": 0.5852,
+      "step": 134260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9807825199907865e-06,
+      "loss": 0.599,
+      "step": 134270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.977491855605647e-06,
+      "loss": 0.5992,
+      "step": 134280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.974201191220507e-06,
+      "loss": 0.5858,
+      "step": 134290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.970910526835368e-06,
+      "loss": 0.6293,
+      "step": 134300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.967619862450229e-06,
+      "loss": 0.6147,
+      "step": 134310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9643291980650894e-06,
+      "loss": 0.6825,
+      "step": 134320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.96103853367995e-06,
+      "loss": 0.5277,
+      "step": 134330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.957747869294811e-06,
+      "loss": 0.6367,
+      "step": 134340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9544572049096716e-06,
+      "loss": 0.6565,
+      "step": 134350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.951166540524533e-06,
+      "loss": 0.5857,
+      "step": 134360
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.947875876139393e-06,
+      "loss": 0.4934,
+      "step": 134370
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.944585211754254e-06,
+      "loss": 0.5134,
+      "step": 134380
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.941294547369114e-06,
+      "loss": 0.7604,
+      "step": 134390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9380038829839745e-06,
+      "loss": 0.7929,
+      "step": 134400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.934713218598836e-06,
+      "loss": 0.5569,
+      "step": 134410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.931422554213696e-06,
+      "loss": 0.6001,
+      "step": 134420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.928131889828557e-06,
+      "loss": 0.7808,
+      "step": 134430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.924841225443417e-06,
+      "loss": 0.4701,
+      "step": 134440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9215505610582775e-06,
+      "loss": 0.7296,
+      "step": 134450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.918259896673139e-06,
+      "loss": 0.6488,
+      "step": 134460
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.914969232287999e-06,
+      "loss": 0.5785,
+      "step": 134470
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9116785679028596e-06,
+      "loss": 0.4705,
+      "step": 134480
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.90838790351772e-06,
+      "loss": 0.597,
+      "step": 134490
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.90509723913258e-06,
+      "loss": 0.623,
+      "step": 134500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.901806574747442e-06,
+      "loss": 0.5447,
+      "step": 134510
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.898515910362302e-06,
+      "loss": 0.5983,
+      "step": 134520
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.895225245977163e-06,
+      "loss": 0.6666,
+      "step": 134530
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.891934581592024e-06,
+      "loss": 0.6291,
+      "step": 134540
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.888643917206885e-06,
+      "loss": 0.6365,
+      "step": 134550
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8853532528217455e-06,
+      "loss": 0.6069,
+      "step": 134560
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.882062588436606e-06,
+      "loss": 0.5181,
+      "step": 134570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.878771924051466e-06,
+      "loss": 0.569,
+      "step": 134580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.875481259666327e-06,
+      "loss": 0.5669,
+      "step": 134590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.872190595281188e-06,
+      "loss": 0.5388,
+      "step": 134600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.868899930896048e-06,
+      "loss": 0.5025,
+      "step": 134610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.865609266510909e-06,
+      "loss": 0.5611,
+      "step": 134620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.862318602125769e-06,
+      "loss": 0.632,
+      "step": 134630
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.85902793774063e-06,
+      "loss": 0.5238,
+      "step": 134640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.855737273355491e-06,
+      "loss": 0.5207,
+      "step": 134650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.852446608970351e-06,
+      "loss": 0.5658,
+      "step": 134660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.849155944585212e-06,
+      "loss": 0.5832,
+      "step": 134670
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.845865280200072e-06,
+      "loss": 0.6252,
+      "step": 134680
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.842574615814933e-06,
+      "loss": 0.548,
+      "step": 134690
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.839283951429794e-06,
+      "loss": 0.5484,
+      "step": 134700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.835993287044654e-06,
+      "loss": 0.5254,
+      "step": 134710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.832702622659516e-06,
+      "loss": 0.5078,
+      "step": 134720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.829411958274376e-06,
+      "loss": 0.6057,
+      "step": 134730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.826121293889236e-06,
+      "loss": 0.5984,
+      "step": 134740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.822830629504098e-06,
+      "loss": 0.6695,
+      "step": 134750
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.819539965118958e-06,
+      "loss": 0.5805,
+      "step": 134760
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8162493007338185e-06,
+      "loss": 0.568,
+      "step": 134770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.812958636348679e-06,
+      "loss": 0.5366,
+      "step": 134780
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.809667971963539e-06,
+      "loss": 0.5221,
+      "step": 134790
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.806377307578401e-06,
+      "loss": 0.5834,
+      "step": 134800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.803086643193261e-06,
+      "loss": 0.6215,
+      "step": 134810
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.7997959788081215e-06,
+      "loss": 0.6905,
+      "step": 134820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.796505314422982e-06,
+      "loss": 0.7029,
+      "step": 134830
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.793214650037842e-06,
+      "loss": 0.5033,
+      "step": 134840
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.789923985652704e-06,
+      "loss": 0.7659,
+      "step": 134850
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.786633321267564e-06,
+      "loss": 0.7039,
+      "step": 134860
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.7833426568824244e-06,
+      "loss": 0.5137,
+      "step": 134870
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.780051992497285e-06,
+      "loss": 0.4625,
+      "step": 134880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.776761328112146e-06,
+      "loss": 0.5708,
+      "step": 134890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.7734706637270065e-06,
+      "loss": 0.586,
+      "step": 134900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.770179999341868e-06,
+      "loss": 0.6002,
+      "step": 134910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.766889334956728e-06,
+      "loss": 0.5016,
+      "step": 134920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.763598670571589e-06,
+      "loss": 0.5555,
+      "step": 134930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.76030800618645e-06,
+      "loss": 0.7431,
+      "step": 134940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.75701734180131e-06,
+      "loss": 0.6003,
+      "step": 134950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.753726677416171e-06,
+      "loss": 0.5796,
+      "step": 134960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.750436013031031e-06,
+      "loss": 0.4908,
+      "step": 134970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.747145348645892e-06,
+      "loss": 0.5615,
+      "step": 134980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.743854684260753e-06,
+      "loss": 0.4819,
+      "step": 134990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.740564019875613e-06,
+      "loss": 0.7673,
+      "step": 135000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.737273355490474e-06,
+      "loss": 0.5852,
+      "step": 135010
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.733982691105334e-06,
+      "loss": 0.4607,
+      "step": 135020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7306920267201945e-06,
+      "loss": 0.5315,
+      "step": 135030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.727401362335056e-06,
+      "loss": 0.5838,
+      "step": 135040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.724110697949916e-06,
+      "loss": 0.6523,
+      "step": 135050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.720820033564777e-06,
+      "loss": 0.4818,
+      "step": 135060
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.717529369179637e-06,
+      "loss": 0.6446,
+      "step": 135070
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.714238704794498e-06,
+      "loss": 0.5146,
+      "step": 135080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.71094804040936e-06,
+      "loss": 0.5488,
+      "step": 135090
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.70765737602422e-06,
+      "loss": 0.637,
+      "step": 135100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7043667116390804e-06,
+      "loss": 0.6013,
+      "step": 135110
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.701076047253941e-06,
+      "loss": 0.6633,
+      "step": 135120
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.697785382868801e-06,
+      "loss": 0.5349,
+      "step": 135130
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6944947184836626e-06,
+      "loss": 0.6097,
+      "step": 135140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.691204054098523e-06,
+      "loss": 0.5402,
+      "step": 135150
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.687913389713383e-06,
+      "loss": 0.6163,
+      "step": 135160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.684622725328244e-06,
+      "loss": 0.7108,
+      "step": 135170
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.681332060943104e-06,
+      "loss": 0.5543,
+      "step": 135180
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6780413965579655e-06,
+      "loss": 0.5759,
+      "step": 135190
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.674750732172826e-06,
+      "loss": 0.5956,
+      "step": 135200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.671460067787686e-06,
+      "loss": 0.7011,
+      "step": 135210
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.668169403402547e-06,
+      "loss": 0.7088,
+      "step": 135220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.664878739017408e-06,
+      "loss": 0.7061,
+      "step": 135230
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6615880746322685e-06,
+      "loss": 0.7508,
+      "step": 135240
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.658297410247129e-06,
+      "loss": 0.5467,
+      "step": 135250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.65500674586199e-06,
+      "loss": 0.5509,
+      "step": 135260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6517160814768506e-06,
+      "loss": 0.5736,
+      "step": 135270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.648425417091712e-06,
+      "loss": 0.5457,
+      "step": 135280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.645134752706572e-06,
+      "loss": 0.4638,
+      "step": 135290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.641844088321433e-06,
+      "loss": 0.5773,
+      "step": 135300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.638553423936293e-06,
+      "loss": 0.7941,
+      "step": 135310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6352627595511535e-06,
+      "loss": 0.5476,
+      "step": 135320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.631972095166015e-06,
+      "loss": 0.6224,
+      "step": 135330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.628681430780875e-06,
+      "loss": 0.7089,
+      "step": 135340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.625390766395736e-06,
+      "loss": 0.5716,
+      "step": 135350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.622100102010596e-06,
+      "loss": 0.5877,
+      "step": 135360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6188094376254565e-06,
+      "loss": 0.5153,
+      "step": 135370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.615518773240318e-06,
+      "loss": 0.6753,
+      "step": 135380
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.612228108855178e-06,
+      "loss": 0.6273,
+      "step": 135390
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6089374444700386e-06,
+      "loss": 0.6045,
+      "step": 135400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.605646780084899e-06,
+      "loss": 0.5476,
+      "step": 135410
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.602356115699759e-06,
+      "loss": 0.5592,
+      "step": 135420
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.599065451314621e-06,
+      "loss": 0.6326,
+      "step": 135430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.595774786929481e-06,
+      "loss": 0.5563,
+      "step": 135440
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.592484122544342e-06,
+      "loss": 0.6034,
+      "step": 135450
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.589193458159203e-06,
+      "loss": 0.5081,
+      "step": 135460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.585902793774063e-06,
+      "loss": 0.4932,
+      "step": 135470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5826121293889245e-06,
+      "loss": 0.7084,
+      "step": 135480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.579321465003785e-06,
+      "loss": 0.7573,
+      "step": 135490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.576030800618645e-06,
+      "loss": 0.5319,
+      "step": 135500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.572740136233506e-06,
+      "loss": 0.7737,
+      "step": 135510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.569449471848366e-06,
+      "loss": 0.6291,
+      "step": 135520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5661588074632274e-06,
+      "loss": 0.5636,
+      "step": 135530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.562868143078088e-06,
+      "loss": 0.5849,
+      "step": 135540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.559577478692948e-06,
+      "loss": 0.6626,
+      "step": 135550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.556286814307809e-06,
+      "loss": 0.6005,
+      "step": 135560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.552996149922669e-06,
+      "loss": 0.5526,
+      "step": 135570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.54970548553753e-06,
+      "loss": 0.774,
+      "step": 135580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.546414821152391e-06,
+      "loss": 0.5546,
+      "step": 135590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.543124156767251e-06,
+      "loss": 0.6571,
+      "step": 135600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.539833492382112e-06,
+      "loss": 0.5825,
+      "step": 135610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.536542827996973e-06,
+      "loss": 0.5234,
+      "step": 135620
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.533252163611833e-06,
+      "loss": 0.647,
+      "step": 135630
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.529961499226695e-06,
+      "loss": 0.6409,
+      "step": 135640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.526670834841555e-06,
+      "loss": 0.5104,
+      "step": 135650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5233801704564154e-06,
+      "loss": 0.5811,
+      "step": 135660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.520089506071277e-06,
+      "loss": 0.5326,
+      "step": 135670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.516798841686137e-06,
+      "loss": 0.5805,
+      "step": 135680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5135081773009975e-06,
+      "loss": 0.5269,
+      "step": 135690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.510217512915858e-06,
+      "loss": 0.5595,
+      "step": 135700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.506926848530718e-06,
+      "loss": 0.6927,
+      "step": 135710
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.50363618414558e-06,
+      "loss": 0.6,
+      "step": 135720
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.50034551976044e-06,
+      "loss": 0.5775,
+      "step": 135730
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4970548553753005e-06,
+      "loss": 0.577,
+      "step": 135740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.493764190990161e-06,
+      "loss": 0.7086,
+      "step": 135750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.490473526605021e-06,
+      "loss": 0.6259,
+      "step": 135760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.487182862219883e-06,
+      "loss": 0.5373,
+      "step": 135770
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.483892197834743e-06,
+      "loss": 0.5708,
+      "step": 135780
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4806015334496034e-06,
+      "loss": 0.6158,
+      "step": 135790
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.477310869064464e-06,
+      "loss": 0.5679,
+      "step": 135800
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.474020204679325e-06,
+      "loss": 1.0051,
+      "step": 135810
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4707295402941855e-06,
+      "loss": 0.6468,
+      "step": 135820
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.467438875909047e-06,
+      "loss": 0.5624,
+      "step": 135830
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.464148211523907e-06,
+      "loss": 0.5119,
+      "step": 135840
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.460857547138768e-06,
+      "loss": 0.5191,
+      "step": 135850
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.457566882753628e-06,
+      "loss": 0.5818,
+      "step": 135860
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.454276218368489e-06,
+      "loss": 0.5777,
+      "step": 135870
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.45098555398335e-06,
+      "loss": 0.4516,
+      "step": 135880
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.44769488959821e-06,
+      "loss": 0.6739,
+      "step": 135890
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.444404225213071e-06,
+      "loss": 0.7043,
+      "step": 135900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.441113560827931e-06,
+      "loss": 0.7549,
+      "step": 135910
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.437822896442792e-06,
+      "loss": 0.6,
+      "step": 135920
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.434532232057653e-06,
+      "loss": 0.5566,
+      "step": 135930
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.431241567672513e-06,
+      "loss": 0.5107,
+      "step": 135940
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.4279509032873736e-06,
+      "loss": 0.547,
+      "step": 135950
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.424660238902235e-06,
+      "loss": 0.549,
+      "step": 135960
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.421369574517095e-06,
+      "loss": 0.624,
+      "step": 135970
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.418078910131956e-06,
+      "loss": 0.7419,
+      "step": 135980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.414788245746816e-06,
+      "loss": 0.6491,
+      "step": 135990
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.411497581361677e-06,
+      "loss": 0.6529,
+      "step": 136000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.408206916976538e-06,
+      "loss": 0.5622,
+      "step": 136010
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.404916252591399e-06,
+      "loss": 0.5456,
+      "step": 136020
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.4016255882062595e-06,
+      "loss": 0.5682,
+      "step": 136030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.39833492382112e-06,
+      "loss": 0.5408,
+      "step": 136040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.39504425943598e-06,
+      "loss": 0.5555,
+      "step": 136050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3917535950508416e-06,
+      "loss": 0.698,
+      "step": 136060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.388462930665702e-06,
+      "loss": 0.7081,
+      "step": 136070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.385172266280562e-06,
+      "loss": 0.6897,
+      "step": 136080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.381881601895423e-06,
+      "loss": 0.6211,
+      "step": 136090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.378590937510283e-06,
+      "loss": 0.628,
+      "step": 136100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3753002731251445e-06,
+      "loss": 0.5725,
+      "step": 136110
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.372009608740005e-06,
+      "loss": 0.6844,
+      "step": 136120
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.368718944354865e-06,
+      "loss": 0.5435,
+      "step": 136130
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.365428279969726e-06,
+      "loss": 0.525,
+      "step": 136140
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.362137615584586e-06,
+      "loss": 0.725,
+      "step": 136150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3588469511994475e-06,
+      "loss": 0.5045,
+      "step": 136160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.355556286814308e-06,
+      "loss": 0.5749,
+      "step": 136170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.352265622429168e-06,
+      "loss": 0.5434,
+      "step": 136180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3489749580440296e-06,
+      "loss": 0.6138,
+      "step": 136190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.34568429365889e-06,
+      "loss": 0.5395,
+      "step": 136200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.342393629273751e-06,
+      "loss": 0.563,
+      "step": 136210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.339102964888612e-06,
+      "loss": 0.5185,
+      "step": 136220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.335812300503472e-06,
+      "loss": 0.5411,
+      "step": 136230
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3325216361183325e-06,
+      "loss": 0.6097,
+      "step": 136240
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.329230971733193e-06,
+      "loss": 0.6031,
+      "step": 136250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.325940307348054e-06,
+      "loss": 0.6784,
+      "step": 136260
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.322649642962915e-06,
+      "loss": 0.4946,
+      "step": 136270
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.319358978577775e-06,
+      "loss": 0.6223,
+      "step": 136280
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3160683141926355e-06,
+      "loss": 0.5957,
+      "step": 136290
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.312777649807497e-06,
+      "loss": 0.6129,
+      "step": 136300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.309486985422357e-06,
+      "loss": 0.4994,
+      "step": 136310
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.306196321037218e-06,
+      "loss": 0.5199,
+      "step": 136320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.302905656652078e-06,
+      "loss": 0.5462,
+      "step": 136330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.299614992266938e-06,
+      "loss": 0.6477,
+      "step": 136340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.2963243278818e-06,
+      "loss": 0.7135,
+      "step": 136350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.29303366349666e-06,
+      "loss": 0.6628,
+      "step": 136360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.2897429991115205e-06,
+      "loss": 0.6154,
+      "step": 136370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.286452334726382e-06,
+      "loss": 0.5928,
+      "step": 136380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.283161670341242e-06,
+      "loss": 0.6252,
+      "step": 136390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.2798710059561035e-06,
+      "loss": 0.5419,
+      "step": 136400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.276580341570964e-06,
+      "loss": 0.589,
+      "step": 136410
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.273289677185824e-06,
+      "loss": 0.6049,
+      "step": 136420
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.269999012800685e-06,
+      "loss": 0.4768,
+      "step": 136430
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.266708348415545e-06,
+      "loss": 0.5187,
+      "step": 136440
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2634176840304064e-06,
+      "loss": 0.5395,
+      "step": 136450
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.260127019645267e-06,
+      "loss": 0.5457,
+      "step": 136460
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.256836355260127e-06,
+      "loss": 0.5535,
+      "step": 136470
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.253545690874988e-06,
+      "loss": 0.6236,
+      "step": 136480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.250255026489848e-06,
+      "loss": 0.5924,
+      "step": 136490
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.246964362104709e-06,
+      "loss": 0.572,
+      "step": 136500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.24367369771957e-06,
+      "loss": 0.9381,
+      "step": 136510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.24038303333443e-06,
+      "loss": 0.7522,
+      "step": 136520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.237092368949291e-06,
+      "loss": 0.5476,
+      "step": 136530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.233801704564151e-06,
+      "loss": 0.6178,
+      "step": 136540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.230511040179012e-06,
+      "loss": 0.5064,
+      "step": 136550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.227220375793873e-06,
+      "loss": 0.5126,
+      "step": 136560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.223929711408734e-06,
+      "loss": 0.59,
+      "step": 136570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2206390470235944e-06,
+      "loss": 0.522,
+      "step": 136580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.217348382638455e-06,
+      "loss": 0.6492,
+      "step": 136590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.214057718253316e-06,
+      "loss": 0.4904,
+      "step": 136600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2107670538681765e-06,
+      "loss": 0.5366,
+      "step": 136610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.207476389483037e-06,
+      "loss": 0.5474,
+      "step": 136620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.204185725097897e-06,
+      "loss": 0.522,
+      "step": 136630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.200895060712758e-06,
+      "loss": 0.5953,
+      "step": 136640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.197604396327619e-06,
+      "loss": 0.5886,
+      "step": 136650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1943137319424795e-06,
+      "loss": 0.5334,
+      "step": 136660
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.19102306755734e-06,
+      "loss": 0.6803,
+      "step": 136670
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1877324031722e-06,
+      "loss": 0.6186,
+      "step": 136680
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.184441738787062e-06,
+      "loss": 0.5349,
+      "step": 136690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.181151074401922e-06,
+      "loss": 0.5356,
+      "step": 136700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1778604100167824e-06,
+      "loss": 0.5943,
+      "step": 136710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.174569745631643e-06,
+      "loss": 0.5009,
+      "step": 136720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.171279081246503e-06,
+      "loss": 0.6175,
+      "step": 136730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1679884168613646e-06,
+      "loss": 0.5281,
+      "step": 136740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.164697752476225e-06,
+      "loss": 0.6404,
+      "step": 136750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.161407088091086e-06,
+      "loss": 0.6024,
+      "step": 136760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.158116423705947e-06,
+      "loss": 0.5189,
+      "step": 136770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.154825759320807e-06,
+      "loss": 0.6118,
+      "step": 136780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.151535094935668e-06,
+      "loss": 0.5354,
+      "step": 136790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.148244430550529e-06,
+      "loss": 0.5999,
+      "step": 136800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.144953766165389e-06,
+      "loss": 0.5795,
+      "step": 136810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.14166310178025e-06,
+      "loss": 0.6457,
+      "step": 136820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.13837243739511e-06,
+      "loss": 0.6672,
+      "step": 136830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.135081773009971e-06,
+      "loss": 0.5844,
+      "step": 136840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.131791108624832e-06,
+      "loss": 0.5195,
+      "step": 136850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.128500444239692e-06,
+      "loss": 0.6828,
+      "step": 136860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1252097798545526e-06,
+      "loss": 0.6114,
+      "step": 136870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.121919115469413e-06,
+      "loss": 0.6418,
+      "step": 136880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.118628451084274e-06,
+      "loss": 0.6676,
+      "step": 136890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.115337786699135e-06,
+      "loss": 0.6465,
+      "step": 136900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.112047122313995e-06,
+      "loss": 0.5959,
+      "step": 136910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1087564579288555e-06,
+      "loss": 0.6897,
+      "step": 136920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.105465793543717e-06,
+      "loss": 0.5829,
+      "step": 136930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.102175129158577e-06,
+      "loss": 0.6402,
+      "step": 136940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0988844647734385e-06,
+      "loss": 0.5376,
+      "step": 136950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.095593800388299e-06,
+      "loss": 0.6063,
+      "step": 136960
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.092303136003159e-06,
+      "loss": 0.5197,
+      "step": 136970
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.08901247161802e-06,
+      "loss": 0.5425,
+      "step": 136980
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.085721807232881e-06,
+      "loss": 0.6031,
+      "step": 136990
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.082431142847741e-06,
+      "loss": 0.5702,
+      "step": 137000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.079140478462602e-06,
+      "loss": 0.5472,
+      "step": 137010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.075849814077462e-06,
+      "loss": 0.5924,
+      "step": 137020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0725591496923235e-06,
+      "loss": 0.5366,
+      "step": 137030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.069268485307184e-06,
+      "loss": 0.5894,
+      "step": 137040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.065977820922044e-06,
+      "loss": 0.7161,
+      "step": 137050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.062687156536905e-06,
+      "loss": 0.4998,
+      "step": 137060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.059396492151765e-06,
+      "loss": 0.5452,
+      "step": 137070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0561058277666265e-06,
+      "loss": 0.5055,
+      "step": 137080
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.052815163381487e-06,
+      "loss": 0.6337,
+      "step": 137090
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.049524498996347e-06,
+      "loss": 0.7005,
+      "step": 137100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.046233834611208e-06,
+      "loss": 0.6204,
+      "step": 137110
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.042943170226069e-06,
+      "loss": 0.5884,
+      "step": 137120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.039652505840929e-06,
+      "loss": 0.6419,
+      "step": 137130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.036361841455791e-06,
+      "loss": 0.6522,
+      "step": 137140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.033071177070651e-06,
+      "loss": 0.5637,
+      "step": 137150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0297805126855115e-06,
+      "loss": 0.6683,
+      "step": 137160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.026489848300372e-06,
+      "loss": 0.5567,
+      "step": 137170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.023199183915233e-06,
+      "loss": 0.6685,
+      "step": 137180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.019908519530094e-06,
+      "loss": 0.5331,
+      "step": 137190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.016617855144954e-06,
+      "loss": 0.6067,
+      "step": 137200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0133271907598145e-06,
+      "loss": 0.6236,
+      "step": 137210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.010036526374675e-06,
+      "loss": 0.5838,
+      "step": 137220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.006745861989536e-06,
+      "loss": 0.597,
+      "step": 137230
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.003455197604397e-06,
+      "loss": 0.7522,
+      "step": 137240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.000164533219257e-06,
+      "loss": 0.573,
+      "step": 137250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9968738688341174e-06,
+      "loss": 0.5675,
+      "step": 137260
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.993583204448978e-06,
+      "loss": 0.5452,
+      "step": 137270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.990292540063839e-06,
+      "loss": 0.5702,
+      "step": 137280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9870018756786995e-06,
+      "loss": 0.6076,
+      "step": 137290
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.98371121129356e-06,
+      "loss": 0.5446,
+      "step": 137300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.980420546908421e-06,
+      "loss": 0.6702,
+      "step": 137310
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.977129882523282e-06,
+      "loss": 0.6656,
+      "step": 137320
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.973839218138143e-06,
+      "loss": 0.5845,
+      "step": 137330
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.970548553753003e-06,
+      "loss": 0.6103,
+      "step": 137340
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.967257889367864e-06,
+      "loss": 0.576,
+      "step": 137350
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.963967224982724e-06,
+      "loss": 0.6726,
+      "step": 137360
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9606765605975854e-06,
+      "loss": 0.5769,
+      "step": 137370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.957385896212446e-06,
+      "loss": 0.529,
+      "step": 137380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.954095231827306e-06,
+      "loss": 0.6347,
+      "step": 137390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.950804567442167e-06,
+      "loss": 0.5592,
+      "step": 137400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.947513903057027e-06,
+      "loss": 0.6616,
+      "step": 137410
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.944223238671888e-06,
+      "loss": 0.5457,
+      "step": 137420
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.940932574286749e-06,
+      "loss": 0.7828,
+      "step": 137430
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.937641909901609e-06,
+      "loss": 0.5802,
+      "step": 137440
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.93435124551647e-06,
+      "loss": 0.5824,
+      "step": 137450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.93106058113133e-06,
+      "loss": 0.6509,
+      "step": 137460
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.927769916746191e-06,
+      "loss": 0.5525,
+      "step": 137470
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.924479252361052e-06,
+      "loss": 0.6007,
+      "step": 137480
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.921188587975912e-06,
+      "loss": 0.6667,
+      "step": 137490
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9178979235907734e-06,
+      "loss": 0.6744,
+      "step": 137500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.914607259205634e-06,
+      "loss": 0.5388,
+      "step": 137510
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.911316594820495e-06,
+      "loss": 0.5778,
+      "step": 137520
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9080259304353556e-06,
+      "loss": 0.5479,
+      "step": 137530
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.904735266050216e-06,
+      "loss": 0.638,
+      "step": 137540
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.901444601665076e-06,
+      "loss": 0.5803,
+      "step": 137550
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.898153937279937e-06,
+      "loss": 0.5114,
+      "step": 137560
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894863272894798e-06,
+      "loss": 0.6443,
+      "step": 137570
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8915726085096585e-06,
+      "loss": 0.6184,
+      "step": 137580
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.888281944124519e-06,
+      "loss": 0.6428,
+      "step": 137590
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.884991279739379e-06,
+      "loss": 0.644,
+      "step": 137600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.88170061535424e-06,
+      "loss": 0.5516,
+      "step": 137610
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.878409950969101e-06,
+      "loss": 0.5902,
+      "step": 137620
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8751192865839615e-06,
+      "loss": 0.6667,
+      "step": 137630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.871828622198822e-06,
+      "loss": 0.5376,
+      "step": 137640
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.868537957813682e-06,
+      "loss": 0.589,
+      "step": 137650
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.865247293428543e-06,
+      "loss": 0.6017,
+      "step": 137660
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.861956629043404e-06,
+      "loss": 0.4811,
+      "step": 137670
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.858665964658264e-06,
+      "loss": 0.5982,
+      "step": 137680
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.855375300273126e-06,
+      "loss": 0.603,
+      "step": 137690
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.852084635887986e-06,
+      "loss": 0.6215,
+      "step": 137700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8487939715028465e-06,
+      "loss": 0.714,
+      "step": 137710
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.845503307117708e-06,
+      "loss": 0.5466,
+      "step": 137720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.842212642732568e-06,
+      "loss": 0.5243,
+      "step": 137730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.838921978347429e-06,
+      "loss": 0.6275,
+      "step": 137740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.835631313962289e-06,
+      "loss": 0.4983,
+      "step": 137750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.83234064957715e-06,
+      "loss": 0.611,
+      "step": 137760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.829049985192011e-06,
+      "loss": 0.557,
+      "step": 137770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.825759320806871e-06,
+      "loss": 0.7075,
+      "step": 137780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8224686564217316e-06,
+      "loss": 0.5581,
+      "step": 137790
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.819177992036592e-06,
+      "loss": 0.6432,
+      "step": 137800
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.815887327651453e-06,
+      "loss": 0.4886,
+      "step": 137810
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.812596663266314e-06,
+      "loss": 0.6031,
+      "step": 137820
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.809305998881174e-06,
+      "loss": 0.5786,
+      "step": 137830
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8060153344960345e-06,
+      "loss": 0.4361,
+      "step": 137840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.802724670110895e-06,
+      "loss": 0.5184,
+      "step": 137850
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.799434005725756e-06,
+      "loss": 0.5526,
+      "step": 137860
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.796143341340617e-06,
+      "loss": 0.4659,
+      "step": 137870
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.792852676955478e-06,
+      "loss": 0.4823,
+      "step": 137880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.789562012570338e-06,
+      "loss": 0.5262,
+      "step": 137890
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.786271348185199e-06,
+      "loss": 0.5752,
+      "step": 137900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.78298068380006e-06,
+      "loss": 0.6476,
+      "step": 137910
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.77969001941492e-06,
+      "loss": 0.6999,
+      "step": 137920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.776399355029781e-06,
+      "loss": 0.6333,
+      "step": 137930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.773108690644641e-06,
+      "loss": 0.6563,
+      "step": 137940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.769818026259502e-06,
+      "loss": 0.6178,
+      "step": 137950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.766527361874363e-06,
+      "loss": 0.5853,
+      "step": 137960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.763236697489223e-06,
+      "loss": 0.5595,
+      "step": 137970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.759946033104084e-06,
+      "loss": 0.6195,
+      "step": 137980
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.756655368718944e-06,
+      "loss": 0.6658,
+      "step": 137990
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.753364704333805e-06,
+      "loss": 0.5041,
+      "step": 138000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.750074039948666e-06,
+      "loss": 0.6006,
+      "step": 138010
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.746783375563526e-06,
+      "loss": 0.5768,
+      "step": 138020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.743492711178387e-06,
+      "loss": 0.5993,
+      "step": 138030
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.740202046793248e-06,
+      "loss": 0.5444,
+      "step": 138040
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7369113824081084e-06,
+      "loss": 0.6274,
+      "step": 138050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.73362071802297e-06,
+      "loss": 0.6196,
+      "step": 138060
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.73033005363783e-06,
+      "loss": 0.556,
+      "step": 138070
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7270393892526905e-06,
+      "loss": 0.7061,
+      "step": 138080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.723748724867551e-06,
+      "loss": 0.6956,
+      "step": 138090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.720458060482412e-06,
+      "loss": 0.6371,
+      "step": 138100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.717167396097273e-06,
+      "loss": 0.472,
+      "step": 138110
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.713876731712133e-06,
+      "loss": 0.5967,
+      "step": 138120
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7105860673269935e-06,
+      "loss": 0.5867,
+      "step": 138130
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.707295402941854e-06,
+      "loss": 0.5784,
+      "step": 138140
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.704004738556715e-06,
+      "loss": 0.5561,
+      "step": 138150
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.700714074171576e-06,
+      "loss": 0.5578,
+      "step": 138160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.697423409786436e-06,
+      "loss": 0.5547,
+      "step": 138170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6941327454012964e-06,
+      "loss": 0.5885,
+      "step": 138180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.690842081016157e-06,
+      "loss": 0.5582,
+      "step": 138190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.687551416631018e-06,
+      "loss": 0.6503,
+      "step": 138200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6842607522458785e-06,
+      "loss": 0.5368,
+      "step": 138210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.680970087860739e-06,
+      "loss": 0.5513,
+      "step": 138220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6776794234756e-06,
+      "loss": 0.5814,
+      "step": 138230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.674388759090461e-06,
+      "loss": 0.5344,
+      "step": 138240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.671098094705322e-06,
+      "loss": 0.9527,
+      "step": 138250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.667807430320182e-06,
+      "loss": 0.612,
+      "step": 138260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.664516765935043e-06,
+      "loss": 0.542,
+      "step": 138270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.661226101549903e-06,
+      "loss": 0.5534,
+      "step": 138280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.657935437164764e-06,
+      "loss": 0.484,
+      "step": 138290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.654644772779625e-06,
+      "loss": 0.5865,
+      "step": 138300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.651354108394485e-06,
+      "loss": 0.5177,
+      "step": 138310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.648063444009346e-06,
+      "loss": 0.6176,
+      "step": 138320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.644772779624206e-06,
+      "loss": 0.5794,
+      "step": 138330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6414821152390666e-06,
+      "loss": 0.6328,
+      "step": 138340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.638191450853928e-06,
+      "loss": 0.5426,
+      "step": 138350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.634900786468788e-06,
+      "loss": 0.5791,
+      "step": 138360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.631610122083649e-06,
+      "loss": 0.5843,
+      "step": 138370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.628319457698509e-06,
+      "loss": 0.6126,
+      "step": 138380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6250287933133695e-06,
+      "loss": 0.5455,
+      "step": 138390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.621738128928231e-06,
+      "loss": 0.5796,
+      "step": 138400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.618447464543091e-06,
+      "loss": 0.631,
+      "step": 138410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6151568001579525e-06,
+      "loss": 0.562,
+      "step": 138420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.611866135772813e-06,
+      "loss": 0.6623,
+      "step": 138430
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.608575471387674e-06,
+      "loss": 0.6328,
+      "step": 138440
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6052848070025346e-06,
+      "loss": 0.5482,
+      "step": 138450
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.601994142617395e-06,
+      "loss": 0.5467,
+      "step": 138460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.598703478232255e-06,
+      "loss": 0.6427,
+      "step": 138470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.595412813847116e-06,
+      "loss": 0.6758,
+      "step": 138480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.592122149461977e-06,
+      "loss": 0.5327,
+      "step": 138490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5888314850768375e-06,
+      "loss": 0.5979,
+      "step": 138500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.585540820691698e-06,
+      "loss": 0.5981,
+      "step": 138510
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.582250156306558e-06,
+      "loss": 0.6038,
+      "step": 138520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.578959491921419e-06,
+      "loss": 0.5423,
+      "step": 138530
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.57566882753628e-06,
+      "loss": 0.6049,
+      "step": 138540
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5723781631511405e-06,
+      "loss": 0.5858,
+      "step": 138550
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.569087498766001e-06,
+      "loss": 0.6016,
+      "step": 138560
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.565796834380861e-06,
+      "loss": 0.5821,
+      "step": 138570
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.562506169995722e-06,
+      "loss": 0.4878,
+      "step": 138580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.559215505610583e-06,
+      "loss": 0.5631,
+      "step": 138590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.555924841225443e-06,
+      "loss": 0.6818,
+      "step": 138600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.552634176840305e-06,
+      "loss": 0.5849,
+      "step": 138610
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.549343512455165e-06,
+      "loss": 0.5591,
+      "step": 138620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5460528480700255e-06,
+      "loss": 0.5249,
+      "step": 138630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.542762183684887e-06,
+      "loss": 0.4951,
+      "step": 138640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.539471519299747e-06,
+      "loss": 0.5911,
+      "step": 138650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.536180854914608e-06,
+      "loss": 0.5865,
+      "step": 138660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.532890190529468e-06,
+      "loss": 0.6161,
+      "step": 138670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5295995261443285e-06,
+      "loss": 0.7049,
+      "step": 138680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.52630886175919e-06,
+      "loss": 0.7104,
+      "step": 138690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.52301819737405e-06,
+      "loss": 0.601,
+      "step": 138700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5197275329889106e-06,
+      "loss": 0.5562,
+      "step": 138710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.516436868603771e-06,
+      "loss": 0.5464,
+      "step": 138720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.513146204218631e-06,
+      "loss": 0.5787,
+      "step": 138730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.509855539833493e-06,
+      "loss": 0.5651,
+      "step": 138740
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.506564875448353e-06,
+      "loss": 0.5724,
+      "step": 138750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5032742110632135e-06,
+      "loss": 0.5651,
+      "step": 138760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.499983546678074e-06,
+      "loss": 0.6195,
+      "step": 138770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.496692882292935e-06,
+      "loss": 0.5804,
+      "step": 138780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.493402217907796e-06,
+      "loss": 0.5601,
+      "step": 138790
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.490111553522657e-06,
+      "loss": 0.6906,
+      "step": 138800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.486820889137517e-06,
+      "loss": 0.6416,
+      "step": 138810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.483530224752378e-06,
+      "loss": 0.5688,
+      "step": 138820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.480239560367239e-06,
+      "loss": 0.6355,
+      "step": 138830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4769488959820994e-06,
+      "loss": 0.4761,
+      "step": 138840
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.47365823159696e-06,
+      "loss": 0.6023,
+      "step": 138850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.47036756721182e-06,
+      "loss": 0.5404,
+      "step": 138860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.467076902826681e-06,
+      "loss": 0.5219,
+      "step": 138870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.463786238441542e-06,
+      "loss": 0.612,
+      "step": 138880
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.460495574056402e-06,
+      "loss": 0.557,
+      "step": 138890
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.457204909671263e-06,
+      "loss": 0.5247,
+      "step": 138900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.453914245286123e-06,
+      "loss": 0.6114,
+      "step": 138910
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.450623580900984e-06,
+      "loss": 0.6302,
+      "step": 138920
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.447332916515845e-06,
+      "loss": 0.5039,
+      "step": 138930
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.444042252130705e-06,
+      "loss": 0.5844,
+      "step": 138940
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.440751587745566e-06,
+      "loss": 0.5315,
+      "step": 138950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.437460923360426e-06,
+      "loss": 0.5779,
+      "step": 138960
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4341702589752874e-06,
+      "loss": 0.6307,
+      "step": 138970
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.430879594590148e-06,
+      "loss": 0.5525,
+      "step": 138980
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.427588930205009e-06,
+      "loss": 0.5238,
+      "step": 138990
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4242982658198695e-06,
+      "loss": 0.526,
+      "step": 139000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.42100760143473e-06,
+      "loss": 0.5717,
+      "step": 139010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.41771693704959e-06,
+      "loss": 0.5563,
+      "step": 139020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.414426272664452e-06,
+      "loss": 0.5094,
+      "step": 139030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.411135608279312e-06,
+      "loss": 0.6156,
+      "step": 139040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4078449438941725e-06,
+      "loss": 0.5129,
+      "step": 139050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.404554279509033e-06,
+      "loss": 0.5264,
+      "step": 139060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.401263615123893e-06,
+      "loss": 0.7,
+      "step": 139070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.397972950738755e-06,
+      "loss": 0.7005,
+      "step": 139080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.394682286353615e-06,
+      "loss": 0.5213,
+      "step": 139090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3913916219684754e-06,
+      "loss": 0.5883,
+      "step": 139100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.388100957583336e-06,
+      "loss": 0.7342,
+      "step": 139110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.384810293198196e-06,
+      "loss": 0.6184,
+      "step": 139120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3815196288130576e-06,
+      "loss": 0.5915,
+      "step": 139130
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.378228964427918e-06,
+      "loss": 0.5743,
+      "step": 139140
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.374938300042778e-06,
+      "loss": 0.638,
+      "step": 139150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.37164763565764e-06,
+      "loss": 0.5341,
+      "step": 139160
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3683569712725e-06,
+      "loss": 0.6073,
+      "step": 139170
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.365066306887361e-06,
+      "loss": 0.556,
+      "step": 139180
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.361775642502222e-06,
+      "loss": 0.5907,
+      "step": 139190
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.358484978117082e-06,
+      "loss": 0.7121,
+      "step": 139200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.355194313731943e-06,
+      "loss": 0.5473,
+      "step": 139210
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.351903649346804e-06,
+      "loss": 0.5842,
+      "step": 139220
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.348612984961664e-06,
+      "loss": 0.6672,
+      "step": 139230
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.345322320576525e-06,
+      "loss": 0.5669,
+      "step": 139240
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.342031656191385e-06,
+      "loss": 0.6574,
+      "step": 139250
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3387409918062456e-06,
+      "loss": 0.5699,
+      "step": 139260
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.335450327421107e-06,
+      "loss": 0.7881,
+      "step": 139270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.332159663035967e-06,
+      "loss": 0.43,
+      "step": 139280
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.328868998650828e-06,
+      "loss": 0.6424,
+      "step": 139290
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.325578334265688e-06,
+      "loss": 0.5898,
+      "step": 139300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3222876698805485e-06,
+      "loss": 0.5346,
+      "step": 139310
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.31899700549541e-06,
+      "loss": 0.4923,
+      "step": 139320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.31570634111027e-06,
+      "loss": 0.5936,
+      "step": 139330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.312415676725131e-06,
+      "loss": 0.585,
+      "step": 139340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.309125012339992e-06,
+      "loss": 0.5271,
+      "step": 139350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.305834347954852e-06,
+      "loss": 0.6659,
+      "step": 139360
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3025436835697136e-06,
+      "loss": 0.6856,
+      "step": 139370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.299253019184574e-06,
+      "loss": 0.5024,
+      "step": 139380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.295962354799434e-06,
+      "loss": 0.6126,
+      "step": 139390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.292671690414295e-06,
+      "loss": 0.5278,
+      "step": 139400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.289381026029155e-06,
+      "loss": 0.5226,
+      "step": 139410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2860903616440165e-06,
+      "loss": 0.4542,
+      "step": 139420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.282799697258877e-06,
+      "loss": 0.6488,
+      "step": 139430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.279509032873737e-06,
+      "loss": 0.7065,
+      "step": 139440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.276218368488598e-06,
+      "loss": 0.4627,
+      "step": 139450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.272927704103458e-06,
+      "loss": 0.6026,
+      "step": 139460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2696370397183195e-06,
+      "loss": 0.5257,
+      "step": 139470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.26634637533318e-06,
+      "loss": 0.5125,
+      "step": 139480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.26305571094804e-06,
+      "loss": 0.6527,
+      "step": 139490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.259765046562901e-06,
+      "loss": 0.5534,
+      "step": 139500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.256474382177762e-06,
+      "loss": 0.7391,
+      "step": 139510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.253183717792622e-06,
+      "loss": 0.6753,
+      "step": 139520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.249893053407483e-06,
+      "loss": 0.5758,
+      "step": 139530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.246602389022344e-06,
+      "loss": 0.5975,
+      "step": 139540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2433117246372045e-06,
+      "loss": 0.6253,
+      "step": 139550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.240021060252066e-06,
+      "loss": 0.6489,
+      "step": 139560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.236730395866926e-06,
+      "loss": 0.6129,
+      "step": 139570
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.233439731481787e-06,
+      "loss": 0.472,
+      "step": 139580
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.230149067096647e-06,
+      "loss": 0.5668,
+      "step": 139590
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2268584027115075e-06,
+      "loss": 0.6225,
+      "step": 139600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.223567738326369e-06,
+      "loss": 0.5073,
+      "step": 139610
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.220277073941229e-06,
+      "loss": 0.5548,
+      "step": 139620
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.21698640955609e-06,
+      "loss": 0.5386,
+      "step": 139630
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.21369574517095e-06,
+      "loss": 0.5799,
+      "step": 139640
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2104050807858104e-06,
+      "loss": 0.5455,
+      "step": 139650
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.207114416400672e-06,
+      "loss": 0.5477,
+      "step": 139660
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.203823752015532e-06,
+      "loss": 0.5784,
+      "step": 139670
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2005330876303925e-06,
+      "loss": 0.4123,
+      "step": 139680
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.197242423245253e-06,
+      "loss": 0.573,
+      "step": 139690
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.193951758860113e-06,
+      "loss": 0.623,
+      "step": 139700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.190661094474975e-06,
+      "loss": 0.548,
+      "step": 139710
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.187370430089835e-06,
+      "loss": 0.6246,
+      "step": 139720
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.184079765704696e-06,
+      "loss": 0.6945,
+      "step": 139730
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.180789101319557e-06,
+      "loss": 0.5746,
+      "step": 139740
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.177498436934417e-06,
+      "loss": 0.4714,
+      "step": 139750
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1742077725492784e-06,
+      "loss": 0.4711,
+      "step": 139760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.170917108164139e-06,
+      "loss": 0.5301,
+      "step": 139770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.167626443778999e-06,
+      "loss": 0.5566,
+      "step": 139780
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.16433577939386e-06,
+      "loss": 0.5101,
+      "step": 139790
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.16104511500872e-06,
+      "loss": 0.4507,
+      "step": 139800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.157754450623581e-06,
+      "loss": 0.5584,
+      "step": 139810
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.154463786238442e-06,
+      "loss": 0.5882,
+      "step": 139820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.151173121853302e-06,
+      "loss": 0.5239,
+      "step": 139830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.147882457468163e-06,
+      "loss": 0.5328,
+      "step": 139840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.144591793083024e-06,
+      "loss": 0.4987,
+      "step": 139850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.141301128697884e-06,
+      "loss": 0.6506,
+      "step": 139860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.138010464312745e-06,
+      "loss": 0.7026,
+      "step": 139870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.134719799927605e-06,
+      "loss": 0.6862,
+      "step": 139880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.131429135542466e-06,
+      "loss": 0.6747,
+      "step": 139890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.128138471157327e-06,
+      "loss": 0.502,
+      "step": 139900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.124847806772187e-06,
+      "loss": 0.6235,
+      "step": 139910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1215571423870486e-06,
+      "loss": 0.6162,
+      "step": 139920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.118266478001909e-06,
+      "loss": 0.7186,
+      "step": 139930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.114975813616769e-06,
+      "loss": 0.5335,
+      "step": 139940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.111685149231631e-06,
+      "loss": 0.7058,
+      "step": 139950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.108394484846491e-06,
+      "loss": 0.4903,
+      "step": 139960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1051038204613515e-06,
+      "loss": 0.5391,
+      "step": 139970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.101813156076212e-06,
+      "loss": 0.5521,
+      "step": 139980
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.098522491691072e-06,
+      "loss": 0.6903,
+      "step": 139990
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.095231827305934e-06,
+      "loss": 0.6074,
+      "step": 140000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.091941162920794e-06,
+      "loss": 0.574,
+      "step": 140010
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0886504985356545e-06,
+      "loss": 0.5731,
+      "step": 140020
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.085359834150515e-06,
+      "loss": 0.5458,
+      "step": 140030
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.082069169765375e-06,
+      "loss": 0.5385,
+      "step": 140040
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0787785053802366e-06,
+      "loss": 0.6248,
+      "step": 140050
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.075487840995097e-06,
+      "loss": 0.6071,
+      "step": 140060
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.072197176609957e-06,
+      "loss": 0.5528,
+      "step": 140070
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.068906512224818e-06,
+      "loss": 0.5741,
+      "step": 140080
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.065615847839679e-06,
+      "loss": 0.5674,
+      "step": 140090
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0623251834545395e-06,
+      "loss": 0.4627,
+      "step": 140100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.059034519069401e-06,
+      "loss": 0.5985,
+      "step": 140110
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.055743854684261e-06,
+      "loss": 0.5778,
+      "step": 140120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.052453190299122e-06,
+      "loss": 0.6355,
+      "step": 140130
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.049162525913982e-06,
+      "loss": 0.6278,
+      "step": 140140
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.045871861528843e-06,
+      "loss": 0.5484,
+      "step": 140150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.042581197143704e-06,
+      "loss": 0.5559,
+      "step": 140160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.039290532758564e-06,
+      "loss": 0.4826,
+      "step": 140170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0359998683734246e-06,
+      "loss": 0.5826,
+      "step": 140180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.032709203988285e-06,
+      "loss": 0.4697,
+      "step": 140190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.029418539603146e-06,
+      "loss": 0.4962,
+      "step": 140200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.026127875218007e-06,
+      "loss": 0.612,
+      "step": 140210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.022837210832867e-06,
+      "loss": 0.4786,
+      "step": 140220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0195465464477275e-06,
+      "loss": 0.5548,
+      "step": 140230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.016255882062589e-06,
+      "loss": 0.452,
+      "step": 140240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.012965217677449e-06,
+      "loss": 0.5789,
+      "step": 140250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.00967455329231e-06,
+      "loss": 0.4727,
+      "step": 140260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.00638388890717e-06,
+      "loss": 0.5887,
+      "step": 140270
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.003093224522031e-06,
+      "loss": 0.5467,
+      "step": 140280
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.999802560136892e-06,
+      "loss": 0.5997,
+      "step": 140290
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.996511895751753e-06,
+      "loss": 0.6206,
+      "step": 140300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.993221231366613e-06,
+      "loss": 0.5227,
+      "step": 140310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.989930566981474e-06,
+      "loss": 0.5985,
+      "step": 140320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.986639902596334e-06,
+      "loss": 0.5246,
+      "step": 140330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9833492382111955e-06,
+      "loss": 0.645,
+      "step": 140340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.980058573826056e-06,
+      "loss": 0.5679,
+      "step": 140350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.976767909440916e-06,
+      "loss": 0.5598,
+      "step": 140360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.973477245055777e-06,
+      "loss": 0.5135,
+      "step": 140370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.970186580670637e-06,
+      "loss": 0.5751,
+      "step": 140380
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9668959162854985e-06,
+      "loss": 0.5893,
+      "step": 140390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.963605251900359e-06,
+      "loss": 0.6624,
+      "step": 140400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.960314587515219e-06,
+      "loss": 0.5639,
+      "step": 140410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.95702392313008e-06,
+      "loss": 0.6284,
+      "step": 140420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.95373325874494e-06,
+      "loss": 0.5365,
+      "step": 140430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9504425943598014e-06,
+      "loss": 0.5713,
+      "step": 140440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.947151929974662e-06,
+      "loss": 0.5493,
+      "step": 140450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.943861265589522e-06,
+      "loss": 0.574,
+      "step": 140460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9405706012043835e-06,
+      "loss": 0.5827,
+      "step": 140470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.937279936819244e-06,
+      "loss": 0.658,
+      "step": 140480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.933989272434105e-06,
+      "loss": 0.6103,
+      "step": 140490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.930698608048966e-06,
+      "loss": 0.5819,
+      "step": 140500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.927407943663826e-06,
+      "loss": 0.5845,
+      "step": 140510
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.9241172792786865e-06,
+      "loss": 0.6659,
+      "step": 140520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.920826614893547e-06,
+      "loss": 0.5815,
+      "step": 140530
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.917535950508408e-06,
+      "loss": 0.5429,
+      "step": 140540
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.914245286123269e-06,
+      "loss": 0.5801,
+      "step": 140550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.910954621738129e-06,
+      "loss": 0.5856,
+      "step": 140560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.9076639573529894e-06,
+      "loss": 0.5506,
+      "step": 140570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.904373292967851e-06,
+      "loss": 0.5557,
+      "step": 140580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.901082628582711e-06,
+      "loss": 0.6393,
+      "step": 140590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8977919641975715e-06,
+      "loss": 0.468,
+      "step": 140600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.894501299812432e-06,
+      "loss": 0.6108,
+      "step": 140610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.891210635427292e-06,
+      "loss": 0.6251,
+      "step": 140620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.887919971042154e-06,
+      "loss": 0.5105,
+      "step": 140630
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.884629306657014e-06,
+      "loss": 0.6746,
+      "step": 140640
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8813386422718745e-06,
+      "loss": 0.6486,
+      "step": 140650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.878047977886736e-06,
+      "loss": 0.5784,
+      "step": 140660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.874757313501596e-06,
+      "loss": 0.6271,
+      "step": 140670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8714666491164574e-06,
+      "loss": 0.5697,
+      "step": 140680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.868175984731318e-06,
+      "loss": 0.5819,
+      "step": 140690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.864885320346178e-06,
+      "loss": 0.5961,
+      "step": 140700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.861594655961039e-06,
+      "loss": 0.5501,
+      "step": 140710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.858303991575899e-06,
+      "loss": 0.5908,
+      "step": 140720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.85501332719076e-06,
+      "loss": 0.5558,
+      "step": 140730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.851722662805621e-06,
+      "loss": 0.4805,
+      "step": 140740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.848431998420481e-06,
+      "loss": 0.5785,
+      "step": 140750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.845141334035342e-06,
+      "loss": 0.6067,
+      "step": 140760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.841850669650202e-06,
+      "loss": 0.6385,
+      "step": 140770
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.838560005265063e-06,
+      "loss": 0.5866,
+      "step": 140780
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.835269340879924e-06,
+      "loss": 0.5779,
+      "step": 140790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.831978676494784e-06,
+      "loss": 0.5366,
+      "step": 140800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.828688012109645e-06,
+      "loss": 0.5041,
+      "step": 140810
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.825397347724505e-06,
+      "loss": 0.605,
+      "step": 140820
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.822106683339366e-06,
+      "loss": 0.437,
+      "step": 140830
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.818816018954227e-06,
+      "loss": 0.618,
+      "step": 140840
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.815525354569088e-06,
+      "loss": 0.5633,
+      "step": 140850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.812234690183948e-06,
+      "loss": 0.583,
+      "step": 140860
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.808944025798809e-06,
+      "loss": 0.5352,
+      "step": 140870
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8056533614136697e-06,
+      "loss": 0.4818,
+      "step": 140880
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8023626970285305e-06,
+      "loss": 0.6919,
+      "step": 140890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.799072032643391e-06,
+      "loss": 0.6864,
+      "step": 140900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7957813682582514e-06,
+      "loss": 0.554,
+      "step": 140910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7924907038731126e-06,
+      "loss": 0.5697,
+      "step": 140920
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.789200039487973e-06,
+      "loss": 0.5256,
+      "step": 140930
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7859093751028335e-06,
+      "loss": 0.5465,
+      "step": 140940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.782618710717694e-06,
+      "loss": 0.5793,
+      "step": 140950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7793280463325543e-06,
+      "loss": 0.583,
+      "step": 140960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7760373819474156e-06,
+      "loss": 0.4974,
+      "step": 140970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.772746717562276e-06,
+      "loss": 0.7874,
+      "step": 140980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.769456053177137e-06,
+      "loss": 0.645,
+      "step": 140990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7661653887919973e-06,
+      "loss": 0.6617,
+      "step": 141000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7628747244068577e-06,
+      "loss": 0.5715,
+      "step": 141010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.759584060021719e-06,
+      "loss": 0.5421,
+      "step": 141020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7562933956365794e-06,
+      "loss": 0.5522,
+      "step": 141030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7530027312514398e-06,
+      "loss": 0.6106,
+      "step": 141040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7497120668663e-06,
+      "loss": 0.5682,
+      "step": 141050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.746421402481161e-06,
+      "loss": 0.5979,
+      "step": 141060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.743130738096022e-06,
+      "loss": 0.626,
+      "step": 141070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7398400737108827e-06,
+      "loss": 0.6738,
+      "step": 141080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.736549409325743e-06,
+      "loss": 0.6191,
+      "step": 141090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7332587449406036e-06,
+      "loss": 0.6264,
+      "step": 141100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.729968080555464e-06,
+      "loss": 0.529,
+      "step": 141110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7266774161703253e-06,
+      "loss": 0.5428,
+      "step": 141120
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7233867517851857e-06,
+      "loss": 1.0503,
+      "step": 141130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.720096087400046e-06,
+      "loss": 0.5404,
+      "step": 141140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7168054230149065e-06,
+      "loss": 0.6335,
+      "step": 141150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7135147586297674e-06,
+      "loss": 0.6194,
+      "step": 141160
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.710224094244628e-06,
+      "loss": 0.5528,
+      "step": 141170
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.706933429859489e-06,
+      "loss": 0.6571,
+      "step": 141180
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7036427654743495e-06,
+      "loss": 0.6107,
+      "step": 141190
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.70035210108921e-06,
+      "loss": 0.5305,
+      "step": 141200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6970614367040703e-06,
+      "loss": 0.6693,
+      "step": 141210
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6937707723189316e-06,
+      "loss": 0.6236,
+      "step": 141220
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.690480107933792e-06,
+      "loss": 0.5797,
+      "step": 141230
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6871894435486524e-06,
+      "loss": 0.6001,
+      "step": 141240
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6838987791635133e-06,
+      "loss": 0.5449,
+      "step": 141250
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6806081147783737e-06,
+      "loss": 0.4868,
+      "step": 141260
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.677317450393235e-06,
+      "loss": 0.4336,
+      "step": 141270
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6740267860080954e-06,
+      "loss": 0.6184,
+      "step": 141280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.670736121622956e-06,
+      "loss": 0.6158,
+      "step": 141290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6674454572378162e-06,
+      "loss": 0.4206,
+      "step": 141300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6641547928526775e-06,
+      "loss": 0.5582,
+      "step": 141310
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.660864128467538e-06,
+      "loss": 0.5434,
+      "step": 141320
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6575734640823983e-06,
+      "loss": 0.5532,
+      "step": 141330
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6542827996972587e-06,
+      "loss": 0.5734,
+      "step": 141340
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6509921353121196e-06,
+      "loss": 0.516,
+      "step": 141350
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6477014709269804e-06,
+      "loss": 0.657,
+      "step": 141360
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6444108065418413e-06,
+      "loss": 0.6793,
+      "step": 141370
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6411201421567017e-06,
+      "loss": 0.7056,
+      "step": 141380
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.637829477771562e-06,
+      "loss": 0.5448,
+      "step": 141390
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6345388133864225e-06,
+      "loss": 0.5177,
+      "step": 141400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.631248149001284e-06,
+      "loss": 0.6096,
+      "step": 141410
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6279574846161442e-06,
+      "loss": 0.5482,
+      "step": 141420
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6246668202310046e-06,
+      "loss": 0.5632,
+      "step": 141430
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6213761558458655e-06,
+      "loss": 0.5184,
+      "step": 141440
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.618085491460726e-06,
+      "loss": 0.6404,
+      "step": 141450
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.614794827075587e-06,
+      "loss": 0.5176,
+      "step": 141460
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6115041626904476e-06,
+      "loss": 0.5642,
+      "step": 141470
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.608213498305308e-06,
+      "loss": 0.5069,
+      "step": 141480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6049228339201684e-06,
+      "loss": 0.5917,
+      "step": 141490
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.601632169535029e-06,
+      "loss": 0.5068,
+      "step": 141500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.59834150514989e-06,
+      "loss": 0.5625,
+      "step": 141510
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5950508407647505e-06,
+      "loss": 0.5728,
+      "step": 141520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.591760176379611e-06,
+      "loss": 0.6089,
+      "step": 141530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.588469511994472e-06,
+      "loss": 0.6417,
+      "step": 141540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5851788476093322e-06,
+      "loss": 0.6636,
+      "step": 141550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5818881832241935e-06,
+      "loss": 0.4929,
+      "step": 141560
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.578597518839054e-06,
+      "loss": 0.5266,
+      "step": 141570
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5753068544539143e-06,
+      "loss": 0.6626,
+      "step": 141580
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5720161900687748e-06,
+      "loss": 0.6018,
+      "step": 141590
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.568725525683635e-06,
+      "loss": 0.5004,
+      "step": 141600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5654348612984965e-06,
+      "loss": 0.5554,
+      "step": 141610
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.562144196913357e-06,
+      "loss": 0.5361,
+      "step": 141620
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5588535325282177e-06,
+      "loss": 0.5672,
+      "step": 141630
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.555562868143078e-06,
+      "loss": 0.5055,
+      "step": 141640
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5522722037579394e-06,
+      "loss": 0.5744,
+      "step": 141650
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5489815393728e-06,
+      "loss": 0.7216,
+      "step": 141660
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5456908749876602e-06,
+      "loss": 0.5408,
+      "step": 141670
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5424002106025207e-06,
+      "loss": 0.5485,
+      "step": 141680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.539109546217381e-06,
+      "loss": 0.6377,
+      "step": 141690
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5358188818322424e-06,
+      "loss": 0.5444,
+      "step": 141700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5325282174471028e-06,
+      "loss": 0.6397,
+      "step": 141710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5292375530619636e-06,
+      "loss": 0.5626,
+      "step": 141720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.525946888676824e-06,
+      "loss": 0.5751,
+      "step": 141730
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5226562242916845e-06,
+      "loss": 0.6804,
+      "step": 141740
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5193655599065457e-06,
+      "loss": 0.61,
+      "step": 141750
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.516074895521406e-06,
+      "loss": 0.5736,
+      "step": 141760
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5127842311362666e-06,
+      "loss": 0.5088,
+      "step": 141770
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.509493566751127e-06,
+      "loss": 0.5935,
+      "step": 141780
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5062029023659874e-06,
+      "loss": 0.6605,
+      "step": 141790
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5029122379808487e-06,
+      "loss": 0.6455,
+      "step": 141800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.499621573595709e-06,
+      "loss": 0.5758,
+      "step": 141810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.49633090921057e-06,
+      "loss": 0.5641,
+      "step": 141820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4930402448254304e-06,
+      "loss": 0.543,
+      "step": 141830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4897495804402908e-06,
+      "loss": 0.6669,
+      "step": 141840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.486458916055152e-06,
+      "loss": 0.5949,
+      "step": 141850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4831682516700125e-06,
+      "loss": 0.8208,
+      "step": 141860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.479877587284873e-06,
+      "loss": 0.5647,
+      "step": 141870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4765869228997333e-06,
+      "loss": 0.5504,
+      "step": 141880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.473296258514594e-06,
+      "loss": 0.6158,
+      "step": 141890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.470005594129455e-06,
+      "loss": 0.6026,
+      "step": 141900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.466714929744316e-06,
+      "loss": 0.5434,
+      "step": 141910
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4634242653591763e-06,
+      "loss": 0.598,
+      "step": 141920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4601336009740367e-06,
+      "loss": 0.511,
+      "step": 141930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.456842936588897e-06,
+      "loss": 0.6257,
+      "step": 141940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4535522722037584e-06,
+      "loss": 0.6866,
+      "step": 141950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.450261607818619e-06,
+      "loss": 0.5497,
+      "step": 141960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.446970943433479e-06,
+      "loss": 0.6016,
+      "step": 141970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4436802790483396e-06,
+      "loss": 0.5742,
+      "step": 141980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.440389614663201e-06,
+      "loss": 0.6418,
+      "step": 141990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4370989502780613e-06,
+      "loss": 0.536,
+      "step": 142000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.433808285892922e-06,
+      "loss": 0.6038,
+      "step": 142010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4305176215077826e-06,
+      "loss": 0.5318,
+      "step": 142020
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.427226957122643e-06,
+      "loss": 0.5901,
+      "step": 142030
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4239362927375043e-06,
+      "loss": 0.6213,
+      "step": 142040
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4206456283523647e-06,
+      "loss": 0.588,
+      "step": 142050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.417354963967225e-06,
+      "loss": 0.5662,
+      "step": 142060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4140642995820855e-06,
+      "loss": 0.6758,
+      "step": 142070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4107736351969464e-06,
+      "loss": 0.5518,
+      "step": 142080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4074829708118072e-06,
+      "loss": 0.5967,
+      "step": 142090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.404192306426668e-06,
+      "loss": 0.5262,
+      "step": 142100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4009016420415285e-06,
+      "loss": 0.5305,
+      "step": 142110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.397610977656389e-06,
+      "loss": 0.6477,
+      "step": 142120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3943203132712493e-06,
+      "loss": 0.476,
+      "step": 142130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3910296488861106e-06,
+      "loss": 0.5469,
+      "step": 142140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.387738984500971e-06,
+      "loss": 0.6125,
+      "step": 142150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3844483201158314e-06,
+      "loss": 0.5679,
+      "step": 142160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.381157655730692e-06,
+      "loss": 0.6934,
+      "step": 142170
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3778669913455527e-06,
+      "loss": 0.5688,
+      "step": 142180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3745763269604135e-06,
+      "loss": 0.6992,
+      "step": 142190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3712856625752744e-06,
+      "loss": 0.5123,
+      "step": 142200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.367994998190135e-06,
+      "loss": 0.7419,
+      "step": 142210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3647043338049952e-06,
+      "loss": 0.6309,
+      "step": 142220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3614136694198556e-06,
+      "loss": 0.5462,
+      "step": 142230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.358123005034717e-06,
+      "loss": 0.5455,
+      "step": 142240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3548323406495773e-06,
+      "loss": 0.6478,
+      "step": 142250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3515416762644378e-06,
+      "loss": 0.5574,
+      "step": 142260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3482510118792986e-06,
+      "loss": 0.6085,
+      "step": 142270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.344960347494159e-06,
+      "loss": 0.598,
+      "step": 142280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3416696831090203e-06,
+      "loss": 0.6002,
+      "step": 142290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3383790187238807e-06,
+      "loss": 0.5889,
+      "step": 142300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.335088354338741e-06,
+      "loss": 0.6233,
+      "step": 142310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3317976899536015e-06,
+      "loss": 0.5964,
+      "step": 142320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.328507025568463e-06,
+      "loss": 0.5799,
+      "step": 142330
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3252163611833232e-06,
+      "loss": 0.5239,
+      "step": 142340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3219256967981837e-06,
+      "loss": 0.7403,
+      "step": 142350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.318635032413044e-06,
+      "loss": 0.5988,
+      "step": 142360
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.315344368027905e-06,
+      "loss": 0.5839,
+      "step": 142370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3120537036427658e-06,
+      "loss": 0.6662,
+      "step": 142380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3087630392576266e-06,
+      "loss": 0.5326,
+      "step": 142390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.305472374872487e-06,
+      "loss": 0.5981,
+      "step": 142400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3021817104873474e-06,
+      "loss": 0.6283,
+      "step": 142410
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.298891046102208e-06,
+      "loss": 0.6394,
+      "step": 142420
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.295600381717069e-06,
+      "loss": 0.5886,
+      "step": 142430
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2923097173319296e-06,
+      "loss": 0.5199,
+      "step": 142440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.28901905294679e-06,
+      "loss": 0.5341,
+      "step": 142450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.285728388561651e-06,
+      "loss": 0.573,
+      "step": 142460
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2824377241765112e-06,
+      "loss": 0.5825,
+      "step": 142470
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2791470597913725e-06,
+      "loss": 0.5165,
+      "step": 142480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.275856395406233e-06,
+      "loss": 0.449,
+      "step": 142490
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2725657310210934e-06,
+      "loss": 0.5077,
+      "step": 142500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2692750666359538e-06,
+      "loss": 0.6445,
+      "step": 142510
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.265984402250814e-06,
+      "loss": 0.5999,
+      "step": 142520
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2626937378656755e-06,
+      "loss": 0.718,
+      "step": 142530
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.259403073480536e-06,
+      "loss": 0.5323,
+      "step": 142540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2561124090953963e-06,
+      "loss": 0.6449,
+      "step": 142550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.252821744710257e-06,
+      "loss": 0.6098,
+      "step": 142560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2495310803251176e-06,
+      "loss": 0.4754,
+      "step": 142570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.246240415939979e-06,
+      "loss": 0.4938,
+      "step": 142580
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2429497515548393e-06,
+      "loss": 0.6354,
+      "step": 142590
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2396590871696997e-06,
+      "loss": 0.6765,
+      "step": 142600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.23636842278456e-06,
+      "loss": 0.4876,
+      "step": 142610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2330777583994205e-06,
+      "loss": 0.5703,
+      "step": 142620
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2297870940142818e-06,
+      "loss": 0.5223,
+      "step": 142630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.226496429629142e-06,
+      "loss": 0.6323,
+      "step": 142640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.223205765244003e-06,
+      "loss": 0.5633,
+      "step": 142650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2199151008588635e-06,
+      "loss": 0.6271,
+      "step": 142660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.216624436473724e-06,
+      "loss": 0.4772,
+      "step": 142670
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.213333772088585e-06,
+      "loss": 0.5295,
+      "step": 142680
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2100431077034456e-06,
+      "loss": 0.5667,
+      "step": 142690
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.206752443318306e-06,
+      "loss": 0.5751,
+      "step": 142700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2034617789331664e-06,
+      "loss": 0.7515,
+      "step": 142710
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2001711145480277e-06,
+      "loss": 0.5126,
+      "step": 142720
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.196880450162888e-06,
+      "loss": 0.5911,
+      "step": 142730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1935897857777485e-06,
+      "loss": 0.6475,
+      "step": 142740
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1902991213926094e-06,
+      "loss": 0.655,
+      "step": 142750
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1870084570074698e-06,
+      "loss": 0.5949,
+      "step": 142760
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.183717792622331e-06,
+      "loss": 0.7519,
+      "step": 142770
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1804271282371915e-06,
+      "loss": 0.4322,
+      "step": 142780
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.177136463852052e-06,
+      "loss": 0.6174,
+      "step": 142790
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1738457994669123e-06,
+      "loss": 0.5739,
+      "step": 142800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1705551350817727e-06,
+      "loss": 0.6118,
+      "step": 142810
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.167264470696634e-06,
+      "loss": 0.499,
+      "step": 142820
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1639738063114944e-06,
+      "loss": 0.6681,
+      "step": 142830
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1606831419263553e-06,
+      "loss": 0.5273,
+      "step": 142840
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1573924775412157e-06,
+      "loss": 0.5625,
+      "step": 142850
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.154101813156076e-06,
+      "loss": 0.4807,
+      "step": 142860
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1508111487709374e-06,
+      "loss": 0.6203,
+      "step": 142870
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.147520484385798e-06,
+      "loss": 0.4936,
+      "step": 142880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1442298200006582e-06,
+      "loss": 0.5857,
+      "step": 142890
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1409391556155186e-06,
+      "loss": 0.5821,
+      "step": 142900
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.137648491230379e-06,
+      "loss": 0.5892,
+      "step": 142910
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1343578268452403e-06,
+      "loss": 0.5422,
+      "step": 142920
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1310671624601007e-06,
+      "loss": 0.4881,
+      "step": 142930
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1277764980749616e-06,
+      "loss": 0.6123,
+      "step": 142940
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1244858336898224e-06,
+      "loss": 0.6393,
+      "step": 142950
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.121195169304683e-06,
+      "loss": 0.4891,
+      "step": 142960
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1179045049195433e-06,
+      "loss": 0.5561,
+      "step": 142970
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.114613840534404e-06,
+      "loss": 0.4671,
+      "step": 142980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1113231761492645e-06,
+      "loss": 0.5428,
+      "step": 142990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1080325117641254e-06,
+      "loss": 0.6093,
+      "step": 143000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.104741847378986e-06,
+      "loss": 0.5154,
+      "step": 143010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1014511829938466e-06,
+      "loss": 0.5536,
+      "step": 143020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.0981605186087075e-06,
+      "loss": 0.5983,
+      "step": 143030
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.094869854223568e-06,
+      "loss": 0.4925,
+      "step": 143040
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0915791898384288e-06,
+      "loss": 0.5589,
+      "step": 143050
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.088288525453289e-06,
+      "loss": 0.5075,
+      "step": 143060
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0849978610681496e-06,
+      "loss": 0.5311,
+      "step": 143070
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0817071966830104e-06,
+      "loss": 0.6246,
+      "step": 143080
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.078416532297871e-06,
+      "loss": 0.5766,
+      "step": 143090
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0751258679127317e-06,
+      "loss": 0.5257,
+      "step": 143100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0718352035275925e-06,
+      "loss": 0.4651,
+      "step": 143110
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.068544539142453e-06,
+      "loss": 0.6231,
+      "step": 143120
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.065253874757314e-06,
+      "loss": 0.5282,
+      "step": 143130
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0619632103721742e-06,
+      "loss": 0.5922,
+      "step": 143140
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.058672545987035e-06,
+      "loss": 0.5811,
+      "step": 143150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0553818816018955e-06,
+      "loss": 0.6123,
+      "step": 143160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0520912172167563e-06,
+      "loss": 0.5912,
+      "step": 143170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0488005528316168e-06,
+      "loss": 0.577,
+      "step": 143180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.045509888446477e-06,
+      "loss": 0.5134,
+      "step": 143190
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.042219224061338e-06,
+      "loss": 0.5507,
+      "step": 143200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.038928559676199e-06,
+      "loss": 0.5636,
+      "step": 143210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0356378952910597e-06,
+      "loss": 0.5416,
+      "step": 143220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.03234723090592e-06,
+      "loss": 0.5082,
+      "step": 143230
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0290565665207806e-06,
+      "loss": 0.5869,
+      "step": 143240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0257659021356414e-06,
+      "loss": 0.5327,
+      "step": 143250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.022475237750502e-06,
+      "loss": 0.5853,
+      "step": 143260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0191845733653627e-06,
+      "loss": 0.5346,
+      "step": 143270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.015893908980223e-06,
+      "loss": 0.5388,
+      "step": 143280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.012603244595084e-06,
+      "loss": 0.509,
+      "step": 143290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0093125802099448e-06,
+      "loss": 0.535,
+      "step": 143300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.006021915824805e-06,
+      "loss": 0.5525,
+      "step": 143310
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.002731251439666e-06,
+      "loss": 0.488,
+      "step": 143320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9994405870545265e-06,
+      "loss": 0.5606,
+      "step": 143330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9961499226693873e-06,
+      "loss": 0.7506,
+      "step": 143340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9928592582842477e-06,
+      "loss": 0.7901,
+      "step": 143350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.989568593899108e-06,
+      "loss": 0.5321,
+      "step": 143360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.986277929513969e-06,
+      "loss": 0.6444,
+      "step": 143370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9829872651288294e-06,
+      "loss": 0.4739,
+      "step": 143380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9796966007436903e-06,
+      "loss": 0.6397,
+      "step": 143390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.976405936358551e-06,
+      "loss": 0.64,
+      "step": 143400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9731152719734115e-06,
+      "loss": 0.5129,
+      "step": 143410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9698246075882724e-06,
+      "loss": 0.6017,
+      "step": 143420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9665339432031328e-06,
+      "loss": 0.761,
+      "step": 143430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9632432788179936e-06,
+      "loss": 0.4836,
+      "step": 143440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.959952614432854e-06,
+      "loss": 0.5103,
+      "step": 143450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9566619500477145e-06,
+      "loss": 0.7136,
+      "step": 143460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9533712856625753e-06,
+      "loss": 0.6489,
+      "step": 143470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.950080621277436e-06,
+      "loss": 0.5671,
+      "step": 143480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.946789956892297e-06,
+      "loss": 0.4301,
+      "step": 143490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9434992925071574e-06,
+      "loss": 0.5198,
+      "step": 143500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9402086281220183e-06,
+      "loss": 0.5982,
+      "step": 143510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9369179637368787e-06,
+      "loss": 0.4667,
+      "step": 143520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.933627299351739e-06,
+      "loss": 0.5811,
+      "step": 143530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9303366349666e-06,
+      "loss": 0.5673,
+      "step": 143540
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9270459705814604e-06,
+      "loss": 0.5849,
+      "step": 143550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.923755306196321e-06,
+      "loss": 0.5427,
+      "step": 143560
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9204646418111816e-06,
+      "loss": 0.6185,
+      "step": 143570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9171739774260425e-06,
+      "loss": 0.5468,
+      "step": 143580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9138833130409033e-06,
+      "loss": 0.689,
+      "step": 143590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9105926486557637e-06,
+      "loss": 0.5846,
+      "step": 143600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9073019842706246e-06,
+      "loss": 0.5872,
+      "step": 143610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.904011319885485e-06,
+      "loss": 0.628,
+      "step": 143620
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9007206555003454e-06,
+      "loss": 0.5745,
+      "step": 143630
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8974299911152063e-06,
+      "loss": 0.5999,
+      "step": 143640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8941393267300667e-06,
+      "loss": 0.5154,
+      "step": 143650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8908486623449275e-06,
+      "loss": 0.6542,
+      "step": 143660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8875579979597884e-06,
+      "loss": 0.557,
+      "step": 143670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8842673335746492e-06,
+      "loss": 0.6116,
+      "step": 143680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8809766691895096e-06,
+      "loss": 0.5892,
+      "step": 143690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.87768600480437e-06,
+      "loss": 0.6245,
+      "step": 143700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.874395340419231e-06,
+      "loss": 0.6073,
+      "step": 143710
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8711046760340913e-06,
+      "loss": 0.5231,
+      "step": 143720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.867814011648952e-06,
+      "loss": 0.5836,
+      "step": 143730
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8645233472638126e-06,
+      "loss": 0.5223,
+      "step": 143740
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.861232682878673e-06,
+      "loss": 0.6003,
+      "step": 143750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.857942018493534e-06,
+      "loss": 0.5494,
+      "step": 143760
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8546513541083947e-06,
+      "loss": 0.6547,
+      "step": 143770
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8513606897232555e-06,
+      "loss": 0.4662,
+      "step": 143780
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.848070025338116e-06,
+      "loss": 0.5591,
+      "step": 143790
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8447793609529764e-06,
+      "loss": 0.5097,
+      "step": 143800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8414886965678372e-06,
+      "loss": 0.5826,
+      "step": 143810
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8381980321826976e-06,
+      "loss": 0.5174,
+      "step": 143820
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8349073677975585e-06,
+      "loss": 0.5807,
+      "step": 143830
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.831616703412419e-06,
+      "loss": 0.5355,
+      "step": 143840
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8283260390272798e-06,
+      "loss": 0.5144,
+      "step": 143850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8250353746421406e-06,
+      "loss": 0.5668,
+      "step": 143860
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.821744710257001e-06,
+      "loss": 0.5554,
+      "step": 143870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.818454045871862e-06,
+      "loss": 0.7056,
+      "step": 143880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8151633814867223e-06,
+      "loss": 0.5828,
+      "step": 143890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.811872717101583e-06,
+      "loss": 0.6005,
+      "step": 143900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8085820527164435e-06,
+      "loss": 0.4592,
+      "step": 143910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.805291388331304e-06,
+      "loss": 0.6021,
+      "step": 143920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.802000723946165e-06,
+      "loss": 0.7392,
+      "step": 143930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7987100595610252e-06,
+      "loss": 0.5065,
+      "step": 143940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.795419395175886e-06,
+      "loss": 0.5536,
+      "step": 143950
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.792128730790747e-06,
+      "loss": 0.5375,
+      "step": 143960
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7888380664056073e-06,
+      "loss": 0.622,
+      "step": 143970
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.785547402020468e-06,
+      "loss": 0.568,
+      "step": 143980
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7822567376353286e-06,
+      "loss": 0.5928,
+      "step": 143990
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7789660732501894e-06,
+      "loss": 0.6437,
+      "step": 144000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.77567540886505e-06,
+      "loss": 0.6521,
+      "step": 144010
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7723847444799107e-06,
+      "loss": 0.5512,
+      "step": 144020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.769094080094771e-06,
+      "loss": 0.6083,
+      "step": 144030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.765803415709632e-06,
+      "loss": 0.5845,
+      "step": 144040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.762512751324493e-06,
+      "loss": 0.5637,
+      "step": 144050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7592220869393532e-06,
+      "loss": 0.5408,
+      "step": 144060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.755931422554214e-06,
+      "loss": 0.6008,
+      "step": 144070
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7526407581690745e-06,
+      "loss": 0.7202,
+      "step": 144080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.749350093783935e-06,
+      "loss": 0.5749,
+      "step": 144090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7460594293987958e-06,
+      "loss": 0.6153,
+      "step": 144100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.742768765013656e-06,
+      "loss": 0.4983,
+      "step": 144110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.739478100628517e-06,
+      "loss": 0.5459,
+      "step": 144120
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7361874362433775e-06,
+      "loss": 0.6049,
+      "step": 144130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7328967718582383e-06,
+      "loss": 0.5802,
+      "step": 144140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.729606107473099e-06,
+      "loss": 0.6567,
+      "step": 144150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7263154430879596e-06,
+      "loss": 0.5802,
+      "step": 144160
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7230247787028204e-06,
+      "loss": 0.4479,
+      "step": 144170
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.719734114317681e-06,
+      "loss": 0.5794,
+      "step": 144180
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7164434499325412e-06,
+      "loss": 0.7367,
+      "step": 144190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.713152785547402e-06,
+      "loss": 0.5574,
+      "step": 144200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7098621211622625e-06,
+      "loss": 0.6065,
+      "step": 144210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7065714567771234e-06,
+      "loss": 0.6377,
+      "step": 144220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.703280792391984e-06,
+      "loss": 0.5678,
+      "step": 144230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.699990128006845e-06,
+      "loss": 0.6103,
+      "step": 144240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6966994636217055e-06,
+      "loss": 0.527,
+      "step": 144250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.693408799236566e-06,
+      "loss": 0.6717,
+      "step": 144260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6901181348514267e-06,
+      "loss": 0.6264,
+      "step": 144270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.686827470466287e-06,
+      "loss": 0.6022,
+      "step": 144280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.683536806081148e-06,
+      "loss": 0.6313,
+      "step": 144290
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6802461416960084e-06,
+      "loss": 0.5906,
+      "step": 144300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.676955477310869e-06,
+      "loss": 0.5725,
+      "step": 144310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6736648129257297e-06,
+      "loss": 0.8633,
+      "step": 144320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6703741485405905e-06,
+      "loss": 0.6844,
+      "step": 144330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6670834841554514e-06,
+      "loss": 0.5549,
+      "step": 144340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6637928197703118e-06,
+      "loss": 0.5201,
+      "step": 144350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.660502155385172e-06,
+      "loss": 0.6364,
+      "step": 144360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.657211491000033e-06,
+      "loss": 0.4776,
+      "step": 144370
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6539208266148935e-06,
+      "loss": 0.5701,
+      "step": 144380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6506301622297543e-06,
+      "loss": 0.5407,
+      "step": 144390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6473394978446147e-06,
+      "loss": 0.5759,
+      "step": 144400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6440488334594756e-06,
+      "loss": 0.5296,
+      "step": 144410
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6407581690743364e-06,
+      "loss": 0.6009,
+      "step": 144420
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.637467504689197e-06,
+      "loss": 0.5428,
+      "step": 144430
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6341768403040577e-06,
+      "loss": 0.5403,
+      "step": 144440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.630886175918918e-06,
+      "loss": 0.5091,
+      "step": 144450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.627595511533779e-06,
+      "loss": 0.474,
+      "step": 144460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6243048471486394e-06,
+      "loss": 0.5941,
+      "step": 144470
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6210141827635e-06,
+      "loss": 0.5509,
+      "step": 144480
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6177235183783606e-06,
+      "loss": 0.4573,
+      "step": 144490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6144328539932215e-06,
+      "loss": 0.6257,
+      "step": 144500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6111421896080823e-06,
+      "loss": 0.6047,
+      "step": 144510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6078515252229427e-06,
+      "loss": 0.563,
+      "step": 144520
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.604560860837803e-06,
+      "loss": 0.5272,
+      "step": 144530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.601270196452664e-06,
+      "loss": 0.5135,
+      "step": 144540
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.5979795320675244e-06,
+      "loss": 0.6344,
+      "step": 144550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.5946888676823853e-06,
+      "loss": 0.5586,
+      "step": 144560
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5913982032972457e-06,
+      "loss": 0.5666,
+      "step": 144570
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5881075389121065e-06,
+      "loss": 0.6107,
+      "step": 144580
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.584816874526967e-06,
+      "loss": 0.545,
+      "step": 144590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.581526210141828e-06,
+      "loss": 0.569,
+      "step": 144600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5782355457566886e-06,
+      "loss": 0.6746,
+      "step": 144610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.574944881371549e-06,
+      "loss": 0.6034,
+      "step": 144620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.57165421698641e-06,
+      "loss": 0.5392,
+      "step": 144630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5683635526012703e-06,
+      "loss": 0.5299,
+      "step": 144640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5650728882161308e-06,
+      "loss": 0.5833,
+      "step": 144650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5617822238309916e-06,
+      "loss": 0.5364,
+      "step": 144660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.558491559445852e-06,
+      "loss": 0.4939,
+      "step": 144670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.555200895060713e-06,
+      "loss": 0.6786,
+      "step": 144680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5519102306755737e-06,
+      "loss": 0.5866,
+      "step": 144690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.548619566290434e-06,
+      "loss": 0.6721,
+      "step": 144700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.545328901905295e-06,
+      "loss": 0.5627,
+      "step": 144710
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5420382375201554e-06,
+      "loss": 0.6148,
+      "step": 144720
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5387475731350162e-06,
+      "loss": 0.5228,
+      "step": 144730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5354569087498767e-06,
+      "loss": 0.4169,
+      "step": 144740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5321662443647375e-06,
+      "loss": 0.5595,
+      "step": 144750
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.528875579979598e-06,
+      "loss": 0.6257,
+      "step": 144760
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5255849155944583e-06,
+      "loss": 0.5035,
+      "step": 144770
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.522294251209319e-06,
+      "loss": 0.8412,
+      "step": 144780
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.51900358682418e-06,
+      "loss": 0.5209,
+      "step": 144790
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.515712922439041e-06,
+      "loss": 0.5812,
+      "step": 144800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5124222580539013e-06,
+      "loss": 0.5225,
+      "step": 144810
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5091315936687617e-06,
+      "loss": 0.5205,
+      "step": 144820
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5058409292836226e-06,
+      "loss": 0.5912,
+      "step": 144830
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.502550264898483e-06,
+      "loss": 0.4988,
+      "step": 144840
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.499259600513344e-06,
+      "loss": 0.4277,
+      "step": 144850
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4959689361282042e-06,
+      "loss": 0.6275,
+      "step": 144860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.492678271743065e-06,
+      "loss": 0.5537,
+      "step": 144870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.489387607357926e-06,
+      "loss": 0.6622,
+      "step": 144880
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4860969429727863e-06,
+      "loss": 0.7684,
+      "step": 144890
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.482806278587647e-06,
+      "loss": 0.4336,
+      "step": 144900
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4795156142025076e-06,
+      "loss": 0.518,
+      "step": 144910
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4762249498173685e-06,
+      "loss": 0.4605,
+      "step": 144920
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.472934285432229e-06,
+      "loss": 0.5316,
+      "step": 144930
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4696436210470893e-06,
+      "loss": 0.5388,
+      "step": 144940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.46635295666195e-06,
+      "loss": 0.5845,
+      "step": 144950
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4630622922768106e-06,
+      "loss": 0.5258,
+      "step": 144960
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4597716278916714e-06,
+      "loss": 0.552,
+      "step": 144970
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4564809635065323e-06,
+      "loss": 0.6096,
+      "step": 144980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4531902991213927e-06,
+      "loss": 0.6089,
+      "step": 144990
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4498996347362535e-06,
+      "loss": 0.682,
+      "step": 145000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.446608970351114e-06,
+      "loss": 0.5136,
+      "step": 145010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4433183059659748e-06,
+      "loss": 0.5748,
+      "step": 145020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.440027641580835e-06,
+      "loss": 0.5565,
+      "step": 145030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4367369771956956e-06,
+      "loss": 0.4828,
+      "step": 145040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4334463128105565e-06,
+      "loss": 0.6858,
+      "step": 145050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4301556484254173e-06,
+      "loss": 0.6539,
+      "step": 145060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.426864984040278e-06,
+      "loss": 0.5533,
+      "step": 145070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4235743196551386e-06,
+      "loss": 0.582,
+      "step": 145080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4202836552699994e-06,
+      "loss": 0.5617,
+      "step": 145090
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.41699299088486e-06,
+      "loss": 0.5447,
+      "step": 145100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4137023264997203e-06,
+      "loss": 0.5992,
+      "step": 145110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.410411662114581e-06,
+      "loss": 0.3987,
+      "step": 145120
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4071209977294415e-06,
+      "loss": 0.6586,
+      "step": 145130
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4038303333443024e-06,
+      "loss": 0.6062,
+      "step": 145140
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4005396689591628e-06,
+      "loss": 0.6726,
+      "step": 145150
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3972490045740236e-06,
+      "loss": 0.6335,
+      "step": 145160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3939583401888845e-06,
+      "loss": 0.6377,
+      "step": 145170
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.390667675803745e-06,
+      "loss": 0.6447,
+      "step": 145180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3873770114186057e-06,
+      "loss": 0.6079,
+      "step": 145190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.384086347033466e-06,
+      "loss": 0.677,
+      "step": 145200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3807956826483266e-06,
+      "loss": 0.652,
+      "step": 145210
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3775050182631874e-06,
+      "loss": 0.6989,
+      "step": 145220
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.374214353878048e-06,
+      "loss": 0.5277,
+      "step": 145230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3709236894929087e-06,
+      "loss": 0.4734,
+      "step": 145240
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3676330251077695e-06,
+      "loss": 0.6105,
+      "step": 145250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3643423607226304e-06,
+      "loss": 0.6612,
+      "step": 145260
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.361051696337491e-06,
+      "loss": 0.5734,
+      "step": 145270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3577610319523512e-06,
+      "loss": 0.5298,
+      "step": 145280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.354470367567212e-06,
+      "loss": 0.5838,
+      "step": 145290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3511797031820725e-06,
+      "loss": 0.6655,
+      "step": 145300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3478890387969333e-06,
+      "loss": 0.6006,
+      "step": 145310
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3445983744117937e-06,
+      "loss": 0.6387,
+      "step": 145320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.341307710026654e-06,
+      "loss": 0.574,
+      "step": 145330
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.338017045641515e-06,
+      "loss": 0.5858,
+      "step": 145340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.334726381256376e-06,
+      "loss": 0.5763,
+      "step": 145350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3314357168712367e-06,
+      "loss": 0.5731,
+      "step": 145360
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.328145052486097e-06,
+      "loss": 0.531,
+      "step": 145370
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3248543881009575e-06,
+      "loss": 0.529,
+      "step": 145380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3215637237158184e-06,
+      "loss": 0.4494,
+      "step": 145390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.318273059330679e-06,
+      "loss": 0.5845,
+      "step": 145400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3149823949455396e-06,
+      "loss": 0.5594,
+      "step": 145410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3116917305604e-06,
+      "loss": 0.5578,
+      "step": 145420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.308401066175261e-06,
+      "loss": 0.6314,
+      "step": 145430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3051104017901218e-06,
+      "loss": 0.5242,
+      "step": 145440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.301819737404982e-06,
+      "loss": 0.4904,
+      "step": 145450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.298529073019843e-06,
+      "loss": 0.5267,
+      "step": 145460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2952384086347034e-06,
+      "loss": 0.3733,
+      "step": 145470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2919477442495643e-06,
+      "loss": 0.5199,
+      "step": 145480
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2886570798644247e-06,
+      "loss": 0.5952,
+      "step": 145490
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.285366415479285e-06,
+      "loss": 0.57,
+      "step": 145500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.282075751094146e-06,
+      "loss": 0.7222,
+      "step": 145510
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2787850867090064e-06,
+      "loss": 0.5567,
+      "step": 145520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2754944223238672e-06,
+      "loss": 0.612,
+      "step": 145530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.272203757938728e-06,
+      "loss": 0.6113,
+      "step": 145540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2689130935535885e-06,
+      "loss": 0.4096,
+      "step": 145550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2656224291684493e-06,
+      "loss": 0.5724,
+      "step": 145560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2623317647833098e-06,
+      "loss": 0.5371,
+      "step": 145570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2590411003981706e-06,
+      "loss": 0.8997,
+      "step": 145580
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.255750436013031e-06,
+      "loss": 0.5839,
+      "step": 145590
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2524597716278914e-06,
+      "loss": 0.4673,
+      "step": 145600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2491691072427523e-06,
+      "loss": 0.5537,
+      "step": 145610
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.245878442857613e-06,
+      "loss": 0.6357,
+      "step": 145620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.242587778472474e-06,
+      "loss": 0.6512,
+      "step": 145630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2392971140873344e-06,
+      "loss": 0.611,
+      "step": 145640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2360064497021952e-06,
+      "loss": 0.5046,
+      "step": 145650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2327157853170557e-06,
+      "loss": 0.4845,
+      "step": 145660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.229425120931916e-06,
+      "loss": 0.6713,
+      "step": 145670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.226134456546777e-06,
+      "loss": 0.5748,
+      "step": 145680
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2228437921616373e-06,
+      "loss": 0.4814,
+      "step": 145690
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.219553127776498e-06,
+      "loss": 0.5818,
+      "step": 145700
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2162624633913586e-06,
+      "loss": 0.5721,
+      "step": 145710
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2129717990062195e-06,
+      "loss": 0.4719,
+      "step": 145720
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2096811346210803e-06,
+      "loss": 0.5782,
+      "step": 145730
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2063904702359407e-06,
+      "loss": 0.523,
+      "step": 145740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2030998058508016e-06,
+      "loss": 0.6326,
+      "step": 145750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.199809141465662e-06,
+      "loss": 0.5455,
+      "step": 145760
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1965184770805224e-06,
+      "loss": 0.5764,
+      "step": 145770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1932278126953832e-06,
+      "loss": 0.557,
+      "step": 145780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1899371483102437e-06,
+      "loss": 0.5584,
+      "step": 145790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1866464839251045e-06,
+      "loss": 0.6722,
+      "step": 145800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1833558195399654e-06,
+      "loss": 0.6401,
+      "step": 145810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.180065155154826e-06,
+      "loss": 0.5554,
+      "step": 145820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1767744907696866e-06,
+      "loss": 0.623,
+      "step": 145830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.173483826384547e-06,
+      "loss": 0.5311,
+      "step": 145840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.170193161999408e-06,
+      "loss": 0.5305,
+      "step": 145850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1669024976142683e-06,
+      "loss": 0.6764,
+      "step": 145860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.163611833229129e-06,
+      "loss": 0.5213,
+      "step": 145870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1603211688439896e-06,
+      "loss": 0.4768,
+      "step": 145880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.15703050445885e-06,
+      "loss": 0.6005,
+      "step": 145890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.153739840073711e-06,
+      "loss": 0.5874,
+      "step": 145900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1504491756885717e-06,
+      "loss": 0.5217,
+      "step": 145910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1471585113034325e-06,
+      "loss": 0.6034,
+      "step": 145920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.143867846918293e-06,
+      "loss": 0.563,
+      "step": 145930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1405771825331534e-06,
+      "loss": 0.4297,
+      "step": 145940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.137286518148014e-06,
+      "loss": 0.5529,
+      "step": 145950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1339958537628746e-06,
+      "loss": 0.6467,
+      "step": 145960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1307051893777355e-06,
+      "loss": 0.5527,
+      "step": 145970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.127414524992596e-06,
+      "loss": 0.4651,
+      "step": 145980
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1241238606074567e-06,
+      "loss": 0.6381,
+      "step": 145990
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1208331962223176e-06,
+      "loss": 0.5093,
+      "step": 146000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.117542531837178e-06,
+      "loss": 0.4979,
+      "step": 146010
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.114251867452039e-06,
+      "loss": 0.5603,
+      "step": 146020
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1109612030668993e-06,
+      "loss": 0.5375,
+      "step": 146030
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.10767053868176e-06,
+      "loss": 0.5983,
+      "step": 146040
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1043798742966205e-06,
+      "loss": 0.6243,
+      "step": 146050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.101089209911481e-06,
+      "loss": 0.6707,
+      "step": 146060
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.097798545526342e-06,
+      "loss": 0.5419,
+      "step": 146070
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0945078811412026e-06,
+      "loss": 0.4642,
+      "step": 146080
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0912172167560635e-06,
+      "loss": 0.4687,
+      "step": 146090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.087926552370924e-06,
+      "loss": 0.5981,
+      "step": 146100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0846358879857843e-06,
+      "loss": 0.515,
+      "step": 146110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.081345223600645e-06,
+      "loss": 0.5663,
+      "step": 146120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0780545592155056e-06,
+      "loss": 0.59,
+      "step": 146130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0747638948303664e-06,
+      "loss": 0.5942,
+      "step": 146140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.071473230445227e-06,
+      "loss": 0.6174,
+      "step": 146150
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0681825660600877e-06,
+      "loss": 0.6022,
+      "step": 146160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.064891901674948e-06,
+      "loss": 0.7277,
+      "step": 146170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.061601237289809e-06,
+      "loss": 0.6309,
+      "step": 146180
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.05831057290467e-06,
+      "loss": 0.599,
+      "step": 146190
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0550199085195302e-06,
+      "loss": 0.5621,
+      "step": 146200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.051729244134391e-06,
+      "loss": 0.5786,
+      "step": 146210
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0484385797492515e-06,
+      "loss": 0.482,
+      "step": 146220
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.045147915364112e-06,
+      "loss": 0.4216,
+      "step": 146230
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0418572509789728e-06,
+      "loss": 0.6301,
+      "step": 146240
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.038566586593833e-06,
+      "loss": 0.6276,
+      "step": 146250
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.035275922208694e-06,
+      "loss": 0.424,
+      "step": 146260
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.031985257823555e-06,
+      "loss": 0.6106,
+      "step": 146270
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0286945934384153e-06,
+      "loss": 0.5498,
+      "step": 146280
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.025403929053276e-06,
+      "loss": 0.5679,
+      "step": 146290
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0221132646681365e-06,
+      "loss": 0.5631,
+      "step": 146300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0188226002829974e-06,
+      "loss": 0.6495,
+      "step": 146310
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.015531935897858e-06,
+      "loss": 0.657,
+      "step": 146320
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0122412715127187e-06,
+      "loss": 0.6639,
+      "step": 146330
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.008950607127579e-06,
+      "loss": 0.5392,
+      "step": 146340
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0056599427424395e-06,
+      "loss": 0.5145,
+      "step": 146350
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0023692783573003e-06,
+      "loss": 0.6751,
+      "step": 146360
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.999078613972161e-06,
+      "loss": 0.6238,
+      "step": 146370
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.995787949587022e-06,
+      "loss": 0.4571,
+      "step": 146380
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9924972852018824e-06,
+      "loss": 0.5258,
+      "step": 146390
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.989206620816743e-06,
+      "loss": 0.6964,
+      "step": 146400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9859159564316037e-06,
+      "loss": 0.603,
+      "step": 146410
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.982625292046464e-06,
+      "loss": 0.5895,
+      "step": 146420
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.979334627661325e-06,
+      "loss": 0.5774,
+      "step": 146430
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9760439632761854e-06,
+      "loss": 0.5097,
+      "step": 146440
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9727532988910462e-06,
+      "loss": 0.4959,
+      "step": 146450
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.969462634505907e-06,
+      "loss": 0.6576,
+      "step": 146460
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9661719701207675e-06,
+      "loss": 0.5958,
+      "step": 146470
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9628813057356283e-06,
+      "loss": 0.5506,
+      "step": 146480
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9595906413504888e-06,
+      "loss": 0.4801,
+      "step": 146490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9562999769653496e-06,
+      "loss": 0.5829,
+      "step": 146500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.95300931258021e-06,
+      "loss": 0.6183,
+      "step": 146510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9497186481950705e-06,
+      "loss": 0.5378,
+      "step": 146520
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9464279838099313e-06,
+      "loss": 0.5598,
+      "step": 146530
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9431373194247917e-06,
+      "loss": 0.6475,
+      "step": 146540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9398466550396526e-06,
+      "loss": 0.5922,
+      "step": 146550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9365559906545134e-06,
+      "loss": 0.617,
+      "step": 146560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.933265326269374e-06,
+      "loss": 0.6192,
+      "step": 146570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9299746618842347e-06,
+      "loss": 0.5977,
+      "step": 146580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.926683997499095e-06,
+      "loss": 0.5915,
+      "step": 146590
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.923393333113956e-06,
+      "loss": 0.5217,
+      "step": 146600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9201026687288164e-06,
+      "loss": 0.502,
+      "step": 146610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9168120043436768e-06,
+      "loss": 0.6035,
+      "step": 146620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9135213399585376e-06,
+      "loss": 0.6225,
+      "step": 146630
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9102306755733985e-06,
+      "loss": 0.5478,
+      "step": 146640
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.906940011188259e-06,
+      "loss": 0.5809,
+      "step": 146650
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9036493468031195e-06,
+      "loss": 0.5549,
+      "step": 146660
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9003586824179801e-06,
+      "loss": 0.5325,
+      "step": 146670
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.897068018032841e-06,
+      "loss": 0.4494,
+      "step": 146680
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8937773536477014e-06,
+      "loss": 0.5975,
+      "step": 146690
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8904866892625623e-06,
+      "loss": 0.4796,
+      "step": 146700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8871960248774229e-06,
+      "loss": 0.6221,
+      "step": 146710
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8839053604922837e-06,
+      "loss": 0.6098,
+      "step": 146720
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8806146961071442e-06,
+      "loss": 0.7129,
+      "step": 146730
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8773240317220046e-06,
+      "loss": 0.5886,
+      "step": 146740
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8740333673368654e-06,
+      "loss": 0.5643,
+      "step": 146750
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.870742702951726e-06,
+      "loss": 0.6925,
+      "step": 146760
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8674520385665869e-06,
+      "loss": 0.5779,
+      "step": 146770
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8641613741814473e-06,
+      "loss": 0.4508,
+      "step": 146780
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8608707097963077e-06,
+      "loss": 0.5434,
+      "step": 146790
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8575800454111686e-06,
+      "loss": 0.5951,
+      "step": 146800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8542893810260292e-06,
+      "loss": 0.5621,
+      "step": 146810
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.85099871664089e-06,
+      "loss": 0.9192,
+      "step": 146820
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8477080522557505e-06,
+      "loss": 0.5082,
+      "step": 146830
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.844417387870611e-06,
+      "loss": 0.5452,
+      "step": 146840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8411267234854717e-06,
+      "loss": 0.619,
+      "step": 146850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8378360591003324e-06,
+      "loss": 0.541,
+      "step": 146860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8345453947151932e-06,
+      "loss": 0.5671,
+      "step": 146870
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8312547303300536e-06,
+      "loss": 0.6209,
+      "step": 146880
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8279640659449145e-06,
+      "loss": 0.6033,
+      "step": 146890
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8246734015597751e-06,
+      "loss": 0.6058,
+      "step": 146900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8213827371746355e-06,
+      "loss": 0.4861,
+      "step": 146910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8180920727894964e-06,
+      "loss": 0.6132,
+      "step": 146920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8148014084043568e-06,
+      "loss": 0.4993,
+      "step": 146930
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8115107440192176e-06,
+      "loss": 0.4598,
+      "step": 146940
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8082200796340783e-06,
+      "loss": 0.685,
+      "step": 146950
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8049294152489387e-06,
+      "loss": 0.4595,
+      "step": 146960
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8016387508637995e-06,
+      "loss": 0.4766,
+      "step": 146970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.79834808647866e-06,
+      "loss": 0.5416,
+      "step": 146980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7950574220935208e-06,
+      "loss": 0.5935,
+      "step": 146990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7917667577083814e-06,
+      "loss": 0.709,
+      "step": 147000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7884760933232419e-06,
+      "loss": 0.5591,
+      "step": 147010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7851854289381027e-06,
+      "loss": 0.614,
+      "step": 147020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7818947645529633e-06,
+      "loss": 0.5516,
+      "step": 147030
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7786041001678242e-06,
+      "loss": 0.5406,
+      "step": 147040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7753134357826846e-06,
+      "loss": 0.6288,
+      "step": 147050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7720227713975454e-06,
+      "loss": 0.6545,
+      "step": 147060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7687321070124059e-06,
+      "loss": 0.586,
+      "step": 147070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7654414426272665e-06,
+      "loss": 0.6075,
+      "step": 147080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7621507782421273e-06,
+      "loss": 0.7205,
+      "step": 147090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7588601138569878e-06,
+      "loss": 0.5014,
+      "step": 147100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7555694494718486e-06,
+      "loss": 0.6083,
+      "step": 147110
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.752278785086709e-06,
+      "loss": 0.6447,
+      "step": 147120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7489881207015697e-06,
+      "loss": 0.5325,
+      "step": 147130
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7456974563164305e-06,
+      "loss": 0.663,
+      "step": 147140
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.742406791931291e-06,
+      "loss": 0.6864,
+      "step": 147150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7391161275461518e-06,
+      "loss": 0.5566,
+      "step": 147160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7358254631610122e-06,
+      "loss": 0.5293,
+      "step": 147170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7325347987758728e-06,
+      "loss": 0.7121,
+      "step": 147180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7292441343907337e-06,
+      "loss": 0.5363,
+      "step": 147190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.725953470005594e-06,
+      "loss": 0.6287,
+      "step": 147200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.722662805620455e-06,
+      "loss": 0.5874,
+      "step": 147210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7193721412353156e-06,
+      "loss": 0.637,
+      "step": 147220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7160814768501764e-06,
+      "loss": 0.5522,
+      "step": 147230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7127908124650368e-06,
+      "loss": 0.4617,
+      "step": 147240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7095001480798972e-06,
+      "loss": 0.5596,
+      "step": 147250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.706209483694758e-06,
+      "loss": 0.5728,
+      "step": 147260
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7029188193096187e-06,
+      "loss": 0.4829,
+      "step": 147270
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6996281549244796e-06,
+      "loss": 0.6292,
+      "step": 147280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.69633749053934e-06,
+      "loss": 0.72,
+      "step": 147290
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6930468261542004e-06,
+      "loss": 0.5564,
+      "step": 147300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6897561617690612e-06,
+      "loss": 0.5447,
+      "step": 147310
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6864654973839219e-06,
+      "loss": 0.5592,
+      "step": 147320
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6831748329987827e-06,
+      "loss": 0.6397,
+      "step": 147330
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6798841686136431e-06,
+      "loss": 0.4772,
+      "step": 147340
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6765935042285036e-06,
+      "loss": 0.5832,
+      "step": 147350
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6733028398433644e-06,
+      "loss": 0.6269,
+      "step": 147360
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.670012175458225e-06,
+      "loss": 0.5717,
+      "step": 147370
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6667215110730859e-06,
+      "loss": 0.4976,
+      "step": 147380
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6634308466879463e-06,
+      "loss": 0.6383,
+      "step": 147390
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6601401823028071e-06,
+      "loss": 0.4653,
+      "step": 147400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6568495179176678e-06,
+      "loss": 0.5977,
+      "step": 147410
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6535588535325282e-06,
+      "loss": 0.5844,
+      "step": 147420
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.650268189147389e-06,
+      "loss": 0.5754,
+      "step": 147430
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6469775247622495e-06,
+      "loss": 0.5866,
+      "step": 147440
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6436868603771103e-06,
+      "loss": 0.5254,
+      "step": 147450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.640396195991971e-06,
+      "loss": 0.5971,
+      "step": 147460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6371055316068314e-06,
+      "loss": 0.495,
+      "step": 147470
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6338148672216922e-06,
+      "loss": 0.627,
+      "step": 147480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6305242028365526e-06,
+      "loss": 0.4781,
+      "step": 147490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6272335384514135e-06,
+      "loss": 0.4938,
+      "step": 147500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.623942874066274e-06,
+      "loss": 0.7192,
+      "step": 147510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6206522096811345e-06,
+      "loss": 0.6526,
+      "step": 147520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6173615452959954e-06,
+      "loss": 0.6378,
+      "step": 147530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6140708809108558e-06,
+      "loss": 0.5879,
+      "step": 147540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6107802165257166e-06,
+      "loss": 0.4697,
+      "step": 147550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6074895521405773e-06,
+      "loss": 0.5697,
+      "step": 147560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.604198887755438e-06,
+      "loss": 0.373,
+      "step": 147570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6009082233702985e-06,
+      "loss": 0.5314,
+      "step": 147580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5976175589851592e-06,
+      "loss": 0.701,
+      "step": 147590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.59432689460002e-06,
+      "loss": 0.6569,
+      "step": 147600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5910362302148804e-06,
+      "loss": 0.5359,
+      "step": 147610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5877455658297413e-06,
+      "loss": 0.5546,
+      "step": 147620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5844549014446017e-06,
+      "loss": 0.636,
+      "step": 147630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5811642370594623e-06,
+      "loss": 0.5295,
+      "step": 147640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5778735726743232e-06,
+      "loss": 0.5761,
+      "step": 147650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5745829082891836e-06,
+      "loss": 0.5165,
+      "step": 147660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5712922439040444e-06,
+      "loss": 0.5294,
+      "step": 147670
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5680015795189048e-06,
+      "loss": 0.5799,
+      "step": 147680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5647109151337655e-06,
+      "loss": 0.5721,
+      "step": 147690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5614202507486263e-06,
+      "loss": 0.5705,
+      "step": 147700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.558129586363487e-06,
+      "loss": 0.6284,
+      "step": 147710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5548389219783474e-06,
+      "loss": 0.5885,
+      "step": 147720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.551548257593208e-06,
+      "loss": 0.6544,
+      "step": 147730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5482575932080689e-06,
+      "loss": 0.5989,
+      "step": 147740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5449669288229295e-06,
+      "loss": 0.5439,
+      "step": 147750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5416762644377901e-06,
+      "loss": 0.6111,
+      "step": 147760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5383856000526505e-06,
+      "loss": 0.6325,
+      "step": 147770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5350949356675114e-06,
+      "loss": 0.5269,
+      "step": 147780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.531804271282372e-06,
+      "loss": 0.5367,
+      "step": 147790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5285136068972326e-06,
+      "loss": 0.6213,
+      "step": 147800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5252229425120933e-06,
+      "loss": 0.4946,
+      "step": 147810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.521932278126954e-06,
+      "loss": 0.6076,
+      "step": 147820
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5186416137418145e-06,
+      "loss": 0.5497,
+      "step": 147830
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5153509493566752e-06,
+      "loss": 0.6152,
+      "step": 147840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5120602849715358e-06,
+      "loss": 0.547,
+      "step": 147850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5087696205863964e-06,
+      "loss": 0.5065,
+      "step": 147860
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.505478956201257e-06,
+      "loss": 0.4226,
+      "step": 147870
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.502188291816118e-06,
+      "loss": 0.6427,
+      "step": 147880
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4988976274309783e-06,
+      "loss": 0.6053,
+      "step": 147890
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.495606963045839e-06,
+      "loss": 0.5534,
+      "step": 147900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4923162986606996e-06,
+      "loss": 0.5476,
+      "step": 147910
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4890256342755604e-06,
+      "loss": 0.5656,
+      "step": 147920
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.485734969890421e-06,
+      "loss": 0.5378,
+      "step": 147930
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4824443055052815e-06,
+      "loss": 0.4992,
+      "step": 147940
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4791536411201421e-06,
+      "loss": 0.7874,
+      "step": 147950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4758629767350028e-06,
+      "loss": 0.6109,
+      "step": 147960
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4725723123498636e-06,
+      "loss": 0.5913,
+      "step": 147970
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4692816479647242e-06,
+      "loss": 0.5564,
+      "step": 147980
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4659909835795849e-06,
+      "loss": 0.5801,
+      "step": 147990
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4627003191944453e-06,
+      "loss": 0.5201,
+      "step": 148000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4594096548093061e-06,
+      "loss": 0.5217,
+      "step": 148010
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4561189904241668e-06,
+      "loss": 0.6089,
+      "step": 148020
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4528283260390274e-06,
+      "loss": 0.5823,
+      "step": 148030
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.449537661653888e-06,
+      "loss": 0.5757,
+      "step": 148040
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4462469972687484e-06,
+      "loss": 0.4599,
+      "step": 148050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4429563328836093e-06,
+      "loss": 0.5359,
+      "step": 148060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.43966566849847e-06,
+      "loss": 0.4096,
+      "step": 148070
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4363750041133306e-06,
+      "loss": 0.7003,
+      "step": 148080
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4330843397281912e-06,
+      "loss": 0.6226,
+      "step": 148090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4297936753430518e-06,
+      "loss": 0.6076,
+      "step": 148100
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4265030109579125e-06,
+      "loss": 0.6479,
+      "step": 148110
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.423212346572773e-06,
+      "loss": 0.7011,
+      "step": 148120
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4199216821876337e-06,
+      "loss": 0.6389,
+      "step": 148130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4166310178024944e-06,
+      "loss": 0.481,
+      "step": 148140
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.413340353417355e-06,
+      "loss": 0.4358,
+      "step": 148150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4100496890322158e-06,
+      "loss": 0.549,
+      "step": 148160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4067590246470762e-06,
+      "loss": 0.667,
+      "step": 148170
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4034683602619369e-06,
+      "loss": 0.5457,
+      "step": 148180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4001776958767975e-06,
+      "loss": 0.6163,
+      "step": 148190
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3968870314916584e-06,
+      "loss": 0.5097,
+      "step": 148200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.393596367106519e-06,
+      "loss": 0.5111,
+      "step": 148210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3903057027213794e-06,
+      "loss": 0.5405,
+      "step": 148220
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.38701503833624e-06,
+      "loss": 0.6127,
+      "step": 148230
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3837243739511007e-06,
+      "loss": 0.549,
+      "step": 148240
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3804337095659615e-06,
+      "loss": 0.4812,
+      "step": 148250
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3771430451808221e-06,
+      "loss": 0.6976,
+      "step": 148260
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3738523807956828e-06,
+      "loss": 0.5044,
+      "step": 148270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3705617164105432e-06,
+      "loss": 0.6502,
+      "step": 148280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.367271052025404e-06,
+      "loss": 0.5445,
+      "step": 148290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3639803876402647e-06,
+      "loss": 0.5898,
+      "step": 148300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3606897232551253e-06,
+      "loss": 0.5691,
+      "step": 148310
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.357399058869986e-06,
+      "loss": 0.6327,
+      "step": 148320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3541083944848466e-06,
+      "loss": 0.7073,
+      "step": 148330
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3508177300997072e-06,
+      "loss": 0.58,
+      "step": 148340
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3475270657145678e-06,
+      "loss": 0.5238,
+      "step": 148350
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3442364013294285e-06,
+      "loss": 0.5524,
+      "step": 148360
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.340945736944289e-06,
+      "loss": 0.4866,
+      "step": 148370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3376550725591497e-06,
+      "loss": 0.5845,
+      "step": 148380
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3343644081740104e-06,
+      "loss": 0.4973,
+      "step": 148390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.331073743788871e-06,
+      "loss": 0.5248,
+      "step": 148400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3277830794037316e-06,
+      "loss": 0.5901,
+      "step": 148410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3244924150185923e-06,
+      "loss": 0.5423,
+      "step": 148420
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.321201750633453e-06,
+      "loss": 0.6391,
+      "step": 148430
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3179110862483137e-06,
+      "loss": 0.5283,
+      "step": 148440
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3146204218631742e-06,
+      "loss": 0.6943,
+      "step": 148450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3113297574780348e-06,
+      "loss": 0.5413,
+      "step": 148460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3080390930928954e-06,
+      "loss": 0.6926,
+      "step": 148470
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3047484287077563e-06,
+      "loss": 0.5461,
+      "step": 148480
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.301457764322617e-06,
+      "loss": 0.5772,
+      "step": 148490
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2981670999374775e-06,
+      "loss": 0.5689,
+      "step": 148500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.294876435552338e-06,
+      "loss": 0.6549,
+      "step": 148510
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2915857711671988e-06,
+      "loss": 0.5488,
+      "step": 148520
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2882951067820594e-06,
+      "loss": 0.5405,
+      "step": 148530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.28500444239692e-06,
+      "loss": 0.6144,
+      "step": 148540
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2817137780117807e-06,
+      "loss": 0.6945,
+      "step": 148550
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2784231136266411e-06,
+      "loss": 0.6073,
+      "step": 148560
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.275132449241502e-06,
+      "loss": 0.6051,
+      "step": 148570
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2718417848563626e-06,
+      "loss": 0.5744,
+      "step": 148580
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2685511204712232e-06,
+      "loss": 0.5718,
+      "step": 148590
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2652604560860839e-06,
+      "loss": 0.5436,
+      "step": 148600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2619697917009445e-06,
+      "loss": 0.515,
+      "step": 148610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2586791273158051e-06,
+      "loss": 0.4725,
+      "step": 148620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2553884629306658e-06,
+      "loss": 0.6218,
+      "step": 148630
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2520977985455264e-06,
+      "loss": 0.5201,
+      "step": 148640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.248807134160387e-06,
+      "loss": 0.5323,
+      "step": 148650
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2455164697752476e-06,
+      "loss": 0.6065,
+      "step": 148660
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2422258053901083e-06,
+      "loss": 0.5665,
+      "step": 148670
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.238935141004969e-06,
+      "loss": 0.6615,
+      "step": 148680
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2356444766198295e-06,
+      "loss": 0.6014,
+      "step": 148690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2323538122346902e-06,
+      "loss": 0.584,
+      "step": 148700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.229063147849551e-06,
+      "loss": 0.5796,
+      "step": 148710
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2257724834644117e-06,
+      "loss": 0.5923,
+      "step": 148720
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.222481819079272e-06,
+      "loss": 0.5915,
+      "step": 148730
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2191911546941327e-06,
+      "loss": 0.5489,
+      "step": 148740
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2159004903089933e-06,
+      "loss": 0.5379,
+      "step": 148750
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2126098259238542e-06,
+      "loss": 0.5382,
+      "step": 148760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2093191615387148e-06,
+      "loss": 0.5648,
+      "step": 148770
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2060284971535754e-06,
+      "loss": 0.6748,
+      "step": 148780
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2027378327684359e-06,
+      "loss": 0.602,
+      "step": 148790
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1994471683832967e-06,
+      "loss": 0.5202,
+      "step": 148800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1961565039981573e-06,
+      "loss": 0.5384,
+      "step": 148810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.192865839613018e-06,
+      "loss": 0.5726,
+      "step": 148820
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1895751752278786e-06,
+      "loss": 0.6222,
+      "step": 148830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.186284510842739e-06,
+      "loss": 0.599,
+      "step": 148840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1829938464575999e-06,
+      "loss": 0.5601,
+      "step": 148850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1797031820724605e-06,
+      "loss": 0.5255,
+      "step": 148860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1764125176873211e-06,
+      "loss": 0.4708,
+      "step": 148870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1731218533021818e-06,
+      "loss": 0.4551,
+      "step": 148880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1698311889170424e-06,
+      "loss": 0.5826,
+      "step": 148890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.166540524531903e-06,
+      "loss": 0.6606,
+      "step": 148900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1632498601467637e-06,
+      "loss": 0.5541,
+      "step": 148910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1599591957616243e-06,
+      "loss": 0.572,
+      "step": 148920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.156668531376485e-06,
+      "loss": 0.62,
+      "step": 148930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1533778669913456e-06,
+      "loss": 0.6272,
+      "step": 148940
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1500872026062064e-06,
+      "loss": 0.6111,
+      "step": 148950
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1467965382210668e-06,
+      "loss": 0.5895,
+      "step": 148960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1435058738359275e-06,
+      "loss": 0.6131,
+      "step": 148970
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.140215209450788e-06,
+      "loss": 0.5429,
+      "step": 148980
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.136924545065649e-06,
+      "loss": 0.5578,
+      "step": 148990
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1336338806805096e-06,
+      "loss": 0.5452,
+      "step": 149000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.13034321629537e-06,
+      "loss": 0.5255,
+      "step": 149010
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1270525519102306e-06,
+      "loss": 0.6116,
+      "step": 149020
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1237618875250913e-06,
+      "loss": 0.5016,
+      "step": 149030
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.120471223139952e-06,
+      "loss": 0.4928,
+      "step": 149040
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1171805587548127e-06,
+      "loss": 0.5037,
+      "step": 149050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1138898943696734e-06,
+      "loss": 0.6002,
+      "step": 149060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1105992299845338e-06,
+      "loss": 0.572,
+      "step": 149070
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1073085655993946e-06,
+      "loss": 0.5498,
+      "step": 149080
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1040179012142553e-06,
+      "loss": 0.544,
+      "step": 149090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1007272368291159e-06,
+      "loss": 0.5275,
+      "step": 149100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0974365724439765e-06,
+      "loss": 0.6012,
+      "step": 149110
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0941459080588372e-06,
+      "loss": 0.5223,
+      "step": 149120
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0908552436736978e-06,
+      "loss": 0.5769,
+      "step": 149130
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0875645792885584e-06,
+      "loss": 0.5237,
+      "step": 149140
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.084273914903419e-06,
+      "loss": 0.5316,
+      "step": 149150
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0809832505182797e-06,
+      "loss": 0.5791,
+      "step": 149160
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0776925861331403e-06,
+      "loss": 0.5258,
+      "step": 149170
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.074401921748001e-06,
+      "loss": 0.606,
+      "step": 149180
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0711112573628616e-06,
+      "loss": 0.5605,
+      "step": 149190
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0678205929777222e-06,
+      "loss": 0.569,
+      "step": 149200
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0645299285925828e-06,
+      "loss": 0.5327,
+      "step": 149210
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0612392642074437e-06,
+      "loss": 0.4801,
+      "step": 149220
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0579485998223043e-06,
+      "loss": 0.6848,
+      "step": 149230
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0546579354371647e-06,
+      "loss": 0.5711,
+      "step": 149240
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0513672710520254e-06,
+      "loss": 0.5475,
+      "step": 149250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.048076606666886e-06,
+      "loss": 0.5356,
+      "step": 149260
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0447859422817468e-06,
+      "loss": 0.6385,
+      "step": 149270
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0414952778966075e-06,
+      "loss": 0.6147,
+      "step": 149280
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.038204613511468e-06,
+      "loss": 0.5506,
+      "step": 149290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0349139491263285e-06,
+      "loss": 0.6531,
+      "step": 149300
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0316232847411894e-06,
+      "loss": 0.6152,
+      "step": 149310
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.02833262035605e-06,
+      "loss": 0.6694,
+      "step": 149320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0250419559709106e-06,
+      "loss": 0.5678,
+      "step": 149330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0217512915857713e-06,
+      "loss": 0.5407,
+      "step": 149340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0184606272006317e-06,
+      "loss": 0.6246,
+      "step": 149350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0151699628154925e-06,
+      "loss": 0.4904,
+      "step": 149360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0118792984303532e-06,
+      "loss": 0.5227,
+      "step": 149370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0085886340452138e-06,
+      "loss": 0.6213,
+      "step": 149380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0052979696600744e-06,
+      "loss": 0.5981,
+      "step": 149390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.002007305274935e-06,
+      "loss": 0.6683,
+      "step": 149400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.987166408897957e-07,
+      "loss": 0.5575,
+      "step": 149410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.954259765046563e-07,
+      "loss": 0.4595,
+      "step": 149420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.92135312119517e-07,
+      "loss": 0.4383,
+      "step": 149430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.888446477343776e-07,
+      "loss": 0.6713,
+      "step": 149440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.855539833492382e-07,
+      "loss": 0.5353,
+      "step": 149450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.822633189640989e-07,
+      "loss": 0.545,
+      "step": 149460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.789726545789595e-07,
+      "loss": 0.6437,
+      "step": 149470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.756819901938201e-07,
+      "loss": 0.4789,
+      "step": 149480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.723913258086808e-07,
+      "loss": 0.4882,
+      "step": 149490
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.691006614235416e-07,
+      "loss": 0.5074,
+      "step": 149500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.658099970384022e-07,
+      "loss": 0.5393,
+      "step": 149510
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.625193326532627e-07,
+      "loss": 0.579,
+      "step": 149520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.592286682681233e-07,
+      "loss": 0.4817,
+      "step": 149530
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.55938003882984e-07,
+      "loss": 0.5503,
+      "step": 149540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.526473394978447e-07,
+      "loss": 0.539,
+      "step": 149550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.493566751127054e-07,
+      "loss": 0.6182,
+      "step": 149560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.46066010727566e-07,
+      "loss": 0.552,
+      "step": 149570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.427753463424265e-07,
+      "loss": 0.6415,
+      "step": 149580
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.394846819572872e-07,
+      "loss": 0.5928,
+      "step": 149590
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.361940175721478e-07,
+      "loss": 0.5551,
+      "step": 149600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.329033531870086e-07,
+      "loss": 0.495,
+      "step": 149610
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.296126888018692e-07,
+      "loss": 0.616,
+      "step": 149620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.263220244167297e-07,
+      "loss": 0.6423,
+      "step": 149630
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.230313600315903e-07,
+      "loss": 0.4938,
+      "step": 149640
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.197406956464511e-07,
+      "loss": 0.4872,
+      "step": 149650
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.164500312613117e-07,
+      "loss": 0.7321,
+      "step": 149660
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.131593668761723e-07,
+      "loss": 0.6155,
+      "step": 149670
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.098687024910331e-07,
+      "loss": 0.7374,
+      "step": 149680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.065780381058935e-07,
+      "loss": 0.4741,
+      "step": 149690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.032873737207542e-07,
+      "loss": 0.6349,
+      "step": 149700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.999967093356149e-07,
+      "loss": 0.4814,
+      "step": 149710
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.967060449504756e-07,
+      "loss": 0.5706,
+      "step": 149720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.934153805653362e-07,
+      "loss": 0.52,
+      "step": 149730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.901247161801969e-07,
+      "loss": 0.542,
+      "step": 149740
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.868340517950574e-07,
+      "loss": 0.5573,
+      "step": 149750
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.83543387409918e-07,
+      "loss": 0.5385,
+      "step": 149760
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.802527230247788e-07,
+      "loss": 0.6449,
+      "step": 149770
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.769620586396394e-07,
+      "loss": 0.5992,
+      "step": 149780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.736713942545001e-07,
+      "loss": 0.6084,
+      "step": 149790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.703807298693606e-07,
+      "loss": 0.6369,
+      "step": 149800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.670900654842213e-07,
+      "loss": 0.5851,
+      "step": 149810
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.637994010990819e-07,
+      "loss": 0.6092,
+      "step": 149820
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.605087367139426e-07,
+      "loss": 0.6248,
+      "step": 149830
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.572180723288033e-07,
+      "loss": 0.5105,
+      "step": 149840
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.539274079436639e-07,
+      "loss": 0.6371,
+      "step": 149850
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.506367435585245e-07,
+      "loss": 0.7695,
+      "step": 149860
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.473460791733851e-07,
+      "loss": 0.4891,
+      "step": 149870
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.440554147882458e-07,
+      "loss": 0.6134,
+      "step": 149880
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.407647504031065e-07,
+      "loss": 0.5535,
+      "step": 149890
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.374740860179671e-07,
+      "loss": 0.6017,
+      "step": 149900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.341834216328276e-07,
+      "loss": 0.5927,
+      "step": 149910
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.308927572476883e-07,
+      "loss": 0.6818,
+      "step": 149920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.27602092862549e-07,
+      "loss": 0.5857,
+      "step": 149930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.243114284774096e-07,
+      "loss": 0.7294,
+      "step": 149940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.210207640922703e-07,
+      "loss": 0.5849,
+      "step": 149950
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.17730099707131e-07,
+      "loss": 0.5676,
+      "step": 149960
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.144394353219915e-07,
+      "loss": 0.5209,
+      "step": 149970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.111487709368522e-07,
+      "loss": 0.6042,
+      "step": 149980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.078581065517128e-07,
+      "loss": 0.6111,
+      "step": 149990
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.045674421665735e-07,
+      "loss": 0.7412,
+      "step": 150000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.012767777814342e-07,
+      "loss": 0.5668,
+      "step": 150010
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.979861133962948e-07,
+      "loss": 0.6033,
+      "step": 150020
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.946954490111553e-07,
+      "loss": 0.4964,
+      "step": 150030
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.91404784626016e-07,
+      "loss": 0.6074,
+      "step": 150040
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.881141202408767e-07,
+      "loss": 0.6034,
+      "step": 150050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.848234558557373e-07,
+      "loss": 0.5018,
+      "step": 150060
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.815327914705981e-07,
+      "loss": 0.4198,
+      "step": 150070
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.782421270854586e-07,
+      "loss": 0.6118,
+      "step": 150080
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.749514627003192e-07,
+      "loss": 0.4375,
+      "step": 150090
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.716607983151798e-07,
+      "loss": 0.5641,
+      "step": 150100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.683701339300405e-07,
+      "loss": 0.615,
+      "step": 150110
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.650794695449012e-07,
+      "loss": 0.5954,
+      "step": 150120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.617888051597617e-07,
+      "loss": 0.6003,
+      "step": 150130
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.584981407746225e-07,
+      "loss": 0.6584,
+      "step": 150140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.55207476389483e-07,
+      "loss": 0.6314,
+      "step": 150150
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.519168120043437e-07,
+      "loss": 0.4829,
+      "step": 150160
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.486261476192044e-07,
+      "loss": 0.5521,
+      "step": 150170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.45335483234065e-07,
+      "loss": 0.6367,
+      "step": 150180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.420448188489256e-07,
+      "loss": 0.6611,
+      "step": 150190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.387541544637863e-07,
+      "loss": 0.5106,
+      "step": 150200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.354634900786469e-07,
+      "loss": 0.5716,
+      "step": 150210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.321728256935075e-07,
+      "loss": 0.5327,
+      "step": 150220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.288821613083683e-07,
+      "loss": 0.4531,
+      "step": 150230
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.255914969232288e-07,
+      "loss": 0.5683,
+      "step": 150240
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.223008325380894e-07,
+      "loss": 0.6633,
+      "step": 150250
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.190101681529502e-07,
+      "loss": 0.5243,
+      "step": 150260
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.157195037678107e-07,
+      "loss": 0.5842,
+      "step": 150270
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.124288393826714e-07,
+      "loss": 0.5144,
+      "step": 150280
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.09138174997532e-07,
+      "loss": 0.5524,
+      "step": 150290
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.058475106123927e-07,
+      "loss": 0.5506,
+      "step": 150300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.025568462272533e-07,
+      "loss": 0.513,
+      "step": 150310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.99266181842114e-07,
+      "loss": 0.5573,
+      "step": 150320
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.959755174569746e-07,
+      "loss": 0.5388,
+      "step": 150330
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.926848530718352e-07,
+      "loss": 0.5839,
+      "step": 150340
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.893941886866959e-07,
+      "loss": 0.6353,
+      "step": 150350
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.861035243015565e-07,
+      "loss": 0.6518,
+      "step": 150360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.828128599164172e-07,
+      "loss": 0.5762,
+      "step": 150370
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.795221955312778e-07,
+      "loss": 0.4997,
+      "step": 150380
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.762315311461384e-07,
+      "loss": 0.6062,
+      "step": 150390
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.729408667609991e-07,
+      "loss": 0.595,
+      "step": 150400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.696502023758597e-07,
+      "loss": 0.6442,
+      "step": 150410
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.663595379907204e-07,
+      "loss": 0.4713,
+      "step": 150420
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.63068873605581e-07,
+      "loss": 0.6323,
+      "step": 150430
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.597782092204417e-07,
+      "loss": 0.6272,
+      "step": 150440
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.564875448353023e-07,
+      "loss": 0.5863,
+      "step": 150450
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.531968804501629e-07,
+      "loss": 0.5766,
+      "step": 150460
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.499062160650236e-07,
+      "loss": 0.4914,
+      "step": 150470
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.466155516798842e-07,
+      "loss": 0.4393,
+      "step": 150480
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.433248872947448e-07,
+      "loss": 0.5715,
+      "step": 150490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.400342229096055e-07,
+      "loss": 0.5787,
+      "step": 150500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.367435585244662e-07,
+      "loss": 0.5636,
+      "step": 150510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.334528941393267e-07,
+      "loss": 0.6181,
+      "step": 150520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.301622297541875e-07,
+      "loss": 0.5451,
+      "step": 150530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.268715653690481e-07,
+      "loss": 0.6161,
+      "step": 150540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.235809009839086e-07,
+      "loss": 0.6054,
+      "step": 150550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.202902365987694e-07,
+      "loss": 0.6032,
+      "step": 150560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.1699957221363e-07,
+      "loss": 0.8337,
+      "step": 150570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.137089078284906e-07,
+      "loss": 0.5537,
+      "step": 150580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.104182434433512e-07,
+      "loss": 0.5107,
+      "step": 150590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.071275790582119e-07,
+      "loss": 0.6035,
+      "step": 150600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.038369146730725e-07,
+      "loss": 0.5273,
+      "step": 150610
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.005462502879331e-07,
+      "loss": 0.5386,
+      "step": 150620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.972555859027938e-07,
+      "loss": 0.6028,
+      "step": 150630
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.939649215176544e-07,
+      "loss": 0.592,
+      "step": 150640
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.906742571325151e-07,
+      "loss": 0.6291,
+      "step": 150650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.873835927473757e-07,
+      "loss": 0.4614,
+      "step": 150660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.840929283622364e-07,
+      "loss": 0.6315,
+      "step": 150670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.80802263977097e-07,
+      "loss": 0.5249,
+      "step": 150680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.775115995919576e-07,
+      "loss": 0.5927,
+      "step": 150690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.742209352068183e-07,
+      "loss": 0.5743,
+      "step": 150700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.709302708216789e-07,
+      "loss": 0.5785,
+      "step": 150710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.676396064365396e-07,
+      "loss": 0.5026,
+      "step": 150720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643489420514002e-07,
+      "loss": 0.6256,
+      "step": 150730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.610582776662608e-07,
+      "loss": 0.5956,
+      "step": 150740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.577676132811215e-07,
+      "loss": 0.5187,
+      "step": 150750
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.544769488959821e-07,
+      "loss": 0.5393,
+      "step": 150760
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.511862845108427e-07,
+      "loss": 0.5125,
+      "step": 150770
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.478956201257034e-07,
+      "loss": 0.6491,
+      "step": 150780
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.446049557405641e-07,
+      "loss": 0.6397,
+      "step": 150790
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.413142913554246e-07,
+      "loss": 0.606,
+      "step": 150800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.380236269702854e-07,
+      "loss": 0.519,
+      "step": 150810
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.34732962585146e-07,
+      "loss": 0.6676,
+      "step": 150820
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.314422982000066e-07,
+      "loss": 0.5867,
+      "step": 150830
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.281516338148673e-07,
+      "loss": 0.5579,
+      "step": 150840
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.248609694297279e-07,
+      "loss": 0.5277,
+      "step": 150850
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.215703050445885e-07,
+      "loss": 0.561,
+      "step": 150860
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.182796406594492e-07,
+      "loss": 0.575,
+      "step": 150870
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.149889762743099e-07,
+      "loss": 0.5922,
+      "step": 150880
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.116983118891704e-07,
+      "loss": 0.5909,
+      "step": 150890
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.084076475040311e-07,
+      "loss": 0.5818,
+      "step": 150900
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.051169831188917e-07,
+      "loss": 0.557,
+      "step": 150910
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.018263187337523e-07,
+      "loss": 0.5128,
+      "step": 150920
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.985356543486131e-07,
+      "loss": 0.5745,
+      "step": 150930
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.952449899634736e-07,
+      "loss": 0.5563,
+      "step": 150940
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.919543255783343e-07,
+      "loss": 0.5839,
+      "step": 150950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.88663661193195e-07,
+      "loss": 0.5456,
+      "step": 150960
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.853729968080556e-07,
+      "loss": 0.5665,
+      "step": 150970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.820823324229162e-07,
+      "loss": 0.651,
+      "step": 150980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.787916680377769e-07,
+      "loss": 0.4602,
+      "step": 150990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.755010036526375e-07,
+      "loss": 0.5333,
+      "step": 151000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.722103392674981e-07,
+      "loss": 0.639,
+      "step": 151010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.689196748823588e-07,
+      "loss": 0.4671,
+      "step": 151020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.656290104972194e-07,
+      "loss": 0.6438,
+      "step": 151030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6233834611208007e-07,
+      "loss": 0.5477,
+      "step": 151040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5904768172694075e-07,
+      "loss": 0.3989,
+      "step": 151050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5575701734180133e-07,
+      "loss": 0.5898,
+      "step": 151060
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5246635295666196e-07,
+      "loss": 0.6178,
+      "step": 151070
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.491756885715226e-07,
+      "loss": 0.5653,
+      "step": 151080
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4588502418638323e-07,
+      "loss": 0.5154,
+      "step": 151090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.425943598012439e-07,
+      "loss": 0.6317,
+      "step": 151100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.393036954161045e-07,
+      "loss": 0.5839,
+      "step": 151110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.360130310309652e-07,
+      "loss": 0.6181,
+      "step": 151120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3272236664582586e-07,
+      "loss": 0.7501,
+      "step": 151130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2943170226068644e-07,
+      "loss": 0.4617,
+      "step": 151140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2614103787554707e-07,
+      "loss": 0.5703,
+      "step": 151150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2285037349040776e-07,
+      "loss": 0.5864,
+      "step": 151160
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1955970910526834e-07,
+      "loss": 0.654,
+      "step": 151170
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.16269044720129e-07,
+      "loss": 0.5759,
+      "step": 151180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.129783803349897e-07,
+      "loss": 0.4995,
+      "step": 151190
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.096877159498503e-07,
+      "loss": 0.6605,
+      "step": 151200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0639705156471097e-07,
+      "loss": 0.4979,
+      "step": 151210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0310638717957155e-07,
+      "loss": 0.4997,
+      "step": 151220
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.998157227944322e-07,
+      "loss": 0.5183,
+      "step": 151230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9652505840929287e-07,
+      "loss": 0.5514,
+      "step": 151240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9323439402415345e-07,
+      "loss": 0.5516,
+      "step": 151250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8994372963901413e-07,
+      "loss": 0.5253,
+      "step": 151260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8665306525387477e-07,
+      "loss": 0.6202,
+      "step": 151270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8336240086873545e-07,
+      "loss": 0.5065,
+      "step": 151280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.800717364835961e-07,
+      "loss": 0.5642,
+      "step": 151290
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7678107209845666e-07,
+      "loss": 0.9198,
+      "step": 151300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7349040771331735e-07,
+      "loss": 0.5321,
+      "step": 151310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.70199743328178e-07,
+      "loss": 0.7025,
+      "step": 151320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.669090789430386e-07,
+      "loss": 0.6832,
+      "step": 151330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6361841455789924e-07,
+      "loss": 0.5932,
+      "step": 151340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6032775017275993e-07,
+      "loss": 0.6146,
+      "step": 151350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5703708578762056e-07,
+      "loss": 0.62,
+      "step": 151360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5374642140248114e-07,
+      "loss": 0.5987,
+      "step": 151370
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.504557570173418e-07,
+      "loss": 0.5046,
+      "step": 151380
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4716509263220246e-07,
+      "loss": 0.5439,
+      "step": 151390
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.438744282470631e-07,
+      "loss": 0.5152,
+      "step": 151400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.405837638619237e-07,
+      "loss": 0.6146,
+      "step": 151410
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.372930994767844e-07,
+      "loss": 0.5425,
+      "step": 151420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3400243509164504e-07,
+      "loss": 0.6138,
+      "step": 151430
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.307117707065056e-07,
+      "loss": 0.4964,
+      "step": 151440
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.274211063213663e-07,
+      "loss": 0.55,
+      "step": 151450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2413044193622694e-07,
+      "loss": 0.5082,
+      "step": 151460
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2083977755108757e-07,
+      "loss": 0.5836,
+      "step": 151470
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1754911316594825e-07,
+      "loss": 0.5723,
+      "step": 151480
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.142584487808089e-07,
+      "loss": 0.6629,
+      "step": 151490
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.109677843956695e-07,
+      "loss": 0.5431,
+      "step": 151500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0767712001053015e-07,
+      "loss": 0.6164,
+      "step": 151510
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.043864556253908e-07,
+      "loss": 0.5549,
+      "step": 151520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.010957912402514e-07,
+      "loss": 0.5952,
+      "step": 151530
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9780512685511205e-07,
+      "loss": 0.56,
+      "step": 151540
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9451446246997273e-07,
+      "loss": 0.6225,
+      "step": 151550
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9122379808483336e-07,
+      "loss": 0.6818,
+      "step": 151560
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.87933133699694e-07,
+      "loss": 0.5423,
+      "step": 151570
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8464246931455463e-07,
+      "loss": 0.534,
+      "step": 151580
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8135180492941526e-07,
+      "loss": 0.5334,
+      "step": 151590
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.780611405442759e-07,
+      "loss": 0.5269,
+      "step": 151600
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.747704761591365e-07,
+      "loss": 0.6957,
+      "step": 151610
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.714798117739972e-07,
+      "loss": 0.7307,
+      "step": 151620
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6818914738885784e-07,
+      "loss": 0.5035,
+      "step": 151630
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.648984830037185e-07,
+      "loss": 0.4998,
+      "step": 151640
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.616078186185791e-07,
+      "loss": 0.4805,
+      "step": 151650
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5831715423343974e-07,
+      "loss": 0.5877,
+      "step": 151660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5502648984830037e-07,
+      "loss": 0.5386,
+      "step": 151670
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.51735825463161e-07,
+      "loss": 0.5998,
+      "step": 151680
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.484451610780217e-07,
+      "loss": 0.5733,
+      "step": 151690
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.451544966928823e-07,
+      "loss": 0.4845,
+      "step": 151700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4186383230774295e-07,
+      "loss": 0.5771,
+      "step": 151710
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.385731679226036e-07,
+      "loss": 0.6894,
+      "step": 151720
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3528250353746424e-07,
+      "loss": 0.5863,
+      "step": 151730
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3199183915232488e-07,
+      "loss": 0.5776,
+      "step": 151740
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2870117476718548e-07,
+      "loss": 0.5526,
+      "step": 151750
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2541051038204617e-07,
+      "loss": 0.557,
+      "step": 151760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.221198459969068e-07,
+      "loss": 0.57,
+      "step": 151770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1882918161176743e-07,
+      "loss": 0.5936,
+      "step": 151780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1553851722662804e-07,
+      "loss": 0.6093,
+      "step": 151790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1224785284148872e-07,
+      "loss": 0.4734,
+      "step": 151800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0895718845634935e-07,
+      "loss": 0.6922,
+      "step": 151810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0566652407120996e-07,
+      "loss": 0.5889,
+      "step": 151820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0237585968607064e-07,
+      "loss": 0.5597,
+      "step": 151830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9908519530093128e-07,
+      "loss": 0.5334,
+      "step": 151840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.957945309157919e-07,
+      "loss": 0.5225,
+      "step": 151850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9250386653065254e-07,
+      "loss": 0.5711,
+      "step": 151860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.892132021455132e-07,
+      "loss": 0.6013,
+      "step": 151870
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8592253776037383e-07,
+      "loss": 0.5582,
+      "step": 151880
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8263187337523446e-07,
+      "loss": 0.5812,
+      "step": 151890
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.793412089900951e-07,
+      "loss": 0.5006,
+      "step": 151900
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7605054460495575e-07,
+      "loss": 0.6082,
+      "step": 151910
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7275988021981639e-07,
+      "loss": 0.5478,
+      "step": 151920
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6946921583467702e-07,
+      "loss": 0.6429,
+      "step": 151930
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6617855144953768e-07,
+      "loss": 0.5072,
+      "step": 151940
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.628878870643983e-07,
+      "loss": 0.6137,
+      "step": 151950
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5959722267925894e-07,
+      "loss": 0.6271,
+      "step": 151960
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5630655829411957e-07,
+      "loss": 0.5752,
+      "step": 151970
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5301589390898023e-07,
+      "loss": 0.6216,
+      "step": 151980
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.497252295238409e-07,
+      "loss": 0.5995,
+      "step": 151990
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.464345651387015e-07,
+      "loss": 0.4821,
+      "step": 152000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 152445,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 2.41529675317248e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}