diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5398 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 3558,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.586692570092768e-06,
+      "loss": 0.8625,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.38003885513915e-06,
+      "loss": 0.6788,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0014041683420038e-05,
+      "loss": 0.621,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1173385140185536e-05,
+      "loss": 0.5789,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2072641788091208e-05,
+      "loss": 0.5749,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2807387968466421e-05,
+      "loss": 0.556,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3428607012429058e-05,
+      "loss": 0.5398,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3966731425231921e-05,
+      "loss": 0.5133,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4441390796747306e-05,
+      "loss": 0.4831,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4865988073137594e-05,
+      "loss": 0.4442,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.525008303038526e-05,
+      "loss": 0.4236,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5600734253512805e-05,
+      "loss": 0.4162,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5923302109800277e-05,
+      "loss": 0.424,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6221953297475443e-05,
+      "loss": 0.4247,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6499990901418477e-05,
+      "loss": 0.4126,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.67600777102783e-05,
+      "loss": 0.4205,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7004391712964543e-05,
+      "loss": 0.4237,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.723473708179369e-05,
+      "loss": 0.4081,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7452625108916732e-05,
+      "loss": 0.4196,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7659334358183974e-05,
+      "loss": 0.4221,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7855956125756325e-05,
+      "loss": 0.4111,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8043429315431648e-05,
+      "loss": 0.4069,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.822256755522178e-05,
+      "loss": 0.4147,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8394080538559187e-05,
+      "loss": 0.4025,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8558591006089653e-05,
+      "loss": 0.4197,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.871664839484666e-05,
+      "loss": 0.397,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8868739910074575e-05,
+      "loss": 0.4152,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9015299582521825e-05,
+      "loss": 0.4015,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9156715735658306e-05,
+      "loss": 0.3958,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9293337186464862e-05,
+      "loss": 0.4001,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9425478429102448e-05,
+      "loss": 0.3967,
+      "step": 124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9553423995324687e-05,
+      "loss": 0.3871,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967743214371253e-05,
+      "loss": 0.3916,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9797737998010925e-05,
+      "loss": 0.3962,
+      "step": 136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9914556230427497e-05,
+      "loss": 0.3968,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.4023,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3955,
+      "step": 148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3986,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3857,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3934,
+      "step": 160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3887,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3862,
+      "step": 168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3893,
+      "step": 172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3856,
+      "step": 176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3901,
+      "step": 180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3883,
+      "step": 184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3945,
+      "step": 188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.382,
+      "step": 192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3804,
+      "step": 196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3886,
+      "step": 200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3832,
+      "step": 204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3815,
+      "step": 208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3893,
+      "step": 212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3859,
+      "step": 216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.389,
+      "step": 220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3884,
+      "step": 224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.388,
+      "step": 228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3846,
+      "step": 232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3819,
+      "step": 236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3862,
+      "step": 240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.3798,
+      "step": 244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.3754,
+      "step": 248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.3797,
+      "step": 252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.3773,
+      "step": 256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.3772,
+      "step": 260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.3766,
+      "step": 264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.3797,
+      "step": 268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.3728,
+      "step": 272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.3726,
+      "step": 276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3811,
+      "step": 280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3796,
+      "step": 284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3791,
+      "step": 288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.3717,
+      "step": 292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.377,
+      "step": 296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.3653,
+      "step": 300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.3777,
+      "step": 304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.3724,
+      "step": 308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.3685,
+      "step": 312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3691,
+      "step": 316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3768,
+      "step": 320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3739,
+      "step": 324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.371,
+      "step": 328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.3728,
+      "step": 332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.3681,
+      "step": 336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.3802,
+      "step": 340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.3699,
+      "step": 344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.3719,
+      "step": 348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.3594,
+      "step": 352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.3696,
+      "step": 356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.3733,
+      "step": 360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.3755,
+      "step": 364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.3765,
+      "step": 368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.3705,
+      "step": 372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.3746,
+      "step": 376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.3771,
+      "step": 380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.3708,
+      "step": 384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.3593,
+      "step": 388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.3681,
+      "step": 392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.368,
+      "step": 396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3702,
+      "step": 400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3698,
+      "step": 404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3724,
+      "step": 408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.369,
+      "step": 412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.3748,
+      "step": 416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.3714,
+      "step": 420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3645,
+      "step": 424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3679,
+      "step": 428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3729,
+      "step": 432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.3675,
+      "step": 436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.3742,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.3626,
+      "step": 444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.3732,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.3699,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.3572,
+      "step": 456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.365,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.3572,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.3796,
+      "step": 468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.36,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.3607,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.3688,
+      "step": 480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.3572,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.3576,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.3583,
+      "step": 492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.3684,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.3591,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.3674,
+      "step": 504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.3599,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.3613,
+      "step": 512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.3592,
+      "step": 516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.3606,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.3636,
+      "step": 524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.3651,
+      "step": 528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.3672,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.3572,
+      "step": 536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.367,
+      "step": 540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.3551,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.364,
+      "step": 548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.3681,
+      "step": 552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.3514,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.3646,
+      "step": 560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.3593,
+      "step": 564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.364,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.3598,
+      "step": 572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.3635,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.3643,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.3692,
+      "step": 584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3571,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3564,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.36566540598869324,
+      "eval_runtime": 647.3717,
+      "eval_samples_per_second": 11.213,
+      "eval_steps_per_second": 0.176,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3604,
+      "step": 596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.3635,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.3686,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.3535,
+      "step": 608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.3656,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.3608,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.3555,
+      "step": 620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.3684,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.368,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.361,
+      "step": 632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.3628,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.3602,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.3601,
+      "step": 644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.357,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.3539,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.3612,
+      "step": 656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.3528,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.3671,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.3534,
+      "step": 668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.3507,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.3596,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.3599,
+      "step": 680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.3597,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.3642,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.3562,
+      "step": 692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.356,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.3517,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.3575,
+      "step": 704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.355,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.361,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.3563,
+      "step": 716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.361,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.3523,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.3527,
+      "step": 728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.358,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.3494,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.3584,
+      "step": 740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.3635,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.3625,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.3567,
+      "step": 752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.3597,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.3604,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.368,
+      "step": 764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.3592,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.3601,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.3557,
+      "step": 776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.3567,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.3588,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.3581,
+      "step": 788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.3533,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.3465,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.3599,
+      "step": 800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.347,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.3506,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.3517,
+      "step": 812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.3447,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.3587,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.3558,
+      "step": 824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.3545,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.3548,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.3554,
+      "step": 836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.3537,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.35,
+      "step": 844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.358,
+      "step": 848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.3569,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.3602,
+      "step": 856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.3496,
+      "step": 860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.3471,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.3525,
+      "step": 868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.3489,
+      "step": 872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.3528,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.3493,
+      "step": 880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.3524,
+      "step": 884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.3468,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.3456,
+      "step": 892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.3527,
+      "step": 896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.3523,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.3525,
+      "step": 904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.3517,
+      "step": 908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.3455,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.348,
+      "step": 916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.3504,
+      "step": 920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.3601,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.353,
+      "step": 928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.3548,
+      "step": 932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.3481,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.345,
+      "step": 940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.3471,
+      "step": 944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.3556,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.3516,
+      "step": 952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.3471,
+      "step": 956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.3496,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.3446,
+      "step": 964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.347,
+      "step": 968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.3554,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.3524,
+      "step": 976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.343,
+      "step": 980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.3603,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.356,
+      "step": 988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.3463,
+      "step": 992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.3479,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.3429,
+      "step": 1000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.3486,
+      "step": 1004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.3506,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.3701,
+      "step": 1012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.3495,
+      "step": 1016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.3495,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.3555,
+      "step": 1024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.3487,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.3482,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.3507,
+      "step": 1036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.3512,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.3519,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.3596,
+      "step": 1048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.3492,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.3449,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.3492,
+      "step": 1060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.3456,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.3451,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.3443,
+      "step": 1072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.3597,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.3496,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.3506,
+      "step": 1084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.3496,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.3503,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.3459,
+      "step": 1096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.3573,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.3482,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.3568,
+      "step": 1108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.347,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.3606,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.3376,
+      "step": 1120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.3517,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.35,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.3472,
+      "step": 1132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.3487,
+      "step": 1136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.3489,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.3457,
+      "step": 1144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.3409,
+      "step": 1148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.3522,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.3407,
+      "step": 1156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.3408,
+      "step": 1160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.3444,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.3418,
+      "step": 1168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.3429,
+      "step": 1172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.3523,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.3534,
+      "step": 1180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.3412,
+      "step": 1184
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.348991334438324,
+      "eval_runtime": 644.1984,
+      "eval_samples_per_second": 11.268,
+      "eval_steps_per_second": 0.177,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.3429,
+      "step": 1188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.339,
+      "step": 1192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.3376,
+      "step": 1196
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.3305,
+      "step": 1200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.3425,
+      "step": 1204
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.335,
+      "step": 1208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.3322,
+      "step": 1212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.3324,
+      "step": 1216
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.3331,
+      "step": 1220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.3455,
+      "step": 1224
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.337,
+      "step": 1228
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.345,
+      "step": 1232
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3354,
+      "step": 1236
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3356,
+      "step": 1240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3405,
+      "step": 1244
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3275,
+      "step": 1248
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3351,
+      "step": 1252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3415,
+      "step": 1256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3276,
+      "step": 1260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3313,
+      "step": 1264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3391,
+      "step": 1268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3413,
+      "step": 1272
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3388,
+      "step": 1276
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3289,
+      "step": 1280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3339,
+      "step": 1284
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3408,
+      "step": 1288
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3313,
+      "step": 1292
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3343,
+      "step": 1296
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3357,
+      "step": 1300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3389,
+      "step": 1304
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3332,
+      "step": 1308
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3349,
+      "step": 1312
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3346,
+      "step": 1316
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3358,
+      "step": 1320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3303,
+      "step": 1324
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.338,
+      "step": 1328
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3293,
+      "step": 1332
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3387,
+      "step": 1336
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3334,
+      "step": 1340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3309,
+      "step": 1344
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3388,
+      "step": 1348
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.333,
+      "step": 1352
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3372,
+      "step": 1356
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3333,
+      "step": 1360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3384,
+      "step": 1364
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3284,
+      "step": 1368
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3277,
+      "step": 1372
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3361,
+      "step": 1376
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3324,
+      "step": 1380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3315,
+      "step": 1384
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3351,
+      "step": 1388
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3329,
+      "step": 1392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3326,
+      "step": 1396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3352,
+      "step": 1400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3318,
+      "step": 1404
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3334,
+      "step": 1408
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3238,
+      "step": 1412
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3338,
+      "step": 1416
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3235,
+      "step": 1420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3346,
+      "step": 1424
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3325,
+      "step": 1428
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.3275,
+      "step": 1432
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.3322,
+      "step": 1436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.3292,
+      "step": 1440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.3346,
+      "step": 1444
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.337,
+      "step": 1448
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.3403,
+      "step": 1452
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.3375,
+      "step": 1456
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.3326,
+      "step": 1460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.3366,
+      "step": 1464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3333,
+      "step": 1468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3371,
+      "step": 1472
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3294,
+      "step": 1476
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.3268,
+      "step": 1480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.3274,
+      "step": 1484
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.3335,
+      "step": 1488
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.332,
+      "step": 1492
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.3299,
+      "step": 1496
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.3328,
+      "step": 1500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3273,
+      "step": 1504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3241,
+      "step": 1508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3311,
+      "step": 1512
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.3233,
+      "step": 1516
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.3367,
+      "step": 1520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.3288,
+      "step": 1524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.3399,
+      "step": 1528
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.3304,
+      "step": 1532
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.3292,
+      "step": 1536
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.3275,
+      "step": 1540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.3314,
+      "step": 1544
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.3309,
+      "step": 1548
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.324,
+      "step": 1552
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.3266,
+      "step": 1556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.3302,
+      "step": 1560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.3302,
+      "step": 1564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.33,
+      "step": 1568
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.3286,
+      "step": 1572
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.3357,
+      "step": 1576
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.338,
+      "step": 1580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3262,
+      "step": 1584
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3325,
+      "step": 1588
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3291,
+      "step": 1592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.331,
+      "step": 1596
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.3248,
+      "step": 1600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.3297,
+      "step": 1604
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3301,
+      "step": 1608
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3239,
+      "step": 1612
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3303,
+      "step": 1616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.3364,
+      "step": 1620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.3307,
+      "step": 1624
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.3279,
+      "step": 1628
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.3328,
+      "step": 1632
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.3268,
+      "step": 1636
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.3262,
+      "step": 1640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.335,
+      "step": 1644
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.3238,
+      "step": 1648
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.3238,
+      "step": 1652
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.3269,
+      "step": 1656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.326,
+      "step": 1660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.3273,
+      "step": 1664
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.3357,
+      "step": 1668
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.3299,
+      "step": 1672
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.3313,
+      "step": 1676
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.3334,
+      "step": 1680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.3292,
+      "step": 1684
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.3291,
+      "step": 1688
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.3281,
+      "step": 1692
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.3285,
+      "step": 1696
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.3311,
+      "step": 1700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.3333,
+      "step": 1704
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.3357,
+      "step": 1708
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.3277,
+      "step": 1712
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.3235,
+      "step": 1716
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.3266,
+      "step": 1720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.3335,
+      "step": 1724
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.3316,
+      "step": 1728
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.3319,
+      "step": 1732
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.332,
+      "step": 1736
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.3213,
+      "step": 1740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.3335,
+      "step": 1744
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.3336,
+      "step": 1748
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.3335,
+      "step": 1752
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.3372,
+      "step": 1756
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.3254,
+      "step": 1760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.322,
+      "step": 1764
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.3319,
+      "step": 1768
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.3242,
+      "step": 1772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3264,
+      "step": 1776
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 0.3379732370376587,
+      "eval_runtime": 646.1781,
+      "eval_samples_per_second": 11.234,
+      "eval_steps_per_second": 0.176,
+      "step": 1779
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3282,
+      "step": 1780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3259,
+      "step": 1784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.3252,
+      "step": 1788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.3326,
+      "step": 1792
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.3323,
+      "step": 1796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.3252,
+      "step": 1800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.3279,
+      "step": 1804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.3278,
+      "step": 1808
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.3384,
+      "step": 1812
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.3223,
+      "step": 1816
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.3182,
+      "step": 1820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.3182,
+      "step": 1824
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.3271,
+      "step": 1828
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.326,
+      "step": 1832
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.328,
+      "step": 1836
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.3272,
+      "step": 1840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.3262,
+      "step": 1844
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.3271,
+      "step": 1848
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.3256,
+      "step": 1852
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.3281,
+      "step": 1856
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.3401,
+      "step": 1860
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.334,
+      "step": 1864
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.3207,
+      "step": 1868
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.3203,
+      "step": 1872
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.3256,
+      "step": 1876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.3248,
+      "step": 1880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.3227,
+      "step": 1884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.318,
+      "step": 1888
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.3241,
+      "step": 1892
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.3381,
+      "step": 1896
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.3308,
+      "step": 1900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.3323,
+      "step": 1904
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.3273,
+      "step": 1908
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.3199,
+      "step": 1912
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.3245,
+      "step": 1916
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.3289,
+      "step": 1920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.3216,
+      "step": 1924
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.322,
+      "step": 1928
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.3219,
+      "step": 1932
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.3255,
+      "step": 1936
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.3262,
+      "step": 1940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.3301,
+      "step": 1944
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.3239,
+      "step": 1948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.3297,
+      "step": 1952
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.3202,
+      "step": 1956
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.3309,
+      "step": 1960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.3298,
+      "step": 1964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.3217,
+      "step": 1968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.3302,
+      "step": 1972
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.3225,
+      "step": 1976
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.3213,
+      "step": 1980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.3323,
+      "step": 1984
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.3208,
+      "step": 1988
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.3286,
+      "step": 1992
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.3104,
+      "step": 1996
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.3336,
+      "step": 2000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.3141,
+      "step": 2004
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.3221,
+      "step": 2008
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.3196,
+      "step": 2012
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.325,
+      "step": 2016
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.3304,
+      "step": 2020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.3279,
+      "step": 2024
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.3193,
+      "step": 2028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.3271,
+      "step": 2032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.3297,
+      "step": 2036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.329,
+      "step": 2040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.3253,
+      "step": 2044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.3334,
+      "step": 2048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.3242,
+      "step": 2052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.3233,
+      "step": 2056
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.3167,
+      "step": 2060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.3221,
+      "step": 2064
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.3162,
+      "step": 2068
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.3232,
+      "step": 2072
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.3271,
+      "step": 2076
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.3226,
+      "step": 2080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.321,
+      "step": 2084
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.3308,
+      "step": 2088
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.3304,
+      "step": 2092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.325,
+      "step": 2096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.3304,
+      "step": 2100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.3215,
+      "step": 2104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.3255,
+      "step": 2108
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.3284,
+      "step": 2112
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.325,
+      "step": 2116
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.33,
+      "step": 2120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.3203,
+      "step": 2124
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.323,
+      "step": 2128
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.3181,
+      "step": 2132
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.3236,
+      "step": 2136
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.3185,
+      "step": 2140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.3261,
+      "step": 2144
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.3151,
+      "step": 2148
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.331,
+      "step": 2152
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.3242,
+      "step": 2156
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.3194,
+      "step": 2160
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.323,
+      "step": 2164
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.3242,
+      "step": 2168
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.3159,
+      "step": 2172
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.3246,
+      "step": 2176
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.3206,
+      "step": 2180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.3165,
+      "step": 2184
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.3281,
+      "step": 2188
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.3208,
+      "step": 2192
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.3303,
+      "step": 2196
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.3212,
+      "step": 2200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.3202,
+      "step": 2204
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.3195,
+      "step": 2208
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.322,
+      "step": 2212
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.3202,
+      "step": 2216
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.3141,
+      "step": 2220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.3181,
+      "step": 2224
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.3174,
+      "step": 2228
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.3227,
+      "step": 2232
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.3234,
+      "step": 2236
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.3179,
+      "step": 2240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.3223,
+      "step": 2244
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.3114,
+      "step": 2248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.325,
+      "step": 2252
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.325,
+      "step": 2256
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.3157,
+      "step": 2260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.3219,
+      "step": 2264
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.3228,
+      "step": 2268
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.3198,
+      "step": 2272
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.3186,
+      "step": 2276
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.3236,
+      "step": 2280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.3158,
+      "step": 2284
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.3253,
+      "step": 2288
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.3158,
+      "step": 2292
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.3205,
+      "step": 2296
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.3217,
+      "step": 2300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.3225,
+      "step": 2304
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.3281,
+      "step": 2308
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.3143,
+      "step": 2312
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.3278,
+      "step": 2316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.3218,
+      "step": 2320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.3163,
+      "step": 2324
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.3207,
+      "step": 2328
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.3185,
+      "step": 2332
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.3151,
+      "step": 2336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.3165,
+      "step": 2340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.3222,
+      "step": 2344
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.3227,
+      "step": 2348
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.3172,
+      "step": 2352
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.3224,
+      "step": 2356
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.3265,
+      "step": 2360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.3265,
+      "step": 2364
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.3224,
+      "step": 2368
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.3291,
+      "step": 2372
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.32955560088157654,
+      "eval_runtime": 645.3065,
+      "eval_samples_per_second": 11.249,
+      "eval_steps_per_second": 0.177,
+      "step": 2372
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.298,
+      "step": 2376
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.3048,
+      "step": 2380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.2921,
+      "step": 2384
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.2906,
+      "step": 2388
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.3,
+      "step": 2392
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.2883,
+      "step": 2396
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.2899,
+      "step": 2400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.2953,
+      "step": 2404
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.2899,
+      "step": 2408
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.2872,
+      "step": 2412
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3006,
+      "step": 2416
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.2929,
+      "step": 2420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.2885,
+      "step": 2424
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3013,
+      "step": 2428
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.2954,
+      "step": 2432
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.2915,
+      "step": 2436
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.2884,
+      "step": 2440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.2947,
+      "step": 2444
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.2975,
+      "step": 2448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.2934,
+      "step": 2452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.2987,
+      "step": 2456
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.2904,
+      "step": 2460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2988,
+      "step": 2464
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2949,
+      "step": 2468
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2929,
+      "step": 2472
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.298,
+      "step": 2476
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2984,
+      "step": 2480
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2926,
+      "step": 2484
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2837,
+      "step": 2488
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2942,
+      "step": 2492
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2941,
+      "step": 2496
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2918,
+      "step": 2500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2992,
+      "step": 2504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.293,
+      "step": 2508
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.294,
+      "step": 2512
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2869,
+      "step": 2516
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2966,
+      "step": 2520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2985,
+      "step": 2524
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.297,
+      "step": 2528
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2978,
+      "step": 2532
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3009,
+      "step": 2536
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.302,
+      "step": 2540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2927,
+      "step": 2544
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2933,
+      "step": 2548
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2954,
+      "step": 2552
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3045,
+      "step": 2556
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.302,
+      "step": 2560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2906,
+      "step": 2564
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3068,
+      "step": 2568
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.286,
+      "step": 2572
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2989,
+      "step": 2576
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2902,
+      "step": 2580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2991,
+      "step": 2584
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2934,
+      "step": 2588
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3047,
+      "step": 2592
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3044,
+      "step": 2596
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2977,
+      "step": 2600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3049,
+      "step": 2604
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2921,
+      "step": 2608
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2879,
+      "step": 2612
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2981,
+      "step": 2616
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2935,
+      "step": 2620
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2951,
+      "step": 2624
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2983,
+      "step": 2628
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2998,
+      "step": 2632
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2991,
+      "step": 2636
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2905,
+      "step": 2640
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2973,
+      "step": 2644
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2963,
+      "step": 2648
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2928,
+      "step": 2652
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2925,
+      "step": 2656
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2914,
+      "step": 2660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.3029,
+      "step": 2664
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2964,
+      "step": 2668
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2861,
+      "step": 2672
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2989,
+      "step": 2676
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2996,
+      "step": 2680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2934,
+      "step": 2684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.296,
+      "step": 2688
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2958,
+      "step": 2692
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3003,
+      "step": 2696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2977,
+      "step": 2700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2965,
+      "step": 2704
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.286,
+      "step": 2708
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2911,
+      "step": 2712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2904,
+      "step": 2716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2866,
+      "step": 2720
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2894,
+      "step": 2724
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2944,
+      "step": 2728
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.296,
+      "step": 2732
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.3065,
+      "step": 2736
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2965,
+      "step": 2740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2999,
+      "step": 2744
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2908,
+      "step": 2748
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.3011,
+      "step": 2752
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2997,
+      "step": 2756
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2979,
+      "step": 2760
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2992,
+      "step": 2764
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.3028,
+      "step": 2768
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2904,
+      "step": 2772
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2915,
+      "step": 2776
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3038,
+      "step": 2780
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2961,
+      "step": 2784
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2937,
+      "step": 2788
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.3016,
+      "step": 2792
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2986,
+      "step": 2796
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3021,
+      "step": 2800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2976,
+      "step": 2804
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2936,
+      "step": 2808
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2957,
+      "step": 2812
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2961,
+      "step": 2816
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2937,
+      "step": 2820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2985,
+      "step": 2824
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.3011,
+      "step": 2828
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2894,
+      "step": 2832
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2911,
+      "step": 2836
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2899,
+      "step": 2840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2986,
+      "step": 2844
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2929,
+      "step": 2848
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2976,
+      "step": 2852
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.299,
+      "step": 2856
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2992,
+      "step": 2860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.3013,
+      "step": 2864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2923,
+      "step": 2868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2893,
+      "step": 2872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2912,
+      "step": 2876
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2831,
+      "step": 2880
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2978,
+      "step": 2884
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.3036,
+      "step": 2888
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2983,
+      "step": 2892
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2932,
+      "step": 2896
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2922,
+      "step": 2900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2941,
+      "step": 2904
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.288,
+      "step": 2908
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2954,
+      "step": 2912
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2967,
+      "step": 2916
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.3025,
+      "step": 2920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2923,
+      "step": 2924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2846,
+      "step": 2928
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2931,
+      "step": 2932
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2917,
+      "step": 2936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.295,
+      "step": 2940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2987,
+      "step": 2944
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.298,
+      "step": 2948
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.297,
+      "step": 2952
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.299,
+      "step": 2956
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3002,
+      "step": 2960
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2951,
+      "step": 2964
+    },
+    {
+      "epoch": 2.5,
+      "eval_loss": 0.33106154203414917,
+      "eval_runtime": 644.7114,
+      "eval_samples_per_second": 11.259,
+      "eval_steps_per_second": 0.177,
+      "step": 2965
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2971,
+      "step": 2968
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2917,
+      "step": 2972
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.3033,
+      "step": 2976
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2991,
+      "step": 2980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.2802,
+      "step": 2984
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.2921,
+      "step": 2988
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.294,
+      "step": 2992
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2971,
+      "step": 2996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.3006,
+      "step": 3000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2989,
+      "step": 3004
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.3007,
+      "step": 3008
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.2941,
+      "step": 3012
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.3033,
+      "step": 3016
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2978,
+      "step": 3020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.305,
+      "step": 3024
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2857,
+      "step": 3028
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2924,
+      "step": 3032
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2938,
+      "step": 3036
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2987,
+      "step": 3040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.3055,
+      "step": 3044
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.303,
+      "step": 3048
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.2908,
+      "step": 3052
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.2899,
+      "step": 3056
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.3016,
+      "step": 3060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.2962,
+      "step": 3064
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.2893,
+      "step": 3068
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.3038,
+      "step": 3072
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.2975,
+      "step": 3076
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2883,
+      "step": 3080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.3057,
+      "step": 3084
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2951,
+      "step": 3088
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.2914,
+      "step": 3092
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.3083,
+      "step": 3096
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.3028,
+      "step": 3100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.293,
+      "step": 3104
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.2957,
+      "step": 3108
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.2904,
+      "step": 3112
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.2922,
+      "step": 3116
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.2955,
+      "step": 3120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.2947,
+      "step": 3124
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.297,
+      "step": 3128
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2935,
+      "step": 3132
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2889,
+      "step": 3136
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.2877,
+      "step": 3140
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.2899,
+      "step": 3144
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.2918,
+      "step": 3148
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2942,
+      "step": 3152
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2866,
+      "step": 3156
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2937,
+      "step": 3160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.2824,
+      "step": 3164
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.2947,
+      "step": 3168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.295,
+      "step": 3172
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.2876,
+      "step": 3176
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.2915,
+      "step": 3180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.2976,
+      "step": 3184
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2938,
+      "step": 3188
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2939,
+      "step": 3192
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2948,
+      "step": 3196
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.2897,
+      "step": 3200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.2902,
+      "step": 3204
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.2934,
+      "step": 3208
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.2958,
+      "step": 3212
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.2911,
+      "step": 3216
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.2905,
+      "step": 3220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.2909,
+      "step": 3224
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.296,
+      "step": 3228
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.2908,
+      "step": 3232
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.2961,
+      "step": 3236
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.297,
+      "step": 3240
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.2912,
+      "step": 3244
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.2913,
+      "step": 3248
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.2932,
+      "step": 3252
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.2899,
+      "step": 3256
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.2924,
+      "step": 3260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.29,
+      "step": 3264
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2988,
+      "step": 3268
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2952,
+      "step": 3272
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2856,
+      "step": 3276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.2915,
+      "step": 3280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.2926,
+      "step": 3284
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.2985,
+      "step": 3288
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.2933,
+      "step": 3292
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.2989,
+      "step": 3296
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.2924,
+      "step": 3300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.3032,
+      "step": 3304
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.292,
+      "step": 3308
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.2952,
+      "step": 3312
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.2863,
+      "step": 3316
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.2944,
+      "step": 3320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.2938,
+      "step": 3324
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.2916,
+      "step": 3328
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.2921,
+      "step": 3332
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.2889,
+      "step": 3336
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.3014,
+      "step": 3340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.2906,
+      "step": 3344
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.2923,
+      "step": 3348
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.2933,
+      "step": 3352
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.2872,
+      "step": 3356
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.2976,
+      "step": 3360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.2881,
+      "step": 3364
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.2938,
+      "step": 3368
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.2954,
+      "step": 3372
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.3019,
+      "step": 3376
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.2977,
+      "step": 3380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.2889,
+      "step": 3384
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.2902,
+      "step": 3388
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.2968,
+      "step": 3392
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.3023,
+      "step": 3396
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.2922,
+      "step": 3400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.2861,
+      "step": 3404
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.2951,
+      "step": 3408
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.2995,
+      "step": 3412
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.2873,
+      "step": 3416
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.2956,
+      "step": 3420
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.2997,
+      "step": 3424
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.2928,
+      "step": 3428
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.2989,
+      "step": 3432
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.2948,
+      "step": 3436
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.2807,
+      "step": 3440
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.2934,
+      "step": 3444
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.2943,
+      "step": 3448
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.2956,
+      "step": 3452
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.2961,
+      "step": 3456
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.2979,
+      "step": 3460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.3074,
+      "step": 3464
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.297,
+      "step": 3468
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.2938,
+      "step": 3472
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.2981,
+      "step": 3476
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.2928,
+      "step": 3480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.2872,
+      "step": 3484
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.2968,
+      "step": 3488
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.2858,
+      "step": 3492
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.2803,
+      "step": 3496
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.2941,
+      "step": 3500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.2844,
+      "step": 3504
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.2939,
+      "step": 3508
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.2812,
+      "step": 3512
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.2923,
+      "step": 3516
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.2945,
+      "step": 3520
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.2916,
+      "step": 3524
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.3008,
+      "step": 3528
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.3018,
+      "step": 3532
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.2934,
+      "step": 3536
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.2944,
+      "step": 3540
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.2889,
+      "step": 3544
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.2947,
+      "step": 3548
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.2967,
+      "step": 3552
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.2927,
+      "step": 3556
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.3247070908546448,
+      "eval_runtime": 644.9407,
+      "eval_samples_per_second": 11.255,
+      "eval_steps_per_second": 0.177,
+      "step": 3558
+    }
+  ],
+  "max_steps": 4744,
+  "num_train_epochs": 4,
+  "total_flos": 795536966025216.0,
+  "trial_name": null,
+  "trial_params": null
+}