diff --git "a/checkpoint-10000/trainer_state.json" "b/checkpoint-10000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10000/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.24017677010279564,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999999855889116,
+      "loss": 3.4871,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999999357728057,
+      "loss": 2.9969,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999998503737748,
+      "loss": 2.9734,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999997293918308,
+      "loss": 2.9203,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999995728269912,
+      "loss": 2.7908,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999993806792782,
+      "loss": 2.7973,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499999152948719,
+      "loss": 2.8511,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999988896353463,
+      "loss": 2.8423,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999985907391973,
+      "loss": 2.8411,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999982562603146,
+      "loss": 2.838,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499997886198746,
+      "loss": 2.8751,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999974805545439,
+      "loss": 2.7868,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999970393277663,
+      "loss": 2.8003,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999965625184758,
+      "loss": 2.8343,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999960501267404,
+      "loss": 2.8053,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999955021526329,
+      "loss": 2.7828,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999949185962313,
+      "loss": 2.7822,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999942994576189,
+      "loss": 2.7426,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999936447368836,
+      "loss": 2.7808,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999929544341185,
+      "loss": 2.7522,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999922285494221,
+      "loss": 2.8008,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999914670828975,
+      "loss": 2.748,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999906700346533,
+      "loss": 2.783,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999898374048027,
+      "loss": 2.7879,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999889691934643,
+      "loss": 2.696,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999880654007619,
+      "loss": 2.7354,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999871260268238,
+      "loss": 2.7336,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999861510717839,
+      "loss": 2.7077,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999851405357809,
+      "loss": 2.7408,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999840944189586,
+      "loss": 2.7115,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499983012721466,
+      "loss": 2.6973,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999818954434569,
+      "loss": 2.6644,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999807425850907,
+      "loss": 2.7442,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499979554146531,
+      "loss": 2.7007,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999783301279471,
+      "loss": 2.7284,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999770705295135,
+      "loss": 2.6649,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999757753514091,
+      "loss": 2.7269,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999744445938185,
+      "loss": 2.64,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999730782569309,
+      "loss": 2.6886,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499971676340941,
+      "loss": 2.6911,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999702388460482,
+      "loss": 2.6814,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999687657724571,
+      "loss": 2.7235,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999672571203774,
+      "loss": 2.6898,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999657128900239,
+      "loss": 2.6222,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999641330816162,
+      "loss": 2.6827,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999625176953794,
+      "loss": 2.6634,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999608667315432,
+      "loss": 2.6976,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999591801903428,
+      "loss": 2.6856,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499957458072018,
+      "loss": 2.624,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499955700376814,
+      "loss": 2.6693,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999539071049812,
+      "loss": 2.7058,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999520782567746,
+      "loss": 2.6878,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999502138324544,
+      "loss": 2.6674,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999483138322863,
+      "loss": 2.634,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999463782565405,
+      "loss": 2.6242,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999444071054925,
+      "loss": 2.636,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999424003794229,
+      "loss": 2.6926,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999403580786175,
+      "loss": 2.6564,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999382802033666,
+      "loss": 2.6474,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999361667539663,
+      "loss": 2.6462,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999340177307172,
+      "loss": 2.6069,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999318331339252,
+      "loss": 2.632,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999296129639014,
+      "loss": 2.6533,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999273572209616,
+      "loss": 2.6524,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499925065905427,
+      "loss": 2.6229,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999227390176237,
+      "loss": 2.6587,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999203765578828,
+      "loss": 2.6175,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999179785265406,
+      "loss": 2.6214,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999155449239384,
+      "loss": 2.6338,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999130757504227,
+      "loss": 2.6364,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999105710063449,
+      "loss": 2.6509,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999080306920613,
+      "loss": 2.6465,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999054548079338,
+      "loss": 2.6452,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999028433543286,
+      "loss": 2.6858,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004999001963316179,
+      "loss": 2.6624,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998975137401781,
+      "loss": 2.5747,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499894795580391,
+      "loss": 2.6239,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998920418526438,
+      "loss": 2.612,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998892525573282,
+      "loss": 2.6312,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998864276948413,
+      "loss": 2.6321,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499883567265585,
+      "loss": 2.6128,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998806712699667,
+      "loss": 2.6299,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998777397083982,
+      "loss": 2.6444,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998747725812971,
+      "loss": 2.6058,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998717698890857,
+      "loss": 2.5646,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998687316321911,
+      "loss": 2.6089,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998656578110461,
+      "loss": 2.6257,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998625484260881,
+      "loss": 2.6794,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998594034777594,
+      "loss": 2.6585,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998562229665079,
+      "loss": 2.6161,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998530068927862,
+      "loss": 2.5684,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499849755257052,
+      "loss": 2.6284,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998464680597682,
+      "loss": 2.5965,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998431453014028,
+      "loss": 2.5825,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998397869824284,
+      "loss": 2.6056,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998363931033231,
+      "loss": 2.5434,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998329636645702,
+      "loss": 2.6535,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998294986666576,
+      "loss": 2.7084,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998259981100785,
+      "loss": 2.6648,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998224619953312,
+      "loss": 2.6241,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499818890322919,
+      "loss": 2.5672,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998152830933501,
+      "loss": 2.6968,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998116403071381,
+      "loss": 2.5603,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004998079619648014,
+      "loss": 2.6031,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004998042480668638,
+      "loss": 2.5779,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004998004986138534,
+      "loss": 2.6792,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997967136063043,
+      "loss": 2.6038,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997928930447551,
+      "loss": 2.6469,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997890369297496,
+      "loss": 2.62,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997851452618364,
+      "loss": 2.6517,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997812180415697,
+      "loss": 2.6313,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997772552695085,
+      "loss": 2.62,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997732569462167,
+      "loss": 2.6205,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997692230722633,
+      "loss": 2.5727,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997651536482227,
+      "loss": 2.5626,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997610486746738,
+      "loss": 2.6196,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997569081522012,
+      "loss": 2.6219,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997527320813939,
+      "loss": 2.6443,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997485204628466,
+      "loss": 2.6027,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997442732971584,
+      "loss": 2.6495,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997399905849342,
+      "loss": 2.6099,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997356723267833,
+      "loss": 2.6074,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997313185233204,
+      "loss": 2.6089,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997269291751651,
+      "loss": 2.6539,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997225042829423,
+      "loss": 2.6044,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997180438472816,
+      "loss": 2.5768,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997135478688181,
+      "loss": 2.5801,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997090163481916,
+      "loss": 2.6145,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004997044492860471,
+      "loss": 2.6338,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996998466830345,
+      "loss": 2.6048,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996952085398091,
+      "loss": 2.6023,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000499690534857031,
+      "loss": 2.6037,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996858256353653,
+      "loss": 2.5876,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996810808754824,
+      "loss": 2.5932,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996763005780576,
+      "loss": 2.6014,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996714847437712,
+      "loss": 2.6211,
+      "step": 1360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996666333733088,
+      "loss": 2.6615,
+      "step": 1370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996617464673609,
+      "loss": 2.6057,
+      "step": 1380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996568240266228,
+      "loss": 2.5722,
+      "step": 1390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996518660517954,
+      "loss": 2.5588,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996468725435843,
+      "loss": 2.612,
+      "step": 1410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996418435027002,
+      "loss": 2.5926,
+      "step": 1420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996367789298589,
+      "loss": 2.5635,
+      "step": 1430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996316788257812,
+      "loss": 2.5991,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004996265431911932,
+      "loss": 2.6328,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004996213720268257,
+      "loss": 2.5715,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004996161653334147,
+      "loss": 2.5659,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004996109231117013,
+      "loss": 2.5627,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004996056453624317,
+      "loss": 2.5869,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004996003320863571,
+      "loss": 2.6017,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995949832842337,
+      "loss": 2.532,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995895989568227,
+      "loss": 2.6129,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995841791048906,
+      "loss": 2.6216,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995787237292089,
+      "loss": 2.5878,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995732328305538,
+      "loss": 2.5592,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000499567706409707,
+      "loss": 2.637,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995621444674552,
+      "loss": 2.571,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995565470045898,
+      "loss": 2.5604,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995509140219076,
+      "loss": 2.6172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995452455202105,
+      "loss": 2.6257,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995395415003051,
+      "loss": 2.5337,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995338019630033,
+      "loss": 2.5631,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000499528026909122,
+      "loss": 2.5511,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995222163394834,
+      "loss": 2.609,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995163702549142,
+      "loss": 2.5627,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995104886562466,
+      "loss": 2.5877,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004995045715443178,
+      "loss": 2.5274,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994986189199701,
+      "loss": 2.545,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994926307840505,
+      "loss": 2.5648,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994866071374115,
+      "loss": 2.5424,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994805479809103,
+      "loss": 2.5759,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994744533154094,
+      "loss": 2.6146,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994683231417762,
+      "loss": 2.589,
+      "step": 1730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994621574608833,
+      "loss": 2.6288,
+      "step": 1740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994559562736083,
+      "loss": 2.5964,
+      "step": 1750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994497195808336,
+      "loss": 2.5697,
+      "step": 1760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994434473834472,
+      "loss": 2.5815,
+      "step": 1770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994371396823416,
+      "loss": 2.5621,
+      "step": 1780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994307964784147,
+      "loss": 2.5517,
+      "step": 1790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994244177725693,
+      "loss": 2.6157,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994180035657132,
+      "loss": 2.5651,
+      "step": 1810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994115538587595,
+      "loss": 2.5868,
+      "step": 1820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004994050686526261,
+      "loss": 2.6103,
+      "step": 1830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004993985479482362,
+      "loss": 2.6364,
+      "step": 1840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004993919917465178,
+      "loss": 2.6651,
+      "step": 1850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004993854000484038,
+      "loss": 2.5453,
+      "step": 1860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000499378772854833,
+      "loss": 2.5657,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000499372110166748,
+      "loss": 2.5375,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993654119850975,
+      "loss": 2.5654,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993586783108349,
+      "loss": 2.6298,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993519091449184,
+      "loss": 2.5405,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993451044883116,
+      "loss": 2.6205,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993382643419831,
+      "loss": 2.5851,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993313887069062,
+      "loss": 2.5528,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993244775840596,
+      "loss": 2.5741,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993175309744273,
+      "loss": 2.5482,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993105488789976,
+      "loss": 2.5643,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004993035312987645,
+      "loss": 2.5748,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992964782347269,
+      "loss": 2.5976,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992893896878884,
+      "loss": 2.5567,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000499282265659258,
+      "loss": 2.516,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00049927510614985,
+      "loss": 2.5344,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992679111606829,
+      "loss": 2.6098,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992606806927813,
+      "loss": 2.567,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992534147471738,
+      "loss": 2.5139,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992461133248951,
+      "loss": 2.6271,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992387764269841,
+      "loss": 2.6127,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992314040544851,
+      "loss": 2.5665,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992239962084475,
+      "loss": 2.5802,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992165528899256,
+      "loss": 2.5765,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000499209074099979,
+      "loss": 2.5625,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004992015598396718,
+      "loss": 2.5349,
+      "step": 2120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991940101100739,
+      "loss": 2.5717,
+      "step": 2130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991864249122598,
+      "loss": 2.5886,
+      "step": 2140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991788042473088,
+      "loss": 2.5742,
+      "step": 2150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000499171148116306,
+      "loss": 2.5869,
+      "step": 2160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991634565203407,
+      "loss": 2.5709,
+      "step": 2170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000499155729460508,
+      "loss": 2.6118,
+      "step": 2180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991479669379076,
+      "loss": 2.5576,
+      "step": 2190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991401689536442,
+      "loss": 2.5306,
+      "step": 2200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991323355088279,
+      "loss": 2.5807,
+      "step": 2210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991244666045735,
+      "loss": 2.5359,
+      "step": 2220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991165622420011,
+      "loss": 2.5796,
+      "step": 2230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991086224222358,
+      "loss": 2.5182,
+      "step": 2240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004991006471464075,
+      "loss": 2.6564,
+      "step": 2250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004990926364156513,
+      "loss": 2.638,
+      "step": 2260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004990845902311077,
+      "loss": 2.5452,
+      "step": 2270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004990765085939216,
+      "loss": 2.5622,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990683915052436,
+      "loss": 2.5423,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990602389662287,
+      "loss": 2.5891,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990520509780374,
+      "loss": 2.5968,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990438275418351,
+      "loss": 2.5594,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990355686587923,
+      "loss": 2.5458,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990272743300843,
+      "loss": 2.5304,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000499018944556892,
+      "loss": 2.5505,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990105793404008,
+      "loss": 2.6207,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004990021786818013,
+      "loss": 2.6038,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000498993742582289,
+      "loss": 2.5597,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989852710430651,
+      "loss": 2.5344,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000498976764065335,
+      "loss": 2.6021,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989682216503097,
+      "loss": 2.5924,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989596437992049,
+      "loss": 2.5742,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989510305132416,
+      "loss": 2.5709,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989423817936458,
+      "loss": 2.6258,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989336976416484,
+      "loss": 2.5664,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989249780584853,
+      "loss": 2.5551,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989162230453978,
+      "loss": 2.5816,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004989074326036319,
+      "loss": 2.5314,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988986067344388,
+      "loss": 2.6238,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988897454390747,
+      "loss": 2.6155,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988808487188009,
+      "loss": 2.6048,
+      "step": 2510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988719165748835,
+      "loss": 2.5906,
+      "step": 2520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000498862949008594,
+      "loss": 2.5826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988539460212088,
+      "loss": 2.5686,
+      "step": 2540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988449076140091,
+      "loss": 2.5948,
+      "step": 2550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988358337882816,
+      "loss": 2.5445,
+      "step": 2560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988267245453175,
+      "loss": 2.5417,
+      "step": 2570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988175798864137,
+      "loss": 2.5712,
+      "step": 2580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004988083998128714,
+      "loss": 2.5913,
+      "step": 2590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987991843259975,
+      "loss": 2.5529,
+      "step": 2600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987899334271035,
+      "loss": 2.5581,
+      "step": 2610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987806471175063,
+      "loss": 2.5334,
+      "step": 2620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987713253985273,
+      "loss": 2.5911,
+      "step": 2630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987619682714936,
+      "loss": 2.5962,
+      "step": 2640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987525757377368,
+      "loss": 2.5038,
+      "step": 2650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987431477985938,
+      "loss": 2.5732,
+      "step": 2660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987336844554066,
+      "loss": 2.5667,
+      "step": 2670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987241857095221,
+      "loss": 2.5632,
+      "step": 2680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987146515622921,
+      "loss": 2.545,
+      "step": 2690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004987050820150739,
+      "loss": 2.5234,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986954770692293,
+      "loss": 2.5294,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986858367261256,
+      "loss": 2.6312,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986761609871347,
+      "loss": 2.5496,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986664498536339,
+      "loss": 2.5203,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986567033270055,
+      "loss": 2.6495,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986469214086364,
+      "loss": 2.592,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986371040999192,
+      "loss": 2.532,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986272514022512,
+      "loss": 2.591,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986173633170346,
+      "loss": 2.5353,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004986074398456769,
+      "loss": 2.565,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985974809895903,
+      "loss": 2.6095,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985874867501927,
+      "loss": 2.5576,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985774571289062,
+      "loss": 2.5559,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985673921271584,
+      "loss": 2.5174,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985572917463821,
+      "loss": 2.5949,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985471559880146,
+      "loss": 2.5327,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985369848534988,
+      "loss": 2.6106,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985267783442823,
+      "loss": 2.6212,
+      "step": 2880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985165364618176,
+      "loss": 2.6182,
+      "step": 2890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004985062592075627,
+      "loss": 2.5897,
+      "step": 2900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984959465829804,
+      "loss": 2.5322,
+      "step": 2910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984855985895382,
+      "loss": 2.4541,
+      "step": 2920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984752152287093,
+      "loss": 2.5546,
+      "step": 2930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984647965019714,
+      "loss": 2.5689,
+      "step": 2940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984543424108075,
+      "loss": 2.5218,
+      "step": 2950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984438529567054,
+      "loss": 2.6078,
+      "step": 2960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984333281411582,
+      "loss": 2.5947,
+      "step": 2970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000498422767965664,
+      "loss": 2.5687,
+      "step": 2980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984121724317256,
+      "loss": 2.5415,
+      "step": 2990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004984015415408514,
+      "loss": 2.5362,
+      "step": 3000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983908752945542,
+      "loss": 2.521,
+      "step": 3010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983801736943524,
+      "loss": 2.5902,
+      "step": 3020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000498369436741769,
+      "loss": 2.565,
+      "step": 3030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983586644383322,
+      "loss": 2.5585,
+      "step": 3040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983478567855754,
+      "loss": 2.5787,
+      "step": 3050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983370137850369,
+      "loss": 2.5266,
+      "step": 3060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983261354382598,
+      "loss": 2.4995,
+      "step": 3070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983152217467924,
+      "loss": 2.5792,
+      "step": 3080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004983042727121883,
+      "loss": 2.5772,
+      "step": 3090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004982932883360058,
+      "loss": 2.4929,
+      "step": 3100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004982822686198083,
+      "loss": 2.508,
+      "step": 3110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004982712135651643,
+      "loss": 2.5573,
+      "step": 3120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004982601231736472,
+      "loss": 2.6541,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004982489974468355,
+      "loss": 2.4863,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000498237836386313,
+      "loss": 2.5646,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004982266399936678,
+      "loss": 2.5657,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000498215408270494,
+      "loss": 2.5708,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00049820414121839,
+      "loss": 2.5032,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004981928388389594,
+      "loss": 2.5778,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000498181501133811,
+      "loss": 2.5989,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004981701281045585,
+      "loss": 2.5047,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004981587197528205,
+      "loss": 2.5285,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000498147276080221,
+      "loss": 2.5799,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004981357970883886,
+      "loss": 2.5478,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004981242827789572,
+      "loss": 2.5617,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004981138897052764,
+      "loss": 2.5479,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004981023082969261,
+      "loss": 2.5556,
+      "step": 3270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004980906915757434,
+      "loss": 2.5612,
+      "step": 3280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004980790395433817,
+      "loss": 2.4746,
+      "step": 3290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004980673522014993,
+      "loss": 2.5154,
+      "step": 3300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00049805562955176,
+      "loss": 2.5829,
+      "step": 3310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004980438715958318,
+      "loss": 2.5546,
+      "step": 3320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004980320783353888,
+      "loss": 2.6171,
+      "step": 3330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004980202497721091,
+      "loss": 2.5568,
+      "step": 3340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004980083859076766,
+      "loss": 2.6162,
+      "step": 3350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979964867437797,
+      "loss": 2.5808,
+      "step": 3360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979845522821121,
+      "loss": 2.5665,
+      "step": 3370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979725825243724,
+      "loss": 2.5723,
+      "step": 3380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979605774722644,
+      "loss": 2.5039,
+      "step": 3390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979485371274967,
+      "loss": 2.5566,
+      "step": 3400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979364614917831,
+      "loss": 2.5481,
+      "step": 3410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979243505668421,
+      "loss": 2.545,
+      "step": 3420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004979122043543979,
+      "loss": 2.5013,
+      "step": 3430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000497900022856179,
+      "loss": 2.5336,
+      "step": 3440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978878060739191,
+      "loss": 2.5577,
+      "step": 3450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978755540093573,
+      "loss": 2.5775,
+      "step": 3460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978632666642374,
+      "loss": 2.5011,
+      "step": 3470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978509440403083,
+      "loss": 2.5834,
+      "step": 3480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978385861393237,
+      "loss": 2.5143,
+      "step": 3490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978261929630427,
+      "loss": 2.574,
+      "step": 3500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978137645132294,
+      "loss": 2.5587,
+      "step": 3510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004978013007916523,
+      "loss": 2.5767,
+      "step": 3520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004977888018000858,
+      "loss": 2.5945,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004977762675403087,
+      "loss": 2.5834,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004977636980141051,
+      "loss": 2.6018,
+      "step": 3550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004977510932232639,
+      "loss": 2.5104,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004977384531695794,
+      "loss": 2.4966,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004977257778548506,
+      "loss": 2.5948,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004977130672808815,
+      "loss": 2.5277,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004977003214494813,
+      "loss": 2.5563,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004976875403624641,
+      "loss": 2.5506,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004976747240216491,
+      "loss": 2.5865,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004976618724288604,
+      "loss": 2.583,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004976489855859272,
+      "loss": 2.5752,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004976360634946838,
+      "loss": 2.5702,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004976231061569694,
+      "loss": 2.6195,
+      "step": 3660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004976101135746281,
+      "loss": 2.5673,
+      "step": 3670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975970857495093,
+      "loss": 2.5204,
+      "step": 3680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975840226834672,
+      "loss": 2.5061,
+      "step": 3690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975709243783612,
+      "loss": 2.538,
+      "step": 3700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975577908360554,
+      "loss": 2.536,
+      "step": 3710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975446220584195,
+      "loss": 2.554,
+      "step": 3720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975314180473274,
+      "loss": 2.5227,
+      "step": 3730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975181788046585,
+      "loss": 2.5565,
+      "step": 3740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004975049043322976,
+      "loss": 2.5408,
+      "step": 3750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004974915946321335,
+      "loss": 2.5228,
+      "step": 3760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000497478249706061,
+      "loss": 2.6042,
+      "step": 3770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004974648695559794,
+      "loss": 2.5345,
+      "step": 3780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000497451454183793,
+      "loss": 2.5326,
+      "step": 3790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004974380035914114,
+      "loss": 2.5002,
+      "step": 3800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004974245177807489,
+      "loss": 2.5726,
+      "step": 3810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004974109967537251,
+      "loss": 2.5553,
+      "step": 3820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973974405122644,
+      "loss": 2.5504,
+      "step": 3830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973838490582962,
+      "loss": 2.5844,
+      "step": 3840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973702223937552,
+      "loss": 2.5305,
+      "step": 3850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973565605205807,
+      "loss": 2.5156,
+      "step": 3860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973428634407174,
+      "loss": 2.5365,
+      "step": 3870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973291311561146,
+      "loss": 2.519,
+      "step": 3880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973153636687271,
+      "loss": 2.5673,
+      "step": 3890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004973015609805142,
+      "loss": 2.5224,
+      "step": 3900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004972877230934407,
+      "loss": 2.4995,
+      "step": 3910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000497273850009476,
+      "loss": 2.5978,
+      "step": 3920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004972599417305948,
+      "loss": 2.5886,
+      "step": 3930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004972459982587765,
+      "loss": 2.6197,
+      "step": 3940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004972320195960059,
+      "loss": 2.5968,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004972180057442726,
+      "loss": 2.5245,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004972039567055711,
+      "loss": 2.5286,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004971898724819011,
+      "loss": 2.5388,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004971757530752671,
+      "loss": 2.5169,
+      "step": 3990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004971615984876789,
+      "loss": 2.5923,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004971474087211511,
+      "loss": 2.5335,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004971331837777033,
+      "loss": 2.5831,
+      "step": 4020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004971189236593603,
+      "loss": 2.5268,
+      "step": 4030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004971046283681515,
+      "loss": 2.5427,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004970902979061119,
+      "loss": 2.5611,
+      "step": 4050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000497075932275281,
+      "loss": 2.6082,
+      "step": 4060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004970615314777035,
+      "loss": 2.5046,
+      "step": 4070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004970470955154291,
+      "loss": 2.5643,
+      "step": 4080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004970326243905126,
+      "loss": 2.5789,
+      "step": 4090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004970181181050134,
+      "loss": 2.4757,
+      "step": 4100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004970035766609965,
+      "loss": 2.4911,
+      "step": 4110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004969890000605315,
+      "loss": 2.5294,
+      "step": 4120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004969743883056932,
+      "loss": 2.4996,
+      "step": 4130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004969597413985612,
+      "loss": 2.5432,
+      "step": 4140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004969450593412202,
+      "loss": 2.5731,
+      "step": 4150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004969303421357601,
+      "loss": 2.6248,
+      "step": 4160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004969155897842754,
+      "loss": 2.5902,
+      "step": 4170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004969008022888659,
+      "loss": 2.4852,
+      "step": 4180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004968859796516365,
+      "loss": 2.4911,
+      "step": 4190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004968711218746967,
+      "loss": 2.5582,
+      "step": 4200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004968562289601613,
+      "loss": 2.5825,
+      "step": 4210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004968413009101502,
+      "loss": 2.5981,
+      "step": 4220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000496826337726788,
+      "loss": 2.5896,
+      "step": 4230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004968113394122044,
+      "loss": 2.5474,
+      "step": 4240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004967963059685341,
+      "loss": 2.5784,
+      "step": 4250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004967812373979169,
+      "loss": 2.5398,
+      "step": 4260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004967661337024976,
+      "loss": 2.5089,
+      "step": 4270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004967509948844258,
+      "loss": 2.519,
+      "step": 4280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004967358209458564,
+      "loss": 2.5411,
+      "step": 4290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004967206118889491,
+      "loss": 2.5732,
+      "step": 4300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004967053677158686,
+      "loss": 2.5164,
+      "step": 4310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004966900884287845,
+      "loss": 2.5663,
+      "step": 4320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004966747740298717,
+      "loss": 2.5126,
+      "step": 4330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004966594245213098,
+      "loss": 2.5704,
+      "step": 4340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004966440399052836,
+      "loss": 2.5488,
+      "step": 4350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004966286201839828,
+      "loss": 2.5444,
+      "step": 4360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004966131653596021,
+      "loss": 2.5005,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004965976754343413,
+      "loss": 2.5497,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000496582150410405,
+      "loss": 2.6044,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004965665902900029,
+      "loss": 2.6103,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004965509950753498,
+      "loss": 2.4993,
+      "step": 4410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004965353647686653,
+      "loss": 2.4838,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004965196993721743,
+      "loss": 2.4869,
+      "step": 4430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000496503998888106,
+      "loss": 2.5232,
+      "step": 4440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004964882633186955,
+      "loss": 2.514,
+      "step": 4450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004964724926661825,
+      "loss": 2.5312,
+      "step": 4460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004964566869328115,
+      "loss": 2.578,
+      "step": 4470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004964408461208319,
+      "loss": 2.557,
+      "step": 4480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004964249702324989,
+      "loss": 2.4945,
+      "step": 4490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004964090592700718,
+      "loss": 2.5046,
+      "step": 4500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004963931132358153,
+      "loss": 2.5283,
+      "step": 4510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000496377132131999,
+      "loss": 2.5303,
+      "step": 4520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004963611159608975,
+      "loss": 2.4897,
+      "step": 4530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004963450647247906,
+      "loss": 2.5238,
+      "step": 4540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004963289784259627,
+      "loss": 2.5362,
+      "step": 4550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004963128570667034,
+      "loss": 2.5355,
+      "step": 4560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004962967006493074,
+      "loss": 2.5849,
+      "step": 4570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004962805091760741,
+      "loss": 2.5662,
+      "step": 4580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004962642826493082,
+      "loss": 2.4892,
+      "step": 4590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004962480210713193,
+      "loss": 2.5777,
+      "step": 4600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004962317244444217,
+      "loss": 2.5342,
+      "step": 4610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004962153927709353,
+      "loss": 2.5293,
+      "step": 4620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004961990260531843,
+      "loss": 2.5612,
+      "step": 4630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004961826242934983,
+      "loss": 2.4989,
+      "step": 4640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004961661874942117,
+      "loss": 2.557,
+      "step": 4650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004961497156576641,
+      "loss": 2.542,
+      "step": 4660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004961332087862,
+      "loss": 2.5551,
+      "step": 4670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004961166668821687,
+      "loss": 2.5445,
+      "step": 4680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004961000899479247,
+      "loss": 2.5366,
+      "step": 4690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004960834779858275,
+      "loss": 2.5445,
+      "step": 4700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004960668309982415,
+      "loss": 2.5115,
+      "step": 4710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000496050148987536,
+      "loss": 2.562,
+      "step": 4720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004960334319560854,
+      "loss": 2.5237,
+      "step": 4730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004960166799062691,
+      "loss": 2.5659,
+      "step": 4740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004959998928404715,
+      "loss": 2.5704,
+      "step": 4750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004959830707610819,
+      "loss": 2.5654,
+      "step": 4760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004959662136704945,
+      "loss": 2.557,
+      "step": 4770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004959493215711087,
+      "loss": 2.4897,
+      "step": 4780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004959323944653288,
+      "loss": 2.572,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000495915432355564,
+      "loss": 2.5766,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004958984352442286,
+      "loss": 2.5121,
+      "step": 4810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004958814031337419,
+      "loss": 2.4934,
+      "step": 4820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004958643360265279,
+      "loss": 2.6128,
+      "step": 4830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000495847233925016,
+      "loss": 2.5246,
+      "step": 4840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004958300968316402,
+      "loss": 2.5577,
+      "step": 4850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004958129247488397,
+      "loss": 2.484,
+      "step": 4860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004957957176790587,
+      "loss": 2.5616,
+      "step": 4870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004957784756247463,
+      "loss": 2.4936,
+      "step": 4880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004957611985883566,
+      "loss": 2.58,
+      "step": 4890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004957438865723486,
+      "loss": 2.5083,
+      "step": 4900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004957265395791865,
+      "loss": 2.4819,
+      "step": 4910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000495709157611339,
+      "loss": 2.5151,
+      "step": 4920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004956917406712805,
+      "loss": 2.4644,
+      "step": 4930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004956742887614897,
+      "loss": 2.5624,
+      "step": 4940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004956568018844507,
+      "loss": 2.5754,
+      "step": 4950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004956392800426525,
+      "loss": 2.5833,
+      "step": 4960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004956217232385888,
+      "loss": 2.5116,
+      "step": 4970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004956041314747587,
+      "loss": 2.4878,
+      "step": 4980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004955865047536658,
+      "loss": 2.627,
+      "step": 4990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004955688430778193,
+      "loss": 2.4976,
+      "step": 5000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004955511464497329,
+      "loss": 2.5709,
+      "step": 5010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004955334148719252,
+      "loss": 2.5895,
+      "step": 5020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004955156483469201,
+      "loss": 2.5231,
+      "step": 5030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004954978468772463,
+      "loss": 2.5312,
+      "step": 5040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004954800104654375,
+      "loss": 2.5598,
+      "step": 5050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004954621391140326,
+      "loss": 2.5421,
+      "step": 5060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000495444232825575,
+      "loss": 2.5199,
+      "step": 5070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004954262916026135,
+      "loss": 2.4944,
+      "step": 5080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004954083154477015,
+      "loss": 2.5115,
+      "step": 5090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004953903043633977,
+      "loss": 2.5126,
+      "step": 5100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004953722583522658,
+      "loss": 2.446,
+      "step": 5110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004953541774168742,
+      "loss": 2.6025,
+      "step": 5120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004953360615597963,
+      "loss": 2.5208,
+      "step": 5130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004953179107836105,
+      "loss": 2.5167,
+      "step": 5140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004952997250909006,
+      "loss": 2.5042,
+      "step": 5150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004952815044842546,
+      "loss": 2.4997,
+      "step": 5160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000495263248966266,
+      "loss": 2.5308,
+      "step": 5170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004952449585395334,
+      "loss": 2.5414,
+      "step": 5180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004952266332066597,
+      "loss": 2.5362,
+      "step": 5190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004952082729702534,
+      "loss": 2.5335,
+      "step": 5200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004951898778329277,
+      "loss": 2.6089,
+      "step": 5210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004951714477973008,
+      "loss": 2.5703,
+      "step": 5220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000495152982865996,
+      "loss": 2.5374,
+      "step": 5230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004951344830416412,
+      "loss": 2.5378,
+      "step": 5240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004951159483268698,
+      "loss": 2.5335,
+      "step": 5250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004950973787243196,
+      "loss": 2.5149,
+      "step": 5260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004950787742366339,
+      "loss": 2.5265,
+      "step": 5270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004950601348664605,
+      "loss": 2.5606,
+      "step": 5280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004950414606164525,
+      "loss": 2.5707,
+      "step": 5290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004950227514892678,
+      "loss": 2.5363,
+      "step": 5300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004950040074875692,
+      "loss": 2.5186,
+      "step": 5310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004949852286140247,
+      "loss": 2.5089,
+      "step": 5320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004949664148713071,
+      "loss": 2.5773,
+      "step": 5330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004949475662620942,
+      "loss": 2.5507,
+      "step": 5340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004949286827890687,
+      "loss": 2.4891,
+      "step": 5350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004949097644549185,
+      "loss": 2.5419,
+      "step": 5360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004948908112623361,
+      "loss": 2.5012,
+      "step": 5370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004948718232140191,
+      "loss": 2.57,
+      "step": 5380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004948528003126702,
+      "loss": 2.5826,
+      "step": 5390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000494833742560997,
+      "loss": 2.5257,
+      "step": 5400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004948146499617119,
+      "loss": 2.5108,
+      "step": 5410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004947955225175325,
+      "loss": 2.542,
+      "step": 5420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004947763602311812,
+      "loss": 2.5763,
+      "step": 5430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004947571631053854,
+      "loss": 2.5096,
+      "step": 5440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004947379311428774,
+      "loss": 2.5886,
+      "step": 5450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004947186643463945,
+      "loss": 2.4859,
+      "step": 5460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004946993627186792,
+      "loss": 2.5634,
+      "step": 5470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004946800262624786,
+      "loss": 2.5464,
+      "step": 5480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004946606549805448,
+      "loss": 2.5726,
+      "step": 5490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000494641248875635,
+      "loss": 2.5298,
+      "step": 5500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004946218079505114,
+      "loss": 2.5529,
+      "step": 5510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000494602332207941,
+      "loss": 2.5434,
+      "step": 5520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004945828216506956,
+      "loss": 2.5649,
+      "step": 5530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004945632762815526,
+      "loss": 2.5616,
+      "step": 5540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004945436961032937,
+      "loss": 2.5637,
+      "step": 5550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004945240811187057,
+      "loss": 2.5155,
+      "step": 5560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004945044313305805,
+      "loss": 2.4839,
+      "step": 5570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004944847467417149,
+      "loss": 2.5336,
+      "step": 5580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004944650273549108,
+      "loss": 2.6058,
+      "step": 5590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004944452731729746,
+      "loss": 2.5009,
+      "step": 5600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004944254841987182,
+      "loss": 2.5141,
+      "step": 5610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000494405660434958,
+      "loss": 2.588,
+      "step": 5620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004943858018845157,
+      "loss": 2.5345,
+      "step": 5630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004943659085502177,
+      "loss": 2.4981,
+      "step": 5640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004943459804348955,
+      "loss": 2.4701,
+      "step": 5650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004943260175413855,
+      "loss": 2.4917,
+      "step": 5660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000494306019872529,
+      "loss": 2.5416,
+      "step": 5670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004942859874311724,
+      "loss": 2.4962,
+      "step": 5680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004942659202201669,
+      "loss": 2.4898,
+      "step": 5690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004942458182423688,
+      "loss": 2.5407,
+      "step": 5700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004942256815006391,
+      "loss": 2.523,
+      "step": 5710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004942055099978438,
+      "loss": 2.5788,
+      "step": 5720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004941853037368543,
+      "loss": 2.5076,
+      "step": 5730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004941650627205463,
+      "loss": 2.6023,
+      "step": 5740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004941447869518009,
+      "loss": 2.4776,
+      "step": 5750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004941244764335038,
+      "loss": 2.5034,
+      "step": 5760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004941041311685461,
+      "loss": 2.566,
+      "step": 5770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004940837511598233,
+      "loss": 2.5236,
+      "step": 5780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004940633364102364,
+      "loss": 2.5121,
+      "step": 5790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004940428869226908,
+      "loss": 2.5473,
+      "step": 5800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004940224027000971,
+      "loss": 2.5438,
+      "step": 5810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004940018837453711,
+      "loss": 2.5331,
+      "step": 5820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004939813300614333,
+      "loss": 2.5063,
+      "step": 5830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004939607416512089,
+      "loss": 2.4819,
+      "step": 5840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004939401185176283,
+      "loss": 2.5368,
+      "step": 5850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004939194606636271,
+      "loss": 2.5501,
+      "step": 5860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004938987680921454,
+      "loss": 2.5764,
+      "step": 5870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004938780408061283,
+      "loss": 2.5278,
+      "step": 5880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004938572788085262,
+      "loss": 2.5153,
+      "step": 5890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000493836482102294,
+      "loss": 2.5262,
+      "step": 5900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004938156506903916,
+      "loss": 2.4982,
+      "step": 5910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004937947845757843,
+      "loss": 2.5665,
+      "step": 5920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004937738837614419,
+      "loss": 2.5907,
+      "step": 5930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004937529482503391,
+      "loss": 2.5652,
+      "step": 5940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004937319780454559,
+      "loss": 2.5768,
+      "step": 5950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004937109731497769,
+      "loss": 2.5628,
+      "step": 5960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004936899335662917,
+      "loss": 2.5346,
+      "step": 5970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004936688592979951,
+      "loss": 2.6023,
+      "step": 5980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004936477503478865,
+      "loss": 2.6156,
+      "step": 5990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004936266067189705,
+      "loss": 2.5852,
+      "step": 6000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004936054284142563,
+      "loss": 2.6043,
+      "step": 6010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004935842154367583,
+      "loss": 2.5359,
+      "step": 6020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004935629677894959,
+      "loss": 2.6143,
+      "step": 6030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004935416854754933,
+      "loss": 2.5196,
+      "step": 6040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004935203684977795,
+      "loss": 2.5343,
+      "step": 6050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004934990168593887,
+      "loss": 2.5528,
+      "step": 6060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00049347763056336,
+      "loss": 2.5252,
+      "step": 6070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004934562096127371,
+      "loss": 2.572,
+      "step": 6080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000493434754010569,
+      "loss": 2.5583,
+      "step": 6090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004934132637599097,
+      "loss": 2.5042,
+      "step": 6100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004933917388638177,
+      "loss": 2.5118,
+      "step": 6110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004933701793253568,
+      "loss": 2.534,
+      "step": 6120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004933485851475955,
+      "loss": 2.5335,
+      "step": 6130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004933269563336075,
+      "loss": 2.5667,
+      "step": 6140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004933052928864709,
+      "loss": 2.5293,
+      "step": 6150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004932835948092696,
+      "loss": 2.5083,
+      "step": 6160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004932618621050917,
+      "loss": 2.5234,
+      "step": 6170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004932400947770304,
+      "loss": 2.5295,
+      "step": 6180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000493218292828184,
+      "loss": 2.5481,
+      "step": 6190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004931964562616554,
+      "loss": 2.5662,
+      "step": 6200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004931745850805528,
+      "loss": 2.4865,
+      "step": 6210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004931526792879892,
+      "loss": 2.5432,
+      "step": 6220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004931307388870824,
+      "loss": 2.4741,
+      "step": 6230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004931087638809553,
+      "loss": 2.5365,
+      "step": 6240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004930867542727354,
+      "loss": 2.5459,
+      "step": 6250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004930647100655556,
+      "loss": 2.498,
+      "step": 6260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004930426312625536,
+      "loss": 2.5288,
+      "step": 6270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004930205178668716,
+      "loss": 2.551,
+      "step": 6280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004929983698816572,
+      "loss": 2.5706,
+      "step": 6290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004929761873100628,
+      "loss": 2.5562,
+      "step": 6300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004929539701552456,
+      "loss": 2.5513,
+      "step": 6310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004929317184203678,
+      "loss": 2.5474,
+      "step": 6320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004929094321085967,
+      "loss": 2.5221,
+      "step": 6330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004928871112231041,
+      "loss": 2.575,
+      "step": 6340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004928647557670672,
+      "loss": 2.5772,
+      "step": 6350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004928423657436676,
+      "loss": 2.523,
+      "step": 6360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004928199411560924,
+      "loss": 2.5617,
+      "step": 6370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004927974820075332,
+      "loss": 2.5132,
+      "step": 6380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004927749883011866,
+      "loss": 2.5358,
+      "step": 6390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004927524600402543,
+      "loss": 2.5094,
+      "step": 6400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004927298972279428,
+      "loss": 2.5867,
+      "step": 6410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004927072998674633,
+      "loss": 2.4521,
+      "step": 6420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004926846679620323,
+      "loss": 2.5832,
+      "step": 6430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004926620015148711,
+      "loss": 2.5095,
+      "step": 6440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004926393005292057,
+      "loss": 2.606,
+      "step": 6450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004926165650082671,
+      "loss": 2.5507,
+      "step": 6460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004925937949552914,
+      "loss": 2.539,
+      "step": 6470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004925709903735196,
+      "loss": 2.474,
+      "step": 6480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004925481512661975,
+      "loss": 2.5245,
+      "step": 6490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004925252776365755,
+      "loss": 2.5624,
+      "step": 6500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004925023694879097,
+      "loss": 2.5657,
+      "step": 6510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004924794268234603,
+      "loss": 2.5445,
+      "step": 6520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004924564496464931,
+      "loss": 2.4908,
+      "step": 6530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004924334379602781,
+      "loss": 2.5565,
+      "step": 6540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004924103917680909,
+      "loss": 2.6322,
+      "step": 6550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004923873110732115,
+      "loss": 2.5302,
+      "step": 6560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004923641958789252,
+      "loss": 2.5552,
+      "step": 6570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000492341046188522,
+      "loss": 2.5387,
+      "step": 6580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004923178620052966,
+      "loss": 2.5912,
+      "step": 6590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004922946433325491,
+      "loss": 2.5186,
+      "step": 6600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004922713901735842,
+      "loss": 2.6154,
+      "step": 6610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004922481025317115,
+      "loss": 2.499,
+      "step": 6620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004922247804102456,
+      "loss": 2.5363,
+      "step": 6630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004922037610236175,
+      "loss": 2.5905,
+      "step": 6640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004921803734000738,
+      "loss": 2.5299,
+      "step": 6650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004921569513065769,
+      "loss": 2.5492,
+      "step": 6660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004921334947464604,
+      "loss": 2.508,
+      "step": 6670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004921100037230632,
+      "loss": 2.4594,
+      "step": 6680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004920864782397285,
+      "loss": 2.5114,
+      "step": 6690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004920629182998049,
+      "loss": 2.5465,
+      "step": 6700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004920393239066459,
+      "loss": 2.5416,
+      "step": 6710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004920156950636093,
+      "loss": 2.5362,
+      "step": 6720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004919920317740584,
+      "loss": 2.4794,
+      "step": 6730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004919683340413616,
+      "loss": 2.4981,
+      "step": 6740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004919446018688913,
+      "loss": 2.5181,
+      "step": 6750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004919208352600256,
+      "loss": 2.5417,
+      "step": 6760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004918970342181473,
+      "loss": 2.5154,
+      "step": 6770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004918731987466439,
+      "loss": 2.5431,
+      "step": 6780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004918493288489081,
+      "loss": 2.5907,
+      "step": 6790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004918254245283372,
+      "loss": 2.5224,
+      "step": 6800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004918014857883335,
+      "loss": 2.5333,
+      "step": 6810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004917775126323045,
+      "loss": 2.574,
+      "step": 6820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004917535050636622,
+      "loss": 2.5394,
+      "step": 6830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004917294630858235,
+      "loss": 2.4887,
+      "step": 6840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004917053867022105,
+      "loss": 2.5104,
+      "step": 6850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00049168127591625,
+      "loss": 2.468,
+      "step": 6860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004916571307313737,
+      "loss": 2.5628,
+      "step": 6870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004916329511510183,
+      "loss": 2.5458,
+      "step": 6880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004916087371786251,
+      "loss": 2.4946,
+      "step": 6890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004915844888176409,
+      "loss": 2.5286,
+      "step": 6900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004915602060715168,
+      "loss": 2.5253,
+      "step": 6910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004915358889437088,
+      "loss": 2.4748,
+      "step": 6920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004915115374376784,
+      "loss": 2.5094,
+      "step": 6930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004914871515568913,
+      "loss": 2.5571,
+      "step": 6940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004914627313048185,
+      "loss": 2.5289,
+      "step": 6950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004914382766849357,
+      "loss": 2.5304,
+      "step": 6960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004914137877007237,
+      "loss": 2.4569,
+      "step": 6970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004913892643556678,
+      "loss": 2.537,
+      "step": 6980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004913647066532589,
+      "loss": 2.518,
+      "step": 6990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004913401145969919,
+      "loss": 2.5473,
+      "step": 7000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004913154881903673,
+      "loss": 2.5467,
+      "step": 7010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004912908274368902,
+      "loss": 2.5051,
+      "step": 7020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004912661323400703,
+      "loss": 2.5434,
+      "step": 7030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004912414029034229,
+      "loss": 2.5079,
+      "step": 7040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004912166391304676,
+      "loss": 2.5231,
+      "step": 7050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004911918410247289,
+      "loss": 2.5327,
+      "step": 7060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004911670085897367,
+      "loss": 2.4631,
+      "step": 7070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004911421418290253,
+      "loss": 2.4736,
+      "step": 7080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000491117240746134,
+      "loss": 2.5329,
+      "step": 7090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000491092305344607,
+      "loss": 2.5367,
+      "step": 7100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004910673356279934,
+      "loss": 2.4862,
+      "step": 7110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004910423315998472,
+      "loss": 2.5973,
+      "step": 7120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004910172932637274,
+      "loss": 2.476,
+      "step": 7130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004909922206231974,
+      "loss": 2.5796,
+      "step": 7140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004909671136818262,
+      "loss": 2.5209,
+      "step": 7150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004909419724431872,
+      "loss": 2.5255,
+      "step": 7160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004909167969108588,
+      "loss": 2.528,
+      "step": 7170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004908915870884241,
+      "loss": 2.5318,
+      "step": 7180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004908663429794715,
+      "loss": 2.5298,
+      "step": 7190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000490841064587594,
+      "loss": 2.5611,
+      "step": 7200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004908157519163893,
+      "loss": 2.5744,
+      "step": 7210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004907904049694605,
+      "loss": 2.4975,
+      "step": 7220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000490765023750415,
+      "loss": 2.5416,
+      "step": 7230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004907396082628656,
+      "loss": 2.5316,
+      "step": 7240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004907141585104296,
+      "loss": 2.4884,
+      "step": 7250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004906886744967293,
+      "loss": 2.4595,
+      "step": 7260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004906631562253918,
+      "loss": 2.6201,
+      "step": 7270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004906376037000494,
+      "loss": 2.5497,
+      "step": 7280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004906120169243389,
+      "loss": 2.5548,
+      "step": 7290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000490586395901902,
+      "loss": 2.5192,
+      "step": 7300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004905607406363856,
+      "loss": 2.4508,
+      "step": 7310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000490535051131441,
+      "loss": 2.5351,
+      "step": 7320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000490509327390725,
+      "loss": 2.5056,
+      "step": 7330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004904835694178986,
+      "loss": 2.5102,
+      "step": 7340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004904577772166281,
+      "loss": 2.5897,
+      "step": 7350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004904319507905844,
+      "loss": 2.5587,
+      "step": 7360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004904060901434436,
+      "loss": 2.5699,
+      "step": 7370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004903801952788865,
+      "loss": 2.4632,
+      "step": 7380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004903542662005986,
+      "loss": 2.5195,
+      "step": 7390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004903309007804499,
+      "loss": 2.4926,
+      "step": 7400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004903049067062451,
+      "loss": 2.5151,
+      "step": 7410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004902788784290256,
+      "loss": 2.5064,
+      "step": 7420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004902528159524959,
+      "loss": 2.5318,
+      "step": 7430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004902267192803657,
+      "loss": 2.4508,
+      "step": 7440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004902005884163492,
+      "loss": 2.5051,
+      "step": 7450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004901744233641659,
+      "loss": 2.574,
+      "step": 7460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004901482241275396,
+      "loss": 2.5474,
+      "step": 7470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004901219907101995,
+      "loss": 2.5259,
+      "step": 7480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004900957231158793,
+      "loss": 2.5176,
+      "step": 7490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000490069421348318,
+      "loss": 2.4825,
+      "step": 7500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004900430854112589,
+      "loss": 2.5205,
+      "step": 7510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004900167153084504,
+      "loss": 2.5239,
+      "step": 7520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004899903110436461,
+      "loss": 2.5319,
+      "step": 7530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004899638726206039,
+      "loss": 2.548,
+      "step": 7540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000489937400043087,
+      "loss": 2.5479,
+      "step": 7550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000489910893314863,
+      "loss": 2.5116,
+      "step": 7560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000489884352439705,
+      "loss": 2.5425,
+      "step": 7570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004898577774213905,
+      "loss": 2.5161,
+      "step": 7580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004898311682637018,
+      "loss": 2.4576,
+      "step": 7590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004898045249704264,
+      "loss": 2.4789,
+      "step": 7600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004897778475453565,
+      "loss": 2.548,
+      "step": 7610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004897511359922891,
+      "loss": 2.4826,
+      "step": 7620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000489724390315026,
+      "loss": 2.5211,
+      "step": 7630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004896976105173742,
+      "loss": 2.5603,
+      "step": 7640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000489670796603145,
+      "loss": 2.5257,
+      "step": 7650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004896439485761551,
+      "loss": 2.5226,
+      "step": 7660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004896170664402257,
+      "loss": 2.5378,
+      "step": 7670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004895901501991831,
+      "loss": 2.5309,
+      "step": 7680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004895631998568583,
+      "loss": 2.5318,
+      "step": 7690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004895362154170871,
+      "loss": 2.5586,
+      "step": 7700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004895091968837105,
+      "loss": 2.5026,
+      "step": 7710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004894821442605736,
+      "loss": 2.5712,
+      "step": 7720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004894550575515273,
+      "loss": 2.5526,
+      "step": 7730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004894279367604268,
+      "loss": 2.5771,
+      "step": 7740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004894007818911321,
+      "loss": 2.5358,
+      "step": 7750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004893735929475083,
+      "loss": 2.5122,
+      "step": 7760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004893463699334254,
+      "loss": 2.5698,
+      "step": 7770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004893191128527577,
+      "loss": 2.5597,
+      "step": 7780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004892918217093852,
+      "loss": 2.5425,
+      "step": 7790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004892644965071919,
+      "loss": 2.596,
+      "step": 7800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004892371372500674,
+      "loss": 2.5331,
+      "step": 7810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004892097439419057,
+      "loss": 2.5161,
+      "step": 7820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004891823165866055,
+      "loss": 2.5049,
+      "step": 7830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004891548551880706,
+      "loss": 2.5163,
+      "step": 7840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004891273597502101,
+      "loss": 2.5393,
+      "step": 7850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004890998302769369,
+      "loss": 2.5318,
+      "step": 7860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004890722667721697,
+      "loss": 2.4846,
+      "step": 7870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004890446692398315,
+      "loss": 2.5587,
+      "step": 7880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004890170376838503,
+      "loss": 2.4969,
+      "step": 7890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000488989372108159,
+      "loss": 2.5095,
+      "step": 7900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004889616725166951,
+      "loss": 2.4844,
+      "step": 7910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004889339389134016,
+      "loss": 2.5708,
+      "step": 7920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004889061713022253,
+      "loss": 2.5844,
+      "step": 7930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004888783696871188,
+      "loss": 2.5585,
+      "step": 7940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004888505340720391,
+      "loss": 2.5356,
+      "step": 7950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004888226644609478,
+      "loss": 2.4999,
+      "step": 7960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000488794760857812,
+      "loss": 2.5001,
+      "step": 7970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004887668232666032,
+      "loss": 2.5503,
+      "step": 7980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004887388516912976,
+      "loss": 2.5694,
+      "step": 7990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004887108461358767,
+      "loss": 2.5327,
+      "step": 8000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004886828066043263,
+      "loss": 2.5373,
+      "step": 8010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004886547331006375,
+      "loss": 2.5168,
+      "step": 8020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004886266256288062,
+      "loss": 2.5754,
+      "step": 8030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004885984841928326,
+      "loss": 2.5211,
+      "step": 8040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004885703087967224,
+      "loss": 2.5467,
+      "step": 8050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004885420994444858,
+      "loss": 2.5561,
+      "step": 8060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004885138561401377,
+      "loss": 2.5086,
+      "step": 8070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004884855788876984,
+      "loss": 2.5144,
+      "step": 8080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004884572676911923,
+      "loss": 2.4998,
+      "step": 8090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004884289225546491,
+      "loss": 2.4899,
+      "step": 8100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004884005434821032,
+      "loss": 2.5413,
+      "step": 8110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000488372130477594,
+      "loss": 2.4236,
+      "step": 8120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004883436835451654,
+      "loss": 2.5394,
+      "step": 8130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004883152026888664,
+      "loss": 2.5144,
+      "step": 8140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004882866879127505,
+      "loss": 2.5319,
+      "step": 8150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048825813922087646,
+      "loss": 2.4748,
+      "step": 8160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004882295566173077,
+      "loss": 2.5425,
+      "step": 8170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048820094010611236,
+      "loss": 2.4608,
+      "step": 8180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004881722896913635,
+      "loss": 2.5557,
+      "step": 8190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004881436053771389,
+      "loss": 2.5616,
+      "step": 8200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004881148871675214,
+      "loss": 2.5322,
+      "step": 8210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004880861350665983,
+      "loss": 2.5417,
+      "step": 8220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004880573490784622,
+      "loss": 2.5991,
+      "step": 8230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048802852920721,
+      "loss": 2.429,
+      "step": 8240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004879996754569439,
+      "loss": 2.513,
+      "step": 8250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048797078783177053,
+      "loss": 2.5191,
+      "step": 8260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048794186633580164,
+      "loss": 2.5401,
+      "step": 8270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048791291097315364,
+      "loss": 2.5169,
+      "step": 8280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004878839217479478,
+      "loss": 2.5629,
+      "step": 8290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048785489866431014,
+      "loss": 2.4349,
+      "step": 8300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048782584172637165,
+      "loss": 2.5095,
+      "step": 8310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048779675093826803,
+      "loss": 2.4796,
+      "step": 8320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048776762630413985,
+      "loss": 2.5365,
+      "step": 8330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048773846782813234,
+      "loss": 2.5191,
+      "step": 8340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048770927551439593,
+      "loss": 2.5618,
+      "step": 8350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048768004936708534,
+      "loss": 2.5195,
+      "step": 8360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004876507893903605,
+      "loss": 2.5409,
+      "step": 8370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048762149558838607,
+      "loss": 2.4873,
+      "step": 8380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048759216796533144,
+      "loss": 2.5492,
+      "step": 8390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004875628065253709,
+      "loss": 2.5033,
+      "step": 8400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048753341127268334,
+      "loss": 2.5711,
+      "step": 8410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048750398221145293,
+      "loss": 2.538,
+      "step": 8420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004874745193458682,
+      "loss": 2.5147,
+      "step": 8430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048744502268012257,
+      "loss": 2.4645,
+      "step": 8440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004874154922184145,
+      "loss": 2.568,
+      "step": 8450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004873859279649471,
+      "loss": 2.5086,
+      "step": 8460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004873563299239281,
+      "loss": 2.4851,
+      "step": 8470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048732669809957056,
+      "loss": 2.5279,
+      "step": 8480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004872970324960918,
+      "loss": 2.5428,
+      "step": 8490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004872673331177142,
+      "loss": 2.5061,
+      "step": 8500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048723759996866504,
+      "loss": 2.5048,
+      "step": 8510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048720783305317616,
+      "loss": 2.4591,
+      "step": 8520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048717803237548446,
+      "loss": 2.5629,
+      "step": 8530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048714819793983143,
+      "loss": 2.4768,
+      "step": 8540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048711832975046346,
+      "loss": 2.5034,
+      "step": 8550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048708842781163176,
+      "loss": 2.5183,
+      "step": 8560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004870584921275923,
+      "loss": 2.518,
+      "step": 8570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000487028522702606,
+      "loss": 2.5804,
+      "step": 8580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004869985195409383,
+      "loss": 2.5195,
+      "step": 8590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004869684826468597,
+      "loss": 2.4545,
+      "step": 8600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004869384120246454,
+      "loss": 2.4323,
+      "step": 8610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004869083076785753,
+      "loss": 2.4878,
+      "step": 8620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048687816961293437,
+      "loss": 2.5676,
+      "step": 8630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048684799783201205,
+      "loss": 2.4564,
+      "step": 8640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048681779234010283,
+      "loss": 2.5636,
+      "step": 8650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048678755314150593,
+      "loss": 2.5314,
+      "step": 8660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048675728024052527,
+      "loss": 2.5547,
+      "step": 8670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048672697364146974,
+      "loss": 2.5618,
+      "step": 8680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004866966333486528,
+      "loss": 2.5123,
+      "step": 8690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048666625936639297,
+      "loss": 2.5665,
+      "step": 8700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048663585169901336,
+      "loss": 2.5149,
+      "step": 8710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048660541035084205,
+      "loss": 2.5337,
+      "step": 8720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004865749353262116,
+      "loss": 2.4607,
+      "step": 8730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048654442662945973,
+      "loss": 2.5306,
+      "step": 8740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004865138842649288,
+      "loss": 2.4574,
+      "step": 8750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048648330823696586,
+      "loss": 2.4983,
+      "step": 8760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048645269854992294,
+      "loss": 2.5543,
+      "step": 8770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048642205520815676,
+      "loss": 2.583,
+      "step": 8780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048639137821602883,
+      "loss": 2.4791,
+      "step": 8790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004863606675779054,
+      "loss": 2.542,
+      "step": 8800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004863299232981576,
+      "loss": 2.4648,
+      "step": 8810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004862991453811614,
+      "loss": 2.4428,
+      "step": 8820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048626833383129744,
+      "loss": 2.5686,
+      "step": 8830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048623748865295104,
+      "loss": 2.5727,
+      "step": 8840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048620660985051263,
+      "loss": 2.5765,
+      "step": 8850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048617569742837713,
+      "loss": 2.4668,
+      "step": 8860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048614475139094446,
+      "loss": 2.5309,
+      "step": 8870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004861137717426191,
+      "loss": 2.573,
+      "step": 8880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004860827584878105,
+      "loss": 2.513,
+      "step": 8890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004860517116309329,
+      "loss": 2.5,
+      "step": 8900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004860206311764052,
+      "loss": 2.5337,
+      "step": 8910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048598951712865103,
+      "loss": 2.5854,
+      "step": 8920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048595836949209896,
+      "loss": 2.5265,
+      "step": 8930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00048592718827118243,
+      "loss": 2.5244,
+      "step": 8940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004858959734703393,
+      "loss": 2.5271,
+      "step": 8950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048586472509401263,
+      "loss": 2.5153,
+      "step": 8960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048583344314664993,
+      "loss": 2.4984,
+      "step": 8970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004858021276327037,
+      "loss": 2.5511,
+      "step": 8980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000485770778556631,
+      "loss": 2.5409,
+      "step": 8990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004857393959228939,
+      "loss": 2.5153,
+      "step": 9000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004857079797359591,
+      "loss": 2.593,
+      "step": 9010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004856765300002982,
+      "loss": 2.5274,
+      "step": 9020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004856450467203875,
+      "loss": 2.521,
+      "step": 9030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004856135299007078,
+      "loss": 2.5454,
+      "step": 9040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004855819795457453,
+      "loss": 2.4956,
+      "step": 9050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004855503956599904,
+      "loss": 2.4448,
+      "step": 9060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004855187782479386,
+      "loss": 2.5293,
+      "step": 9070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048548712731409,
+      "loss": 2.5492,
+      "step": 9080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004854554428629496,
+      "loss": 2.5481,
+      "step": 9090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000485423724899027,
+      "loss": 2.4744,
+      "step": 9100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004853919734268367,
+      "loss": 2.489,
+      "step": 9110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000485360188450898,
+      "loss": 2.5009,
+      "step": 9120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048532836997573493,
+      "loss": 2.5212,
+      "step": 9130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048529651800587617,
+      "loss": 2.5175,
+      "step": 9140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004852646325458553,
+      "loss": 2.5256,
+      "step": 9150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048523271360021065,
+      "loss": 2.4885,
+      "step": 9160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048520076117348533,
+      "loss": 2.5278,
+      "step": 9170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048516877527022704,
+      "loss": 2.5603,
+      "step": 9180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048513675589498864,
+      "loss": 2.5113,
+      "step": 9190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048510470305232733,
+      "loss": 2.4844,
+      "step": 9200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048507261674680523,
+      "loss": 2.4827,
+      "step": 9210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004850404969829894,
+      "loss": 2.5388,
+      "step": 9220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004850083437654512,
+      "loss": 2.5186,
+      "step": 9230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004849761570987674,
+      "loss": 2.5464,
+      "step": 9240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048494393698751904,
+      "loss": 2.57,
+      "step": 9250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048491168343629197,
+      "loss": 2.5314,
+      "step": 9260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000484879396449677,
+      "loss": 2.4585,
+      "step": 9270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048484707603226957,
+      "loss": 2.5167,
+      "step": 9280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004848147221886699,
+      "loss": 2.528,
+      "step": 9290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000484782334923483,
+      "loss": 2.4785,
+      "step": 9300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048474991424131853,
+      "loss": 2.5506,
+      "step": 9310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004847174601467911,
+      "loss": 2.5288,
+      "step": 9320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004846849726445198,
+      "loss": 2.4625,
+      "step": 9330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00048465245173912876,
+      "loss": 2.5387,
+      "step": 9340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004846198974352467,
+      "loss": 2.5236,
+      "step": 9350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004845873097375071,
+      "loss": 2.6822,
+      "step": 9360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048455468865054826,
+      "loss": 2.5064,
+      "step": 9370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048452203417901317,
+      "loss": 2.4583,
+      "step": 9380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004844893463275496,
+      "loss": 2.493,
+      "step": 9390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004844566251008101,
+      "loss": 2.5302,
+      "step": 9400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004844238705034519,
+      "loss": 2.5663,
+      "step": 9410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048439108254013714,
+      "loss": 2.4936,
+      "step": 9420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004843582612155324,
+      "loss": 2.4725,
+      "step": 9430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004843254065343093,
+      "loss": 2.5412,
+      "step": 9440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004842925185011441,
+      "loss": 2.5752,
+      "step": 9450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048425959712071774,
+      "loss": 2.5842,
+      "step": 9460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048422664239771606,
+      "loss": 2.5258,
+      "step": 9470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004841936543368296,
+      "loss": 2.5472,
+      "step": 9480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048416063294275347,
+      "loss": 2.4784,
+      "step": 9490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004841275782201878,
+      "loss": 2.5275,
+      "step": 9500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048409449017383725,
+      "loss": 2.5766,
+      "step": 9510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048406136880841126,
+      "loss": 2.4899,
+      "step": 9520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004840282141286242,
+      "loss": 2.5405,
+      "step": 9530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048399502613919484,
+      "loss": 2.5145,
+      "step": 9540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048396180484484706,
+      "loss": 2.5267,
+      "step": 9550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048392855025030914,
+      "loss": 2.5416,
+      "step": 9560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048389526236031444,
+      "loss": 2.5031,
+      "step": 9570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004838619411796007,
+      "loss": 2.486,
+      "step": 9580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048382858671291073,
+      "loss": 2.4852,
+      "step": 9590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004837951989649919,
+      "loss": 2.5473,
+      "step": 9600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004837617779405963,
+      "loss": 2.5644,
+      "step": 9610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048372832364448084,
+      "loss": 2.4929,
+      "step": 9620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004836948360814071,
+      "loss": 2.5748,
+      "step": 9630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048366131525614137,
+      "loss": 2.4606,
+      "step": 9640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048362776117345487,
+      "loss": 2.5153,
+      "step": 9650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004835941738381233,
+      "loss": 2.5266,
+      "step": 9660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004835605532549272,
+      "loss": 2.5345,
+      "step": 9670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048352689942865196,
+      "loss": 2.5751,
+      "step": 9680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004834932123640875,
+      "loss": 2.5734,
+      "step": 9690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004834594920660286,
+      "loss": 2.5082,
+      "step": 9700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048342573853927455,
+      "loss": 2.4498,
+      "step": 9710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048339195178862985,
+      "loss": 2.5814,
+      "step": 9720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004833581318189032,
+      "loss": 2.5311,
+      "step": 9730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048332427863490837,
+      "loss": 2.5387,
+      "step": 9740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048329039224146364,
+      "loss": 2.529,
+      "step": 9750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004832564726433922,
+      "loss": 2.5404,
+      "step": 9760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00048322251984552183,
+      "loss": 2.5415,
+      "step": 9770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004831885338526852,
+      "loss": 2.4912,
+      "step": 9780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004831545146697195,
+      "loss": 2.4882,
+      "step": 9790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004831204623014667,
+      "loss": 2.5481,
+      "step": 9800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00048308637675277365,
+      "loss": 2.5279,
+      "step": 9810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00048305225802849173,
+      "loss": 2.5243,
+      "step": 9820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004830181061334771,
+      "loss": 2.5333,
+      "step": 9830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004829839210725908,
+      "loss": 2.5383,
+      "step": 9840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00048294970285069826,
+      "loss": 2.5704,
+      "step": 9850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004829154514726699,
+      "loss": 2.5403,
+      "step": 9860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004828811669433808,
+      "loss": 2.535,
+      "step": 9870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004828468492677107,
+      "loss": 2.4692,
+      "step": 9880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00048281249845054413,
+      "loss": 2.5308,
+      "step": 9890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004827781144967703,
+      "loss": 2.4852,
+      "step": 9900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004827436974112831,
+      "loss": 2.4735,
+      "step": 9910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004827092471989811,
+      "loss": 2.5586,
+      "step": 9920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00048267476386476793,
+      "loss": 2.5126,
+      "step": 9930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004826402474135514,
+      "loss": 2.5112,
+      "step": 9940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004826056978502443,
+      "loss": 2.5092,
+      "step": 9950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004825711151797643,
+      "loss": 2.5596,
+      "step": 9960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004825364994070335,
+      "loss": 2.5572,
+      "step": 9970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00048250185053697893,
+      "loss": 2.5777,
+      "step": 9980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00048246716857453203,
+      "loss": 2.536,
+      "step": 9990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004824324535246293,
+      "loss": 2.5377,
+      "step": 10000
+    }
+  ],
+  "max_steps": 83272,
+  "num_train_epochs": 2,
+  "total_flos": 8.253517770868654e+18,
+  "trial_name": null,
+  "trial_params": null
+}