diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4819 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6371202166208737,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 3.5365,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 3.5489,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 3.5323,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 3.4815,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 3.5322,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 3.4405,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 3.4819,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 3.4797,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 3.4162,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 3.3748,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 3.3559,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 3.3819,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 3.2869,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-05,
+      "loss": 3.3594,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.2526,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 3.2382,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-05,
+      "loss": 3.2051,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-05,
+      "loss": 3.1896,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.5e-05,
+      "loss": 3.1497,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 3.1312,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999983838423555e-05,
+      "loss": 3.1318,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999935353798693e-05,
+      "loss": 3.1844,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999854546438851e-05,
+      "loss": 3.1143,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99974141686642e-05,
+      "loss": 3.1435,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99959596581274e-05,
+      "loss": 3.1271,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9994181942181e-05,
+      "loss": 3.1595,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999208103231724e-05,
+      "loss": 3.0622,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998965694211775e-05,
+      "loss": 3.1363,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998690968725338e-05,
+      "loss": 3.0925,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998383928548411e-05,
+      "loss": 3.1098,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998044575665898e-05,
+      "loss": 3.0814,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997672912271582e-05,
+      "loss": 3.0462,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99726894076814e-05,
+      "loss": 3.1178,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996832663767092e-05,
+      "loss": 3.0232,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996364084088809e-05,
+      "loss": 3.1105,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995863204762486e-05,
+      "loss": 3.0949,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995330029026123e-05,
+      "loss": 3.0649,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994764560326505e-05,
+      "loss": 3.0192,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994166802319173e-05,
+      "loss": 3.0056,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99353675886842e-05,
+      "loss": 3.045,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992874434047237e-05,
+      "loss": 3.0226,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992179832137314e-05,
+      "loss": 3.017,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991452957628993e-05,
+      "loss": 3.0404,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99069381522125e-05,
+      "loss": 3.0087,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989902409821661e-05,
+      "loss": 3.0128,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98907874654637e-05,
+      "loss": 3.0279,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988222830720051e-05,
+      "loss": 2.9837,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987334667875891e-05,
+      "loss": 2.9475,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98641426375553e-05,
+      "loss": 2.9592,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985461624309041e-05,
+      "loss": 2.9297,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984476755694887e-05,
+      "loss": 2.9845,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983459664279879e-05,
+      "loss": 2.9412,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.982410356639138e-05,
+      "loss": 2.9676,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98132883955605e-05,
+      "loss": 2.9679,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980215120022223e-05,
+      "loss": 2.8977,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979069205237444e-05,
+      "loss": 3.0213,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977891102609626e-05,
+      "loss": 2.8899,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976680819754768e-05,
+      "loss": 2.9723,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975438364496904e-05,
+      "loss": 2.9178,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974163744868046e-05,
+      "loss": 2.9468,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.97285696910814e-05,
+      "loss": 2.908,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.971518045665009e-05,
+      "loss": 2.9128,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.970146983194295e-05,
+      "loss": 2.9149,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.968743790559415e-05,
+      "loss": 2.9707,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.967308476831488e-05,
+      "loss": 2.9443,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.965841051289288e-05,
+      "loss": 2.9446,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96434152341918e-05,
+      "loss": 2.916,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.962809902915055e-05,
+      "loss": 2.9602,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.961246199678278e-05,
+      "loss": 2.9286,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.959650423817607e-05,
+      "loss": 2.8384,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958022585649149e-05,
+      "loss": 2.8756,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.956362695696274e-05,
+      "loss": 2.915,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.954670764689557e-05,
+      "loss": 2.8671,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.952946803566707e-05,
+      "loss": 2.88,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.951190823472497e-05,
+      "loss": 2.91,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.949402835758687e-05,
+      "loss": 2.8586,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.947582851983961e-05,
+      "loss": 2.906,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.945730883913838e-05,
+      "loss": 2.8762,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.94384694352061e-05,
+      "loss": 2.8768,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.941931042983252e-05,
+      "loss": 2.8941,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.939983194687359e-05,
+      "loss": 2.8198,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.938003411225046e-05,
+      "loss": 2.9177,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.935991705394883e-05,
+      "loss": 2.8462,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.933948090201807e-05,
+      "loss": 2.842,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.931872578857032e-05,
+      "loss": 2.8195,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.929765184777974e-05,
+      "loss": 2.8787,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.927625921588156e-05,
+      "loss": 2.8872,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.925454803117126e-05,
+      "loss": 2.8638,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.923251843400361e-05,
+      "loss": 2.9058,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.921017056679183e-05,
+      "loss": 2.8653,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.918750457400662e-05,
+      "loss": 2.8621,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.916452060217528e-05,
+      "loss": 2.7918,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.914121879988063e-05,
+      "loss": 2.8001,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.911759931776028e-05,
+      "loss": 2.8077,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.909366230850541e-05,
+      "loss": 2.8251,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.906940792685997e-05,
+      "loss": 2.8582,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.904483632961957e-05,
+      "loss": 2.8479,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.90199476756305e-05,
+      "loss": 2.8326,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.899474212578873e-05,
+      "loss": 2.8315,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.896921984303881e-05,
+      "loss": 2.8598,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.89433809923729e-05,
+      "loss": 2.8537,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.891722574082958e-05,
+      "loss": 2.8966,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.889075425749293e-05,
+      "loss": 2.8141,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.886396671349127e-05,
+      "loss": 2.8146,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.883686328199622e-05,
+      "loss": 2.7635,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.880944413822142e-05,
+      "loss": 2.8232,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.878170945942151e-05,
+      "loss": 2.7775,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.875365942489096e-05,
+      "loss": 2.7953,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.872529421596286e-05,
+      "loss": 2.795,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.869661401600783e-05,
+      "loss": 2.8619,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.866761901043274e-05,
+      "loss": 2.7735,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.863830938667962e-05,
+      "loss": 2.7792,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.860868533422434e-05,
+      "loss": 2.7439,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.857874704457544e-05,
+      "loss": 2.7739,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.854849471127294e-05,
+      "loss": 2.878,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.851792852988698e-05,
+      "loss": 2.8094,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.848704869801663e-05,
+      "loss": 2.7561,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.845585541528859e-05,
+      "loss": 2.7466,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.842434888335593e-05,
+      "loss": 2.7493,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.839252930589671e-05,
+      "loss": 2.7701,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.836039688861277e-05,
+      "loss": 2.8069,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.832795183922831e-05,
+      "loss": 2.7478,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.829519436748857e-05,
+      "loss": 2.7962,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.826212468515854e-05,
+      "loss": 2.7282,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.822874300602148e-05,
+      "loss": 2.7446,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.819504954587759e-05,
+      "loss": 2.8204,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.816104452254267e-05,
+      "loss": 2.7895,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.812672815584664e-05,
+      "loss": 2.7493,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.809210066763212e-05,
+      "loss": 2.7641,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.805716228175302e-05,
+      "loss": 2.8104,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.802191322407312e-05,
+      "loss": 2.7556,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.798635372246455e-05,
+      "loss": 2.7818,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.795048400680634e-05,
+      "loss": 2.7879,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.791430430898294e-05,
+      "loss": 2.771,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.787781486288276e-05,
+      "loss": 2.7679,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.784101590439659e-05,
+      "loss": 2.7572,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.78039076714161e-05,
+      "loss": 2.7963,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.776649040383226e-05,
+      "loss": 2.7851,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.772876434353395e-05,
+      "loss": 2.717,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.769072973440616e-05,
+      "loss": 2.7112,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.765238682232862e-05,
+      "loss": 2.7207,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.761373585517408e-05,
+      "loss": 2.7411,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.757477708280678e-05,
+      "loss": 2.7476,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.753551075708076e-05,
+      "loss": 2.7371,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.749593713183833e-05,
+      "loss": 2.7222,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.745605646290836e-05,
+      "loss": 2.7758,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.741586900810463e-05,
+      "loss": 2.7283,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.737537502722421e-05,
+      "loss": 2.6644,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.733457478204569e-05,
+      "loss": 2.7189,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.729346853632762e-05,
+      "loss": 2.7745,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.725205655580668e-05,
+      "loss": 2.6977,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.721033910819602e-05,
+      "loss": 2.7049,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.716831646318353e-05,
+      "loss": 2.6668,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.71259888924301e-05,
+      "loss": 2.7684,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.70833566695678e-05,
+      "loss": 2.6741,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.704042007019825e-05,
+      "loss": 2.7063,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.699717937189067e-05,
+      "loss": 2.6991,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.69536348541802e-05,
+      "loss": 2.7038,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.690978679856611e-05,
+      "loss": 2.7071,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.686563548850981e-05,
+      "loss": 2.7013,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.682118120943324e-05,
+      "loss": 2.7614,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.677642424871691e-05,
+      "loss": 2.6723,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.673136489569801e-05,
+      "loss": 2.6807,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.668600344166864e-05,
+      "loss": 2.6753,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.66403401798738e-05,
+      "loss": 2.7456,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.659437540550965e-05,
+      "loss": 2.6838,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.654810941572145e-05,
+      "loss": 2.7185,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.650154250960175e-05,
+      "loss": 2.7493,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.645467498818838e-05,
+      "loss": 2.6943,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.640750715446258e-05,
+      "loss": 2.7067,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.636003931334692e-05,
+      "loss": 2.6662,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.63122717717035e-05,
+      "loss": 2.7566,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.626420483833181e-05,
+      "loss": 2.616,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.621583882396683e-05,
+      "loss": 2.6537,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.616717404127697e-05,
+      "loss": 2.6933,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.611821080486207e-05,
+      "loss": 2.7639,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.606894943125138e-05,
+      "loss": 2.6798,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.601939023890146e-05,
+      "loss": 2.6854,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.596953354819418e-05,
+      "loss": 2.6772,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.591937968143464e-05,
+      "loss": 2.6723,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.586892896284907e-05,
+      "loss": 2.6909,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.58181817185827e-05,
+      "loss": 2.6651,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.576713827669774e-05,
+      "loss": 2.6508,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.571579896717118e-05,
+      "loss": 2.7314,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.566416412189268e-05,
+      "loss": 2.7292,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.561223407466246e-05,
+      "loss": 2.6193,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.556000916118907e-05,
+      "loss": 2.6594,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.55074897190873e-05,
+      "loss": 2.6729,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.545467608787592e-05,
+      "loss": 2.7101,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.540156860897558e-05,
+      "loss": 2.6037,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.534816762570647e-05,
+      "loss": 2.7042,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.529447348328626e-05,
+      "loss": 2.7188,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.524048652882771e-05,
+      "loss": 2.689,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.518620711133655e-05,
+      "loss": 2.6767,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.513163558170916e-05,
+      "loss": 2.6237,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.507677229273035e-05,
+      "loss": 2.7055,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.502161759907096e-05,
+      "loss": 2.7101,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.496617185728575e-05,
+      "loss": 2.6341,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.491043542581095e-05,
+      "loss": 2.7266,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.485440866496197e-05,
+      "loss": 2.6916,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.479809193693118e-05,
+      "loss": 2.7151,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.474148560578538e-05,
+      "loss": 2.6063,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.468459003746357e-05,
+      "loss": 2.685,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.462740559977463e-05,
+      "loss": 2.6169,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.45699326623948e-05,
+      "loss": 2.6045,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.451217159686541e-05,
+      "loss": 2.7109,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.445412277659037e-05,
+      "loss": 2.6733,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.439578657683388e-05,
+      "loss": 2.6325,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.433716337471793e-05,
+      "loss": 2.6799,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.427825354921984e-05,
+      "loss": 2.6857,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.42190574811699e-05,
+      "loss": 2.6742,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.415957555324881e-05,
+      "loss": 2.688,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.409980814998524e-05,
+      "loss": 2.6574,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.403975565775341e-05,
+      "loss": 2.617,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.397941846477047e-05,
+      "loss": 2.7373,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.391879696109409e-05,
+      "loss": 2.6417,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.38578915386199e-05,
+      "loss": 2.686,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.379670259107895e-05,
+      "loss": 2.5881,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.373523051403518e-05,
+      "loss": 2.6778,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.367347570488287e-05,
+      "loss": 2.6912,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.361143856284405e-05,
+      "loss": 2.6565,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.35491194889659e-05,
+      "loss": 2.658,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.348651888611824e-05,
+      "loss": 2.6208,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.34236371589908e-05,
+      "loss": 2.638,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.336047471409078e-05,
+      "loss": 2.5566,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.329703195974e-05,
+      "loss": 2.6556,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.323330930607246e-05,
+      "loss": 2.6651,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.316930716503154e-05,
+      "loss": 2.6275,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.310502595036749e-05,
+      "loss": 2.6551,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.304046607763457e-05,
+      "loss": 2.6484,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.297562796418853e-05,
+      "loss": 2.6426,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.29105120291838e-05,
+      "loss": 2.6211,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.284511869357086e-05,
+      "loss": 2.6319,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.27794483800935e-05,
+      "loss": 2.6186,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.271350151328597e-05,
+      "loss": 2.6527,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.264727851947045e-05,
+      "loss": 2.625,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.25807798267541e-05,
+      "loss": 2.5686,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.251400586502643e-05,
+      "loss": 2.5855,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.244695706595644e-05,
+      "loss": 2.5983,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.23796338629898e-05,
+      "loss": 2.6167,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.231203669134618e-05,
+      "loss": 2.6019,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.224416598801633e-05,
+      "loss": 2.5794,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.217602219175927e-05,
+      "loss": 2.6468,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.210760574309944e-05,
+      "loss": 2.6471,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.203891708432396e-05,
+      "loss": 2.6195,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.196995665947958e-05,
+      "loss": 2.6508,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.190072491437e-05,
+      "loss": 2.6235,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.183122229655286e-05,
+      "loss": 2.5889,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.176144925533693e-05,
+      "loss": 2.6082,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.169140624177914e-05,
+      "loss": 2.5701,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.162109370868167e-05,
+      "loss": 2.6674,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.15505121105891e-05,
+      "loss": 2.6163,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.147966190378539e-05,
+      "loss": 2.6116,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.140854354629093e-05,
+      "loss": 2.6212,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.133715749785965e-05,
+      "loss": 2.613,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.126550421997598e-05,
+      "loss": 2.6468,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.119358417585189e-05,
+      "loss": 2.6875,
+      "step": 1285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.112139783042389e-05,
+      "loss": 2.5981,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.104894565035002e-05,
+      "loss": 2.6089,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.097622810400688e-05,
+      "loss": 2.5993,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.090324566148655e-05,
+      "loss": 2.5949,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.082999879459355e-05,
+      "loss": 2.658,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.075648797684179e-05,
+      "loss": 2.6,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.068271368345162e-05,
+      "loss": 2.6392,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.060867639134651e-05,
+      "loss": 2.594,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.053437657915027e-05,
+      "loss": 2.5977,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.045981472718369e-05,
+      "loss": 2.6188,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.038499131746162e-05,
+      "loss": 2.5971,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.030990683368976e-05,
+      "loss": 2.6217,
+      "step": 1345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.023456176126156e-05,
+      "loss": 2.6033,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.015895658725509e-05,
+      "loss": 2.6047,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.008309180042985e-05,
+      "loss": 2.6363,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.000696789122366e-05,
+      "loss": 2.6364,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.993058535174948e-05,
+      "loss": 2.5676,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.985394467579222e-05,
+      "loss": 2.7228,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.977704635880553e-05,
+      "loss": 2.6132,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.969989089790862e-05,
+      "loss": 2.5977,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.962247879188303e-05,
+      "loss": 2.5762,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.954481054116946e-05,
+      "loss": 2.6887,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.946688664786442e-05,
+      "loss": 2.6466,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.938870761571711e-05,
+      "loss": 2.5853,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.93102739501261e-05,
+      "loss": 2.573,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.923158615813605e-05,
+      "loss": 2.5942,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.915264474843448e-05,
+      "loss": 2.5571,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.907345023134844e-05,
+      "loss": 2.637,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.899400311884121e-05,
+      "loss": 2.574,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.891430392450902e-05,
+      "loss": 2.6359,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.883435316357775e-05,
+      "loss": 2.6832,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.87541513528995e-05,
+      "loss": 2.5538,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.86736990109494e-05,
+      "loss": 2.5753,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.859299665782205e-05,
+      "loss": 2.6304,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.851204481522839e-05,
+      "loss": 2.6201,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.843084400649217e-05,
+      "loss": 2.4845,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.834939475654663e-05,
+      "loss": 2.6418,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.826769759193105e-05,
+      "loss": 2.582,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.818575304078747e-05,
+      "loss": 2.6164,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.810356163285709e-05,
+      "loss": 2.5092,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.802112389947703e-05,
+      "loss": 2.5895,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.79384403735768e-05,
+      "loss": 2.5769,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.785551158967479e-05,
+      "loss": 2.6272,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.7772338083875e-05,
+      "loss": 2.5754,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.768892039386338e-05,
+      "loss": 2.5814,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.760525905890454e-05,
+      "loss": 2.5858,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.752135461983806e-05,
+      "loss": 2.561,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.743720761907514e-05,
+      "loss": 2.5657,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.735281860059506e-05,
+      "loss": 2.5634,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.726818810994168e-05,
+      "loss": 2.5616,
+      "step": 1535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.718331669421982e-05,
+      "loss": 2.5729,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.709820490209183e-05,
+      "loss": 2.5689,
+      "step": 1545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.701285328377403e-05,
+      "loss": 2.5683,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.692726239103309e-05,
+      "loss": 2.5651,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.68414327771825e-05,
+      "loss": 2.5765,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.675536499707902e-05,
+      "loss": 2.5124,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.666905960711907e-05,
+      "loss": 2.5536,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.658251716523507e-05,
+      "loss": 2.4841,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.649573823089195e-05,
+      "loss": 2.5788,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.64087233650835e-05,
+      "loss": 2.6088,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.632147313032862e-05,
+      "loss": 2.5722,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.623398809066789e-05,
+      "loss": 2.6251,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.614626881165978e-05,
+      "loss": 2.5487,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.605831586037699e-05,
+      "loss": 2.625,
+      "step": 1605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.597012980540288e-05,
+      "loss": 2.5746,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.588171121682771e-05,
+      "loss": 2.5837,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.5793060666245e-05,
+      "loss": 2.5601,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.570417872674779e-05,
+      "loss": 2.6014,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.561506597292498e-05,
+      "loss": 2.6488,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.552572298085765e-05,
+      "loss": 2.5375,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.54361503281152e-05,
+      "loss": 2.585,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.534634859375174e-05,
+      "loss": 2.5722,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.525631835830234e-05,
+      "loss": 2.581,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.516606020377917e-05,
+      "loss": 2.524,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.507557471366787e-05,
+      "loss": 2.5592,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.498486247292373e-05,
+      "loss": 2.5637,
+      "step": 1665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.489392406796784e-05,
+      "loss": 2.6254,
+      "step": 1670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.480276008668342e-05,
+      "loss": 2.5798,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.471137111841192e-05,
+      "loss": 2.5147,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.461975775394925e-05,
+      "loss": 2.5329,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.452792058554199e-05,
+      "loss": 2.5247,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.44358602068835e-05,
+      "loss": 2.5878,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.434357721311009e-05,
+      "loss": 2.5679,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.425107220079727e-05,
+      "loss": 2.5594,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.415834576795573e-05,
+      "loss": 2.5696,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.406539851402763e-05,
+      "loss": 2.5655,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.397223103988263e-05,
+      "loss": 2.5413,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.387884394781402e-05,
+      "loss": 2.5008,
+      "step": 1725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.378523784153485e-05,
+      "loss": 2.5431,
+      "step": 1730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.369141332617402e-05,
+      "loss": 2.5666,
+      "step": 1735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.359737100827237e-05,
+      "loss": 2.5595,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.350311149577873e-05,
+      "loss": 2.5758,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.340863539804604e-05,
+      "loss": 2.5214,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.331394332582736e-05,
+      "loss": 2.524,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.321903589127196e-05,
+      "loss": 2.5844,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.312391370792135e-05,
+      "loss": 2.522,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.302857739070528e-05,
+      "loss": 2.6073,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.293302755593786e-05,
+      "loss": 2.498,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.283726482131344e-05,
+      "loss": 2.5874,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.274128980590275e-05,
+      "loss": 2.5726,
+      "step": 1785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.264510313014878e-05,
+      "loss": 2.5902,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.254870541586289e-05,
+      "loss": 2.5697,
+      "step": 1795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.245209728622065e-05,
+      "loss": 2.6067,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.235527936575795e-05,
+      "loss": 2.5982,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.22582522803669e-05,
+      "loss": 2.5407,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.216101665729174e-05,
+      "loss": 2.5398,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.206357312512485e-05,
+      "loss": 2.6215,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.196592231380268e-05,
+      "loss": 2.5328,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.186806485460164e-05,
+      "loss": 2.5557,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.177000138013407e-05,
+      "loss": 2.5731,
+      "step": 1835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.167173252434406e-05,
+      "loss": 2.5365,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.157325892250351e-05,
+      "loss": 2.5361,
+      "step": 1845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.147458121120788e-05,
+      "loss": 2.5287,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.137570002837209e-05,
+      "loss": 2.5163,
+      "step": 1855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.127661601322646e-05,
+      "loss": 2.4885,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.117732980631256e-05,
+      "loss": 2.5741,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.107784204947904e-05,
+      "loss": 2.4965,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.097815338587748e-05,
+      "loss": 2.5882,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.087826445995828e-05,
+      "loss": 2.4481,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.077817591746642e-05,
+      "loss": 2.5411,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.067788840543738e-05,
+      "loss": 2.5581,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.057740257219286e-05,
+      "loss": 2.5525,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.047671906733665e-05,
+      "loss": 2.5367,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.037583854175043e-05,
+      "loss": 2.5393,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.027476164758952e-05,
+      "loss": 2.5398,
+      "step": 1910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.017348903827868e-05,
+      "loss": 2.598,
+      "step": 1915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.007202136850797e-05,
+      "loss": 2.4959,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.997035929422834e-05,
+      "loss": 2.6141,
+      "step": 1925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.986850347264755e-05,
+      "loss": 2.5012,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.976645456222591e-05,
+      "loss": 2.5286,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.966421322267186e-05,
+      "loss": 2.5149,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.956178011493792e-05,
+      "loss": 2.5775,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.945915590121629e-05,
+      "loss": 2.5362,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.93563412449346e-05,
+      "loss": 2.5677,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.925333681075163e-05,
+      "loss": 2.5623,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.915014326455298e-05,
+      "loss": 2.5345,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.904676127344682e-05,
+      "loss": 2.4833,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.89431915057595e-05,
+      "loss": 2.5733,
+      "step": 1975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.883943463103134e-05,
+      "loss": 2.5644,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.873549132001218e-05,
+      "loss": 2.4898,
+      "step": 1985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.863136224465715e-05,
+      "loss": 2.5107,
+      "step": 1990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.852704807812223e-05,
+      "loss": 2.5224,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.842254949476e-05,
+      "loss": 2.5082,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.831786717011518e-05,
+      "loss": 2.4844,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.821300178092032e-05,
+      "loss": 2.5662,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.810795400509143e-05,
+      "loss": 2.4738,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.80027245217236e-05,
+      "loss": 2.5294,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.78973140110865e-05,
+      "loss": 2.5815,
+      "step": 2025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.779172315462023e-05,
+      "loss": 2.5573,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.768595263493057e-05,
+      "loss": 2.5109,
+      "step": 2035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.758000313578495e-05,
+      "loss": 2.5221,
+      "step": 2040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.747387534210769e-05,
+      "loss": 2.5588,
+      "step": 2045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.736756993997575e-05,
+      "loss": 2.4336,
+      "step": 2050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.726108761661434e-05,
+      "loss": 2.4723,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.715442906039229e-05,
+      "loss": 2.535,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.704759496081782e-05,
+      "loss": 2.5439,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.694058600853387e-05,
+      "loss": 2.4775,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.68334028953138e-05,
+      "loss": 2.5442,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.672604631405686e-05,
+      "loss": 2.5248,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.661851695878366e-05,
+      "loss": 2.6155,
+      "step": 2085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.651081552463176e-05,
+      "loss": 2.5752,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.640294270785116e-05,
+      "loss": 2.5101,
+      "step": 2095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.629489920579977e-05,
+      "loss": 2.4996,
+      "step": 2100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618668571693891e-05,
+      "loss": 2.5165,
+      "step": 2105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.607830294082881e-05,
+      "loss": 2.4761,
+      "step": 2110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.596975157812407e-05,
+      "loss": 2.5279,
+      "step": 2115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.586103233056915e-05,
+      "loss": 2.5754,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.575214590099387e-05,
+      "loss": 2.5445,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.56430929933087e-05,
+      "loss": 2.5687,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.553387431250043e-05,
+      "loss": 2.5171,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.542449056462751e-05,
+      "loss": 2.5296,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.53149424568154e-05,
+      "loss": 2.4644,
+      "step": 2145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.520523069725222e-05,
+      "loss": 2.531,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.509535599518393e-05,
+      "loss": 2.561,
+      "step": 2155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.498531906090988e-05,
+      "loss": 2.5669,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.487512060577822e-05,
+      "loss": 2.5706,
+      "step": 2165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.476476134218124e-05,
+      "loss": 2.5093,
+      "step": 2170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.46542419835508e-05,
+      "loss": 2.4948,
+      "step": 2175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.454356324435375e-05,
+      "loss": 2.5179,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.443272584008726e-05,
+      "loss": 2.5313,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.432173048727415e-05,
+      "loss": 2.5251,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.421057790345843e-05,
+      "loss": 2.5091,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.409926880720047e-05,
+      "loss": 2.4804,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.398780391807243e-05,
+      "loss": 2.5982,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.38761839566537e-05,
+      "loss": 2.5096,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.376440964452603e-05,
+      "loss": 2.5203,
+      "step": 2215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.36524817042691e-05,
+      "loss": 2.5008,
+      "step": 2220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.354040085945569e-05,
+      "loss": 2.5205,
+      "step": 2225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342816783464703e-05,
+      "loss": 2.5532,
+      "step": 2230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.33157833553882e-05,
+      "loss": 2.4872,
+      "step": 2235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.320324814820334e-05,
+      "loss": 2.5489,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.309056294059096e-05,
+      "loss": 2.5336,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.297772846101932e-05,
+      "loss": 2.5369,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.286474543892162e-05,
+      "loss": 2.4666,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.27516146046914e-05,
+      "loss": 2.5197,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.26383366896777e-05,
+      "loss": 2.5219,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.252491242618039e-05,
+      "loss": 2.4867,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.24113425474454e-05,
+      "loss": 2.5454,
+      "step": 2275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.229762778766009e-05,
+      "loss": 2.4834,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.218376888194833e-05,
+      "loss": 2.4881,
+      "step": 2285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.206976656636593e-05,
+      "loss": 2.532,
+      "step": 2290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195562157789572e-05,
+      "loss": 2.4653,
+      "step": 2295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.184133465444286e-05,
+      "loss": 2.4739,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.172690653483012e-05,
+      "loss": 2.4513,
+      "step": 2305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.161233795879304e-05,
+      "loss": 2.5539,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.149762966697509e-05,
+      "loss": 2.4381,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.138278240092304e-05,
+      "loss": 2.5221,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.126779690308202e-05,
+      "loss": 2.5528,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.11526739167908e-05,
+      "loss": 2.4756,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.103741418627696e-05,
+      "loss": 2.5425,
+      "step": 2335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.092201845665206e-05,
+      "loss": 2.5092,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.08064874739069e-05,
+      "loss": 2.4613,
+      "step": 2345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.069082198490656e-05,
+      "loss": 2.5191,
+      "step": 2350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.057502273738573e-05,
+      "loss": 2.4872,
+      "step": 2355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045909047994375e-05,
+      "loss": 2.4849,
+      "step": 2360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.034302596203984e-05,
+      "loss": 2.5193,
+      "step": 2365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.022682993398826e-05,
+      "loss": 2.535,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.011050314695335e-05,
+      "loss": 2.5603,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.999404635294485e-05,
+      "loss": 2.4729,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.98774603048129e-05,
+      "loss": 2.5419,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.976074575624324e-05,
+      "loss": 2.4422,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.964390346175231e-05,
+      "loss": 2.4546,
+      "step": 2395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.952693417668239e-05,
+      "loss": 2.4598,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.940983865719665e-05,
+      "loss": 2.5084,
+      "step": 2405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.929261766027442e-05,
+      "loss": 2.5561,
+      "step": 2410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.917527194370613e-05,
+      "loss": 2.4791,
+      "step": 2415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905780226608847e-05,
+      "loss": 2.5434,
+      "step": 2420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.894020938681952e-05,
+      "loss": 2.5001,
+      "step": 2425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.88224940660938e-05,
+      "loss": 2.5509,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.870465706489738e-05,
+      "loss": 2.5467,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.85866991450029e-05,
+      "loss": 2.4954,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.846862106896481e-05,
+      "loss": 2.5265,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.83504236001142e-05,
+      "loss": 2.5431,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.823210750255404e-05,
+      "loss": 2.5598,
+      "step": 2455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.811367354115422e-05,
+      "loss": 2.4694,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.799512248154654e-05,
+      "loss": 2.5054,
+      "step": 2465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.787645509011978e-05,
+      "loss": 2.4816,
+      "step": 2470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.775767213401482e-05,
+      "loss": 2.5461,
+      "step": 2475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763877438111958e-05,
+      "loss": 2.4212,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751976260006409e-05,
+      "loss": 2.523,
+      "step": 2485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.740063756021557e-05,
+      "loss": 2.5112,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.728140003167339e-05,
+      "loss": 2.455,
+      "step": 2495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.716205078526411e-05,
+      "loss": 2.5422,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.704259059253657e-05,
+      "loss": 2.4659,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.692302022575672e-05,
+      "loss": 2.4748,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.680334045790283e-05,
+      "loss": 2.4998,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.66835520626604e-05,
+      "loss": 2.4328,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.656365581441714e-05,
+      "loss": 2.5124,
+      "step": 2525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.644365248825802e-05,
+      "loss": 2.4182,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.632354285996019e-05,
+      "loss": 2.4805,
+      "step": 2535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.620332770598803e-05,
+      "loss": 2.5136,
+      "step": 2540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60830078034881e-05,
+      "loss": 2.5551,
+      "step": 2545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596258393028415e-05,
+      "loss": 2.4793,
+      "step": 2550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5842056864872e-05,
+      "loss": 2.4842,
+      "step": 2555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.572142738641462e-05,
+      "loss": 2.4513,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5600696274737e-05,
+      "loss": 2.4822,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.547986431032119e-05,
+      "loss": 2.4882,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.535893227430122e-05,
+      "loss": 2.5499,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.523790094845801e-05,
+      "loss": 2.4557,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.511677111521437e-05,
+      "loss": 2.511,
+      "step": 2585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.499554355762993e-05,
+      "loss": 2.5306,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.487421905939607e-05,
+      "loss": 2.52,
+      "step": 2595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.475279840483083e-05,
+      "loss": 2.4901,
+      "step": 2600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.463128237887392e-05,
+      "loss": 2.5071,
+      "step": 2605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450967176708153e-05,
+      "loss": 2.579,
+      "step": 2610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438796735562135e-05,
+      "loss": 2.4971,
+      "step": 2615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.426616993126743e-05,
+      "loss": 2.5702,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.414428028139513e-05,
+      "loss": 2.4756,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.402229919397602e-05,
+      "loss": 2.52,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.390022745757275e-05,
+      "loss": 2.3993,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.377806586133403e-05,
+      "loss": 2.5063,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.365581519498939e-05,
+      "loss": 2.4919,
+      "step": 2645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.353347624884429e-05,
+      "loss": 2.549,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.341104981377479e-05,
+      "loss": 2.4678,
+      "step": 2655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.328853668122257e-05,
+      "loss": 2.4895,
+      "step": 2660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.316593764318977e-05,
+      "loss": 2.5121,
+      "step": 2665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30432534922339e-05,
+      "loss": 2.5164,
+      "step": 2670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.292048502146264e-05,
+      "loss": 2.4327,
+      "step": 2675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.279763302452883e-05,
+      "loss": 2.4712,
+      "step": 2680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.267469829562524e-05,
+      "loss": 2.4445,
+      "step": 2685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.255168162947945e-05,
+      "loss": 2.4735,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.242858382134882e-05,
+      "loss": 2.4882,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.230540566701514e-05,
+      "loss": 2.439,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.218214796277972e-05,
+      "loss": 2.5048,
+      "step": 2705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.205881150545805e-05,
+      "loss": 2.4963,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.193539709237478e-05,
+      "loss": 2.5115,
+      "step": 2715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.181190552135851e-05,
+      "loss": 2.4811,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.168833759073661e-05,
+      "loss": 2.4748,
+      "step": 2725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.15646940993301e-05,
+      "loss": 2.5262,
+      "step": 2730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14409758464485e-05,
+      "loss": 2.4351,
+      "step": 2735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131718363188457e-05,
+      "loss": 2.4975,
+      "step": 2740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.119331825590927e-05,
+      "loss": 2.55,
+      "step": 2745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.106938051926651e-05,
+      "loss": 2.5093,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.094537122316795e-05,
+      "loss": 2.4726,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0821291169287885e-05,
+      "loss": 2.5126,
+      "step": 2760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.069714115975803e-05,
+      "loss": 2.5113,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0572921997162336e-05,
+      "loss": 2.3986,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0448634484531796e-05,
+      "loss": 2.4644,
+      "step": 2775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0324279425339245e-05,
+      "loss": 2.5088,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0199857623494225e-05,
+      "loss": 2.4531,
+      "step": 2785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.007536988333772e-05,
+      "loss": 2.4627,
+      "step": 2790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.995081700963696e-05,
+      "loss": 2.4317,
+      "step": 2795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9826199807580305e-05,
+      "loss": 2.4163,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9701519082771886e-05,
+      "loss": 2.4626,
+      "step": 2805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.957677564122653e-05,
+      "loss": 2.4763,
+      "step": 2810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.945197028936453e-05,
+      "loss": 2.4568,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.932710383400638e-05,
+      "loss": 2.4424,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.920217708236756e-05,
+      "loss": 2.4332,
+      "step": 2825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.90771908420534e-05,
+      "loss": 2.5275,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.895214592105374e-05,
+      "loss": 2.503,
+      "step": 2835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8827043127737815e-05,
+      "loss": 2.4989,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8701883270848964e-05,
+      "loss": 2.5082,
+      "step": 2845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8576667159499435e-05,
+      "loss": 2.4808,
+      "step": 2850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.845139560316513e-05,
+      "loss": 2.4658,
+      "step": 2855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832606941168036e-05,
+      "loss": 2.4405,
+      "step": 2860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820068939523268e-05,
+      "loss": 2.5497,
+      "step": 2865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807525636435757e-05,
+      "loss": 2.5262,
+      "step": 2870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.794977112993323e-05,
+      "loss": 2.4703,
+      "step": 2875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.782423450317534e-05,
+      "loss": 2.4646,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.769864729563186e-05,
+      "loss": 2.4877,
+      "step": 2885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7573010319177634e-05,
+      "loss": 2.5004,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7447324386009327e-05,
+      "loss": 2.5191,
+      "step": 2895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.732159030864007e-05,
+      "loss": 2.4871,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.719580889989422e-05,
+      "loss": 2.514,
+      "step": 2905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7069980972902104e-05,
+      "loss": 2.4289,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.694410734109481e-05,
+      "loss": 2.5312,
+      "step": 2915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.681818881819886e-05,
+      "loss": 2.442,
+      "step": 2920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669222621823098e-05,
+      "loss": 2.4419,
+      "step": 2925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656622035549285e-05,
+      "loss": 2.4767,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6440172044565845e-05,
+      "loss": 2.4705,
+      "step": 2935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.631408210030569e-05,
+      "loss": 2.4408,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6187951337837306e-05,
+      "loss": 2.4542,
+      "step": 2945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.606178057254949e-05,
+      "loss": 2.491,
+      "step": 2950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.593557062008963e-05,
+      "loss": 2.5057,
+      "step": 2955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5809322296358425e-05,
+      "loss": 2.5325,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5683036417504675e-05,
+      "loss": 2.4352,
+      "step": 2965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5556713799919916e-05,
+      "loss": 2.5172,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.54303552602332e-05,
+      "loss": 2.5187,
+      "step": 2975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.530396161530583e-05,
+      "loss": 2.4513,
+      "step": 2980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5177533682226e-05,
+      "loss": 2.4095,
+      "step": 2985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.505107227830363e-05,
+      "loss": 2.5001,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492457822106494e-05,
+      "loss": 2.4889,
+      "step": 2995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4798052328247285e-05,
+      "loss": 2.445,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.467149541779384e-05,
+      "loss": 2.4729,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4544908307848265e-05,
+      "loss": 2.5435,
+      "step": 3010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.441829181674947e-05,
+      "loss": 2.4275,
+      "step": 3015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.429164676302629e-05,
+      "loss": 2.5433,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.416497396539221e-05,
+      "loss": 2.406,
+      "step": 3025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.403827424274007e-05,
+      "loss": 2.3998,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.391154841413678e-05,
+      "loss": 2.468,
+      "step": 3035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3784797298818e-05,
+      "loss": 2.4876,
+      "step": 3040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365802171618288e-05,
+      "loss": 2.4607,
+      "step": 3045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.353122248578871e-05,
+      "loss": 2.4237,
+      "step": 3050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3404400427345665e-05,
+      "loss": 2.4549,
+      "step": 3055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.327755636071152e-05,
+      "loss": 2.5051,
+      "step": 3060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.315069110588631e-05,
+      "loss": 2.458,
+      "step": 3065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.302380548300704e-05,
+      "loss": 2.4539,
+      "step": 3070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.289690031234237e-05,
+      "loss": 2.476,
+      "step": 3075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.276997641428736e-05,
+      "loss": 2.515,
+      "step": 3080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.264303460935811e-05,
+      "loss": 2.4769,
+      "step": 3085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.251607571818652e-05,
+      "loss": 2.513,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.23891005615149e-05,
+      "loss": 2.5087,
+      "step": 3095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.226210996019073e-05,
+      "loss": 2.4987,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.213510473516134e-05,
+      "loss": 2.4424,
+      "step": 3105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.200808570746859e-05,
+      "loss": 2.4414,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.188105369824358e-05,
+      "loss": 2.453,
+      "step": 3115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.17540095287013e-05,
+      "loss": 2.4853,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.162695402013539e-05,
+      "loss": 2.4704,
+      "step": 3125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1499887993912766e-05,
+      "loss": 2.4924,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1372812271468364e-05,
+      "loss": 2.5223,
+      "step": 3135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.124572767429976e-05,
+      "loss": 2.5118,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.111863502396195e-05,
+      "loss": 2.4704,
+      "step": 3145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.099153514206195e-05,
+      "loss": 2.5244,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.086442885025357e-05,
+      "loss": 2.4702,
+      "step": 3155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.073731697023198e-05,
+      "loss": 2.4698,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.061020032372857e-05,
+      "loss": 2.4047,
+      "step": 3165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.04830797325055e-05,
+      "loss": 2.4194,
+      "step": 3170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03559560183504e-05,
+      "loss": 2.4488,
+      "step": 3175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.022883000307117e-05,
+      "loss": 2.4869,
+      "step": 3180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.010170250849048e-05,
+      "loss": 2.4636,
+      "step": 3185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.997457435644064e-05,
+      "loss": 2.457,
+      "step": 3190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9847446368758205e-05,
+      "loss": 2.4465,
+      "step": 3195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.972031936727864e-05,
+      "loss": 2.4596,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.959319417383104e-05,
+      "loss": 2.4294,
+      "step": 3205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9466071610232837e-05,
+      "loss": 2.5443,
+      "step": 3210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9338952498284416e-05,
+      "loss": 2.4314,
+      "step": 3215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.921183765976388e-05,
+      "loss": 2.4529,
+      "step": 3220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.908472791642172e-05,
+      "loss": 2.4449,
+      "step": 3225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.895762408997546e-05,
+      "loss": 2.4325,
+      "step": 3230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883052700210439e-05,
+      "loss": 2.3924,
+      "step": 3235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8703437474444214e-05,
+      "loss": 2.4665,
+      "step": 3240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8576356328581784e-05,
+      "loss": 2.4123,
+      "step": 3245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844928438604977e-05,
+      "loss": 2.4886,
+      "step": 3250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8322222468321335e-05,
+      "loss": 2.4188,
+      "step": 3255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8195171396804825e-05,
+      "loss": 2.4151,
+      "step": 3260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.806813199283851e-05,
+      "loss": 2.4873,
+      "step": 3265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.794110507768518e-05,
+      "loss": 2.4665,
+      "step": 3270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.781409147252692e-05,
+      "loss": 2.4288,
+      "step": 3275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.768709199845975e-05,
+      "loss": 2.4325,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7560107476488394e-05,
+      "loss": 2.4698,
+      "step": 3285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7433138727520856e-05,
+      "loss": 2.4824,
+      "step": 3290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.730618657236319e-05,
+      "loss": 2.463,
+      "step": 3295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717925183171418e-05,
+      "loss": 2.4346,
+      "step": 3300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7052335326160024e-05,
+      "loss": 2.4072,
+      "step": 3305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6925437876169076e-05,
+      "loss": 2.4965,
+      "step": 3310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6798560302086433e-05,
+      "loss": 2.413,
+      "step": 3315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667170342412876e-05,
+      "loss": 2.4827,
+      "step": 3320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.654486806237891e-05,
+      "loss": 2.4368,
+      "step": 3325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6418055036780636e-05,
+      "loss": 2.4497,
+      "step": 3330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.629126516713331e-05,
+      "loss": 2.4877,
+      "step": 3335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.616449927308658e-05,
+      "loss": 2.3907,
+      "step": 3340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6037758174135134e-05,
+      "loss": 2.4938,
+      "step": 3345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.591104268961336e-05,
+      "loss": 2.5219,
+      "step": 3350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.578435363869005e-05,
+      "loss": 2.4877,
+      "step": 3355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565769184036311e-05,
+      "loss": 2.5543,
+      "step": 3360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.55310581134543e-05,
+      "loss": 2.5095,
+      "step": 3365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.540445327660384e-05,
+      "loss": 2.5164,
+      "step": 3370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5277878148265266e-05,
+      "loss": 2.4998,
+      "step": 3375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.51513335467e-05,
+      "loss": 2.4646,
+      "step": 3380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5024820289972166e-05,
+      "loss": 2.5275,
+      "step": 3385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.489833919594323e-05,
+      "loss": 2.4867,
+      "step": 3390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.477189108226671e-05,
+      "loss": 2.4888,
+      "step": 3395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4645476766382975e-05,
+      "loss": 2.4694,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.451909706551386e-05,
+      "loss": 2.4332,
+      "step": 3405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.439275279665747e-05,
+      "loss": 2.4454,
+      "step": 3410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4266444776582805e-05,
+      "loss": 2.4118,
+      "step": 3415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.414017382182457e-05,
+      "loss": 2.4675,
+      "step": 3420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4013940748677824e-05,
+      "loss": 2.432,
+      "step": 3425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3887746373192765e-05,
+      "loss": 2.4323,
+      "step": 3430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3761591511169395e-05,
+      "loss": 2.4853,
+      "step": 3435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.363547697815231e-05,
+      "loss": 2.4549,
+      "step": 3440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.350940358942538e-05,
+      "loss": 2.4614,
+      "step": 3445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3383372160006474e-05,
+      "loss": 2.4261,
+      "step": 3450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.325738350464223e-05,
+      "loss": 2.4382,
+      "step": 3455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.313143843780276e-05,
+      "loss": 2.4806,
+      "step": 3460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.30055377736764e-05,
+      "loss": 2.4526,
+      "step": 3465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2879682326164415e-05,
+      "loss": 2.4854,
+      "step": 3470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.275387290887582e-05,
+      "loss": 2.4108,
+      "step": 3475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.262811033512198e-05,
+      "loss": 2.459,
+      "step": 3480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.250239541791149e-05,
+      "loss": 2.4638,
+      "step": 3485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.237672896994485e-05,
+      "loss": 2.4094,
+      "step": 3490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2251111803609217e-05,
+      "loss": 2.4738,
+      "step": 3495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2125544730973166e-05,
+      "loss": 2.4804,
+      "step": 3500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.200002856378146e-05,
+      "loss": 2.4199,
+      "step": 3505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.187456411344971e-05,
+      "loss": 2.4605,
+      "step": 3510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.174915219105926e-05,
+      "loss": 2.4797,
+      "step": 3515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.162379360735185e-05,
+      "loss": 2.5086,
+      "step": 3520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1498489172724417e-05,
+      "loss": 2.4062,
+      "step": 3525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1373239697223867e-05,
+      "loss": 2.4413,
+      "step": 3530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.124804599054175e-05,
+      "loss": 2.4378,
+      "step": 3535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.112290886200914e-05,
+      "loss": 2.4544,
+      "step": 3540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.099782912059135e-05,
+      "loss": 2.4437,
+      "step": 3545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.087280757488271e-05,
+      "loss": 2.4374,
+      "step": 3550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.074784503310131e-05,
+      "loss": 2.4287,
+      "step": 3555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062294230308385e-05,
+      "loss": 2.4502,
+      "step": 3560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049810019228029e-05,
+      "loss": 2.4278,
+      "step": 3565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.037331950774879e-05,
+      "loss": 2.4773,
+      "step": 3570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.024860105615037e-05,
+      "loss": 2.4424,
+      "step": 3575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0123945643743746e-05,
+      "loss": 2.4163,
+      "step": 3580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.99993540763801e-05,
+      "loss": 2.4628,
+      "step": 3585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.987482715949791e-05,
+      "loss": 2.403,
+      "step": 3590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9750365698117686e-05,
+      "loss": 2.3713,
+      "step": 3595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9625970496836785e-05,
+      "loss": 2.4766,
+      "step": 3600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.950164235982424e-05,
+      "loss": 2.4726,
+      "step": 3605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.937738209081551e-05,
+      "loss": 2.4213,
+      "step": 3610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.925319049310735e-05,
+      "loss": 2.4801,
+      "step": 3615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9129068369552544e-05,
+      "loss": 2.4778,
+      "step": 3620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9005016522554785e-05,
+      "loss": 2.4513,
+      "step": 3625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.888103575406343e-05,
+      "loss": 2.3497,
+      "step": 3630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8757126865568325e-05,
+      "loss": 2.4625,
+      "step": 3635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.863329065809469e-05,
+      "loss": 2.3992,
+      "step": 3640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8509527932197836e-05,
+      "loss": 2.4532,
+      "step": 3645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.838583948795808e-05,
+      "loss": 2.4292,
+      "step": 3650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.826222612497551e-05,
+      "loss": 2.4234,
+      "step": 3655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.813868864236487e-05,
+      "loss": 2.4156,
+      "step": 3660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8015227838750336e-05,
+      "loss": 2.4331,
+      "step": 3665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.789184451226038e-05,
+      "loss": 2.4602,
+      "step": 3670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.776853946052265e-05,
+      "loss": 2.4029,
+      "step": 3675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.764531348065874e-05,
+      "loss": 2.3958,
+      "step": 3680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7522167369279105e-05,
+      "loss": 2.4764,
+      "step": 3685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.739910192247784e-05,
+      "loss": 2.4417,
+      "step": 3690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.727611793582761e-05,
+      "loss": 2.4161,
+      "step": 3695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7153216204374446e-05,
+      "loss": 2.4318,
+      "step": 3700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7030397522632646e-05,
+      "loss": 2.4085,
+      "step": 3705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.690766268457961e-05,
+      "loss": 2.3812,
+      "step": 3710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.678501248365075e-05,
+      "loss": 2.5069,
+      "step": 3715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.666244771273427e-05,
+      "loss": 2.4358,
+      "step": 3720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6539969164166153e-05,
+      "loss": 2.4077,
+      "step": 3725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.641757762972496e-05,
+      "loss": 2.5214,
+      "step": 3730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.629527390062676e-05,
+      "loss": 2.502,
+      "step": 3735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6173058767519985e-05,
+      "loss": 2.3613,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.60509330204803e-05,
+      "loss": 2.4306,
+      "step": 3745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.592889744900556e-05,
+      "loss": 2.4237,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.580695284201063e-05,
+      "loss": 2.3893,
+      "step": 3755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.568509998782238e-05,
+      "loss": 2.4358,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.556333967417447e-05,
+      "loss": 2.4096,
+      "step": 3765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.544167268820237e-05,
+      "loss": 2.4549,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5320099816438176e-05,
+      "loss": 2.427,
+      "step": 3775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.519862184480559e-05,
+      "loss": 2.4802,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5077239558614844e-05,
+      "loss": 2.4517,
+      "step": 3785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.495595374255756e-05,
+      "loss": 2.4289,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4834765180701745e-05,
+      "loss": 2.406,
+      "step": 3795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.471367465648668e-05,
+      "loss": 2.4334,
+      "step": 3800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.459268295271786e-05,
+      "loss": 2.4626,
+      "step": 3805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4471790851561955e-05,
+      "loss": 2.4272,
+      "step": 3810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.435099913454174e-05,
+      "loss": 2.4187,
+      "step": 3815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.423030858253105e-05,
+      "loss": 2.4106,
+      "step": 3820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.410971997574973e-05,
+      "loss": 2.5079,
+      "step": 3825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.398923409375854e-05,
+      "loss": 2.3948,
+      "step": 3830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.386885171545422e-05,
+      "loss": 2.4584,
+      "step": 3835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.374857361906436e-05,
+      "loss": 2.4539,
+      "step": 3840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.362840058214244e-05,
+      "loss": 2.4706,
+      "step": 3845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.350833338156273e-05,
+      "loss": 2.4435,
+      "step": 3850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3388372793515356e-05,
+      "loss": 2.3713,
+      "step": 3855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.326851959350116e-05,
+      "loss": 2.4485,
+      "step": 3860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3148774556326834e-05,
+      "loss": 2.42,
+      "step": 3865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.30291384560998e-05,
+      "loss": 2.4893,
+      "step": 3870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.290961206622325e-05,
+      "loss": 2.4407,
+      "step": 3875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.279019615939114e-05,
+      "loss": 2.3767,
+      "step": 3880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.267089150758319e-05,
+      "loss": 2.4738,
+      "step": 3885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.255169888205988e-05,
+      "loss": 2.4332,
+      "step": 3890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2432619053357537e-05,
+      "loss": 2.4433,
+      "step": 3895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2313652791283255e-05,
+      "loss": 2.3964,
+      "step": 3900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.219480086490997e-05,
+      "loss": 2.4557,
+      "step": 3905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.207606404257146e-05,
+      "loss": 2.5372,
+      "step": 3910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.195744309185743e-05,
+      "loss": 2.4464,
+      "step": 3915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.183893877960851e-05,
+      "loss": 2.4602,
+      "step": 3920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172055187191131e-05,
+      "loss": 2.3905,
+      "step": 3925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.160228313409343e-05,
+      "loss": 2.4858,
+      "step": 3930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148413333071857e-05,
+      "loss": 2.4622,
+      "step": 3935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136610322558158e-05,
+      "loss": 2.5041,
+      "step": 3940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.124819358170347e-05,
+      "loss": 2.4873,
+      "step": 3945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.113040516132655e-05,
+      "loss": 2.4707,
+      "step": 3950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.101273872590943e-05,
+      "loss": 2.4044,
+      "step": 3955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0895195036122135e-05,
+      "loss": 2.4731,
+      "step": 3960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.077777485184122e-05,
+      "loss": 2.4326,
+      "step": 3965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.066047893214478e-05,
+      "loss": 2.3916,
+      "step": 3970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.054330803530762e-05,
+      "loss": 2.4029,
+      "step": 3975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0426262918796296e-05,
+      "loss": 2.4183,
+      "step": 3980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0309344339264232e-05,
+      "loss": 2.3945,
+      "step": 3985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.019255305254688e-05,
+      "loss": 2.4499,
+      "step": 3990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0075889813656743e-05,
+      "loss": 2.4257,
+      "step": 3995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9959355376778575e-05,
+      "loss": 2.4222,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 6278,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 9.527057801281536e+18,
+  "trial_name": null,
+  "trial_params": null
+}