diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18084 @@
+{
+  "best_metric": 1.9482014179229736,
+  "best_model_checkpoint": "saves/Custom/lora/llama2-Medical-Medtext-28-10/checkpoint-14000",
+  "epoch": 1.9999324529703806,
+  "eval_steps": 400,
+  "global_step": 14804,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99932450689003e-06,
+      "loss": 2.8148,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9986490137800594e-06,
+      "loss": 2.6944,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9979735206700893e-06,
+      "loss": 2.688,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.997298027560119e-06,
+      "loss": 2.8038,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9966225344501488e-06,
+      "loss": 2.8559,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9959470413401783e-06,
+      "loss": 2.7153,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.995271548230208e-06,
+      "loss": 2.7136,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9945960551202377e-06,
+      "loss": 2.7289,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9939205620102677e-06,
+      "loss": 2.7078,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.993245068900297e-06,
+      "loss": 2.6738,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.992569575790327e-06,
+      "loss": 2.6968,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9918940826803566e-06,
+      "loss": 2.7469,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9912185895703866e-06,
+      "loss": 2.628,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.990543096460416e-06,
+      "loss": 2.7036,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9898676033504456e-06,
+      "loss": 2.5699,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9891921102404755e-06,
+      "loss": 2.6844,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.988516617130505e-06,
+      "loss": 2.6691,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.987841124020535e-06,
+      "loss": 2.708,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9871656309105645e-06,
+      "loss": 2.7602,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9864901378005944e-06,
+      "loss": 2.8753,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.985814644690624e-06,
+      "loss": 2.6093,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.985139151580654e-06,
+      "loss": 2.7309,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9844636584706834e-06,
+      "loss": 2.6666,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9837881653607133e-06,
+      "loss": 2.6702,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.983112672250743e-06,
+      "loss": 2.7468,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9824371791407724e-06,
+      "loss": 2.6879,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9817616860308023e-06,
+      "loss": 2.6141,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.981086192920832e-06,
+      "loss": 2.6581,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9804106998108618e-06,
+      "loss": 2.5732,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9797352067008917e-06,
+      "loss": 2.6452,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9790597135909212e-06,
+      "loss": 2.6152,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.978384220480951e-06,
+      "loss": 2.5607,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9777087273709807e-06,
+      "loss": 2.6272,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9770332342610106e-06,
+      "loss": 2.7238,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.97635774115104e-06,
+      "loss": 2.6797,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.97568224804107e-06,
+      "loss": 2.6734,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9750067549310996e-06,
+      "loss": 2.5306,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743312618211295e-06,
+      "loss": 2.5373,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.973655768711159e-06,
+      "loss": 2.5462,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972980275601189e-06,
+      "loss": 2.6225,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9723047824912185e-06,
+      "loss": 2.5914,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9716292893812484e-06,
+      "loss": 2.5997,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.970953796271278e-06,
+      "loss": 2.5858,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.970278303161308e-06,
+      "loss": 2.6066,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9696028100513374e-06,
+      "loss": 2.4575,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9689273169413673e-06,
+      "loss": 2.5655,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.968251823831397e-06,
+      "loss": 2.6101,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9675763307214268e-06,
+      "loss": 2.4752,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9669008376114563e-06,
+      "loss": 2.5649,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9662253445014862e-06,
+      "loss": 2.4658,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9655498513915157e-06,
+      "loss": 2.6025,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9648743582815453e-06,
+      "loss": 2.528,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.964198865171575e-06,
+      "loss": 2.4624,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9635233720616047e-06,
+      "loss": 2.5285,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9628478789516346e-06,
+      "loss": 2.5167,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.962172385841664e-06,
+      "loss": 2.4256,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.961496892731694e-06,
+      "loss": 2.4909,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9608213996217236e-06,
+      "loss": 2.4507,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9601459065117535e-06,
+      "loss": 2.5089,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.959470413401783e-06,
+      "loss": 2.4909,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.958794920291813e-06,
+      "loss": 2.3888,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9581194271818425e-06,
+      "loss": 2.3279,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9574439340718725e-06,
+      "loss": 2.3981,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.956768440961902e-06,
+      "loss": 2.4836,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.956092947851932e-06,
+      "loss": 2.4944,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9554174547419614e-06,
+      "loss": 2.5004,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9547419616319914e-06,
+      "loss": 2.3797,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.954066468522021e-06,
+      "loss": 2.4652,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.953390975412051e-06,
+      "loss": 2.4376,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9527154823020803e-06,
+      "loss": 2.4507,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9520399891921103e-06,
+      "loss": 2.4931,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9513644960821398e-06,
+      "loss": 2.5258,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9506890029721697e-06,
+      "loss": 2.3549,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9500135098621992e-06,
+      "loss": 2.4138,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.949338016752229e-06,
+      "loss": 2.4262,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9486625236422587e-06,
+      "loss": 2.3892,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9479870305322886e-06,
+      "loss": 2.275,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.947311537422318e-06,
+      "loss": 2.4148,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.946636044312348e-06,
+      "loss": 2.3715,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9459605512023776e-06,
+      "loss": 2.3765,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.380300521850586,
+      "eval_runtime": 165.4903,
+      "eval_samples_per_second": 3.601,
+      "eval_steps_per_second": 0.453,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9452850580924075e-06,
+      "loss": 2.4244,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.944609564982437e-06,
+      "loss": 2.3635,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.943934071872467e-06,
+      "loss": 2.3548,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9432585787624965e-06,
+      "loss": 2.3199,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9425830856525264e-06,
+      "loss": 2.3976,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.941907592542556e-06,
+      "loss": 2.4299,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.941232099432586e-06,
+      "loss": 2.3214,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9405566063226154e-06,
+      "loss": 2.3571,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9398811132126453e-06,
+      "loss": 2.3554,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.939205620102675e-06,
+      "loss": 2.3911,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9385301269927044e-06,
+      "loss": 2.3898,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9378546338827343e-06,
+      "loss": 2.3447,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.937179140772764e-06,
+      "loss": 2.3268,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9365036476627938e-06,
+      "loss": 2.3788,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9358281545528233e-06,
+      "loss": 2.3091,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.935152661442853e-06,
+      "loss": 2.289,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9344771683328827e-06,
+      "loss": 2.3099,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9338016752229127e-06,
+      "loss": 2.2789,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.933126182112942e-06,
+      "loss": 2.2609,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.932450689002972e-06,
+      "loss": 2.3611,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9317751958930016e-06,
+      "loss": 2.2964,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9310997027830316e-06,
+      "loss": 2.3029,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.930424209673061e-06,
+      "loss": 2.3109,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.929748716563091e-06,
+      "loss": 2.4029,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9290732234531205e-06,
+      "loss": 2.2692,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9283977303431505e-06,
+      "loss": 2.4243,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.92772223723318e-06,
+      "loss": 2.3542,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.92704674412321e-06,
+      "loss": 2.3045,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9263712510132394e-06,
+      "loss": 2.3004,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9256957579032694e-06,
+      "loss": 2.359,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.925020264793299e-06,
+      "loss": 2.3057,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.924344771683329e-06,
+      "loss": 2.3865,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9236692785733583e-06,
+      "loss": 2.2997,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9229937854633883e-06,
+      "loss": 2.2846,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.922318292353418e-06,
+      "loss": 2.3284,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9216427992434477e-06,
+      "loss": 2.258,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9209673061334777e-06,
+      "loss": 2.3458,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.920291813023507e-06,
+      "loss": 2.276,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9196163199135367e-06,
+      "loss": 2.2503,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9189408268035666e-06,
+      "loss": 2.2727,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.918265333693596e-06,
+      "loss": 2.289,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.917589840583626e-06,
+      "loss": 2.2968,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9169143474736556e-06,
+      "loss": 2.3512,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9162388543636855e-06,
+      "loss": 2.3458,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.915563361253715e-06,
+      "loss": 2.2532,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.914887868143745e-06,
+      "loss": 2.2273,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9142123750337745e-06,
+      "loss": 2.3108,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9135368819238044e-06,
+      "loss": 2.3238,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.912861388813834e-06,
+      "loss": 2.2053,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9121858957038635e-06,
+      "loss": 2.386,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9115104025938934e-06,
+      "loss": 2.1914,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.910834909483923e-06,
+      "loss": 2.2638,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.910159416373953e-06,
+      "loss": 2.3571,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9094839232639824e-06,
+      "loss": 2.344,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9088084301540123e-06,
+      "loss": 2.2765,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.908132937044042e-06,
+      "loss": 2.2643,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9074574439340718e-06,
+      "loss": 2.2843,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9067819508241015e-06,
+      "loss": 2.3156,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9061064577141312e-06,
+      "loss": 2.2365,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.905430964604161e-06,
+      "loss": 2.2396,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9047554714941907e-06,
+      "loss": 2.3003,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9040799783842202e-06,
+      "loss": 2.2414,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9034044852742501e-06,
+      "loss": 2.2087,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9027289921642796e-06,
+      "loss": 2.3138,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9020534990543096e-06,
+      "loss": 2.2667,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9013780059443393e-06,
+      "loss": 2.171,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.900702512834369e-06,
+      "loss": 2.2188,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9000270197243988e-06,
+      "loss": 2.2388,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8993515266144285e-06,
+      "loss": 2.2639,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8986760335044582e-06,
+      "loss": 2.1737,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.898000540394488e-06,
+      "loss": 2.2252,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8973250472845177e-06,
+      "loss": 2.1938,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8966495541745474e-06,
+      "loss": 2.2139,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8959740610645771e-06,
+      "loss": 2.2717,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8952985679546068e-06,
+      "loss": 2.2766,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8946230748446366e-06,
+      "loss": 2.2662,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.893947581734666e-06,
+      "loss": 2.1394,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.893272088624696e-06,
+      "loss": 2.187,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8925965955147255e-06,
+      "loss": 2.2881,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8919211024047555e-06,
+      "loss": 2.2189,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.2260634899139404,
+      "eval_runtime": 165.4618,
+      "eval_samples_per_second": 3.602,
+      "eval_steps_per_second": 0.453,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.891245609294785e-06,
+      "loss": 2.2459,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.890570116184815e-06,
+      "loss": 2.1488,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8898946230748444e-06,
+      "loss": 2.2133,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8892191299648744e-06,
+      "loss": 2.2117,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.888543636854904e-06,
+      "loss": 2.1795,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8878681437449338e-06,
+      "loss": 2.2757,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8871926506349634e-06,
+      "loss": 2.2471,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8865171575249933e-06,
+      "loss": 2.3356,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8858416644150228e-06,
+      "loss": 2.263,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8851661713050525e-06,
+      "loss": 2.2264,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8844906781950823e-06,
+      "loss": 2.2703,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.883815185085112e-06,
+      "loss": 2.2639,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8831396919751417e-06,
+      "loss": 2.2495,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8824641988651714e-06,
+      "loss": 2.2448,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8817887057552014e-06,
+      "loss": 2.2363,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8811132126452309e-06,
+      "loss": 2.1892,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8804377195352608e-06,
+      "loss": 2.1822,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8797622264252903e-06,
+      "loss": 2.1871,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8790867333153203e-06,
+      "loss": 2.1882,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8784112402053498e-06,
+      "loss": 2.1884,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8777357470953797e-06,
+      "loss": 2.2064,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8770602539854092e-06,
+      "loss": 2.249,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.876384760875439e-06,
+      "loss": 2.2484,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8757092677654687e-06,
+      "loss": 2.1547,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8750337746554984e-06,
+      "loss": 2.177,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8743582815455281e-06,
+      "loss": 2.1332,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8736827884355579e-06,
+      "loss": 2.2354,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8730072953255876e-06,
+      "loss": 2.1353,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8723318022156173e-06,
+      "loss": 2.2044,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.871656309105647e-06,
+      "loss": 2.2369,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8709808159956768e-06,
+      "loss": 2.1856,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8703053228857065e-06,
+      "loss": 2.1692,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8696298297757362e-06,
+      "loss": 2.2238,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8689543366657657e-06,
+      "loss": 2.2798,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8682788435557957e-06,
+      "loss": 2.2296,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8676033504458252e-06,
+      "loss": 2.2336,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8669278573358551e-06,
+      "loss": 2.2287,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8662523642258847e-06,
+      "loss": 2.244,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8655768711159146e-06,
+      "loss": 2.1761,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8649013780059441e-06,
+      "loss": 2.2065,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.864225884895974e-06,
+      "loss": 2.1739,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8635503917860036e-06,
+      "loss": 2.1753,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8628748986760335e-06,
+      "loss": 2.3107,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8621994055660632e-06,
+      "loss": 2.214,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.861523912456093e-06,
+      "loss": 2.1209,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8608484193461227e-06,
+      "loss": 2.1574,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8601729262361522e-06,
+      "loss": 2.2181,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8594974331261821e-06,
+      "loss": 2.2101,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8588219400162116e-06,
+      "loss": 2.196,
+      "step": 1045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8581464469062416e-06,
+      "loss": 2.2107,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.857470953796271e-06,
+      "loss": 2.1913,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.856795460686301e-06,
+      "loss": 2.1373,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8561199675763305e-06,
+      "loss": 2.0939,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8554444744663605e-06,
+      "loss": 2.2936,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.85476898135639e-06,
+      "loss": 2.11,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.85409348824642e-06,
+      "loss": 2.2067,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8534179951364495e-06,
+      "loss": 2.0939,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8527425020264794e-06,
+      "loss": 2.1118,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.852067008916509e-06,
+      "loss": 2.2529,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8513915158065388e-06,
+      "loss": 2.1717,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8507160226965684e-06,
+      "loss": 2.1909,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.850040529586598e-06,
+      "loss": 2.1772,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8493650364766278e-06,
+      "loss": 2.1615,
+      "step": 1115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8486895433666575e-06,
+      "loss": 2.1519,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8480140502566873e-06,
+      "loss": 2.1842,
+      "step": 1125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.847338557146717e-06,
+      "loss": 2.1566,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8466630640367467e-06,
+      "loss": 2.2503,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8459875709267764e-06,
+      "loss": 2.1565,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8453120778168062e-06,
+      "loss": 2.1437,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.844636584706836e-06,
+      "loss": 2.1883,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8439610915968656e-06,
+      "loss": 2.1408,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8432855984868953e-06,
+      "loss": 2.2447,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8426101053769249e-06,
+      "loss": 2.1307,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8419346122669548e-06,
+      "loss": 2.132,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8412591191569845e-06,
+      "loss": 2.2312,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8405836260470143e-06,
+      "loss": 2.1841,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.839908132937044e-06,
+      "loss": 2.1887,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8392326398270737e-06,
+      "loss": 2.0504,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8385571467171034e-06,
+      "loss": 2.167,
+      "step": 1195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8378816536071332e-06,
+      "loss": 2.1824,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.1631150245666504,
+      "eval_runtime": 165.4999,
+      "eval_samples_per_second": 3.601,
+      "eval_steps_per_second": 0.453,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8372061604971629e-06,
+      "loss": 2.2233,
+      "step": 1205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8365306673871926e-06,
+      "loss": 2.2954,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8358551742772223e-06,
+      "loss": 2.2049,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.835179681167252e-06,
+      "loss": 2.1491,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8345041880572818e-06,
+      "loss": 2.1174,
+      "step": 1225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8338286949473113e-06,
+      "loss": 2.1562,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8331532018373412e-06,
+      "loss": 2.1639,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8324777087273708e-06,
+      "loss": 2.1512,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8318022156174007e-06,
+      "loss": 2.1613,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8311267225074302e-06,
+      "loss": 2.1649,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8304512293974601e-06,
+      "loss": 2.1872,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8297757362874897e-06,
+      "loss": 2.1949,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8291002431775196e-06,
+      "loss": 2.1656,
+      "step": 1265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8284247500675491e-06,
+      "loss": 2.1845,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.827749256957579e-06,
+      "loss": 2.1477,
+      "step": 1275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8270737638476086e-06,
+      "loss": 2.2332,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8263982707376385e-06,
+      "loss": 2.1762,
+      "step": 1285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.825722777627668e-06,
+      "loss": 2.2039,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8250472845176977e-06,
+      "loss": 2.192,
+      "step": 1295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8243717914077275e-06,
+      "loss": 2.2305,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8236962982977572e-06,
+      "loss": 2.063,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.823020805187787e-06,
+      "loss": 2.1416,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8223453120778166e-06,
+      "loss": 2.082,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8216698189678466e-06,
+      "loss": 2.2055,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.820994325857876e-06,
+      "loss": 2.2108,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.820318832747906e-06,
+      "loss": 2.2147,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8196433396379356e-06,
+      "loss": 2.2602,
+      "step": 1335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8189678465279655e-06,
+      "loss": 2.1456,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.818292353417995e-06,
+      "loss": 2.1897,
+      "step": 1345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.817616860308025e-06,
+      "loss": 2.1242,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8169413671980545e-06,
+      "loss": 2.1346,
+      "step": 1355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8162658740880844e-06,
+      "loss": 2.1532,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.815590380978114e-06,
+      "loss": 2.1337,
+      "step": 1365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8149148878681436e-06,
+      "loss": 2.1347,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8142393947581734e-06,
+      "loss": 2.1518,
+      "step": 1375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.813563901648203e-06,
+      "loss": 2.085,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8128884085382328e-06,
+      "loss": 2.2308,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8122129154282625e-06,
+      "loss": 2.189,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8115374223182923e-06,
+      "loss": 2.1436,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.810861929208322e-06,
+      "loss": 2.1599,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8101864360983517e-06,
+      "loss": 2.1162,
+      "step": 1405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8095109429883814e-06,
+      "loss": 2.1438,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8088354498784112e-06,
+      "loss": 2.1114,
+      "step": 1415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.808159956768441e-06,
+      "loss": 2.1274,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8074844636584704e-06,
+      "loss": 2.0931,
+      "step": 1425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8068089705485004e-06,
+      "loss": 2.1427,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8061334774385299e-06,
+      "loss": 2.1255,
+      "step": 1435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8054579843285598e-06,
+      "loss": 2.2059,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8047824912185893e-06,
+      "loss": 2.1694,
+      "step": 1445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8041069981086193e-06,
+      "loss": 2.0896,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8034315049986488e-06,
+      "loss": 2.1429,
+      "step": 1455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8027560118886787e-06,
+      "loss": 2.2017,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8020805187787084e-06,
+      "loss": 2.1535,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8014050256687382e-06,
+      "loss": 2.1342,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8007295325587679e-06,
+      "loss": 2.1528,
+      "step": 1475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8000540394487976e-06,
+      "loss": 2.1151,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7993785463388273e-06,
+      "loss": 2.1482,
+      "step": 1485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7987030532288569e-06,
+      "loss": 2.0806,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7980275601188868e-06,
+      "loss": 2.1843,
+      "step": 1495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7973520670089163e-06,
+      "loss": 2.1359,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7966765738989462e-06,
+      "loss": 2.1303,
+      "step": 1505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7960010807889758e-06,
+      "loss": 2.1373,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7953255876790057e-06,
+      "loss": 2.1413,
+      "step": 1515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7946500945690352e-06,
+      "loss": 2.1424,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7939746014590652e-06,
+      "loss": 2.1073,
+      "step": 1525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7932991083490947e-06,
+      "loss": 2.0807,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7926236152391246e-06,
+      "loss": 2.1708,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7919481221291541e-06,
+      "loss": 2.1456,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791272629019184e-06,
+      "loss": 2.2062,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7905971359092136e-06,
+      "loss": 2.1719,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7899216427992433e-06,
+      "loss": 2.1313,
+      "step": 1555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789246149689273e-06,
+      "loss": 2.0904,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7885706565793028e-06,
+      "loss": 2.1181,
+      "step": 1565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7878951634693325e-06,
+      "loss": 2.236,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7872196703593622e-06,
+      "loss": 2.2107,
+      "step": 1575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786544177249392e-06,
+      "loss": 2.1614,
+      "step": 1580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7858686841394217e-06,
+      "loss": 2.0653,
+      "step": 1585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7851931910294514e-06,
+      "loss": 2.1127,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7845176979194811e-06,
+      "loss": 2.1073,
+      "step": 1595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7838422048095108e-06,
+      "loss": 2.0995,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 2.1323931217193604,
+      "eval_runtime": 165.5394,
+      "eval_samples_per_second": 3.6,
+      "eval_steps_per_second": 0.453,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7831667116995406e-06,
+      "loss": 2.1629,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7824912185895705e-06,
+      "loss": 2.1419,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7818157254796e-06,
+      "loss": 2.1061,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78114023236963e-06,
+      "loss": 2.1188,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7804647392596595e-06,
+      "loss": 2.1862,
+      "step": 1625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7797892461496892e-06,
+      "loss": 2.153,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779113753039719e-06,
+      "loss": 2.0605,
+      "step": 1635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7784382599297486e-06,
+      "loss": 2.2079,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7777627668197784e-06,
+      "loss": 2.1516,
+      "step": 1645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777087273709808e-06,
+      "loss": 2.1234,
+      "step": 1650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7764117805998378e-06,
+      "loss": 2.231,
+      "step": 1655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7757362874898676e-06,
+      "loss": 2.1475,
+      "step": 1660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7750607943798973e-06,
+      "loss": 2.1289,
+      "step": 1665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.774385301269927e-06,
+      "loss": 2.0901,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7737098081599567e-06,
+      "loss": 2.17,
+      "step": 1675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7730343150499865e-06,
+      "loss": 2.2153,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772358821940016e-06,
+      "loss": 2.1964,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.771683328830046e-06,
+      "loss": 2.2056,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7710078357200754e-06,
+      "loss": 2.0911,
+      "step": 1695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7703323426101054e-06,
+      "loss": 2.1362,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7696568495001349e-06,
+      "loss": 2.1819,
+      "step": 1705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7689813563901648e-06,
+      "loss": 2.155,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7683058632801943e-06,
+      "loss": 2.0689,
+      "step": 1715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7676303701702243e-06,
+      "loss": 2.1661,
+      "step": 1720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7669548770602538e-06,
+      "loss": 2.094,
+      "step": 1725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7662793839502837e-06,
+      "loss": 2.061,
+      "step": 1730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7656038908403132e-06,
+      "loss": 2.185,
+      "step": 1735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7649283977303432e-06,
+      "loss": 2.1631,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7642529046203727e-06,
+      "loss": 2.1535,
+      "step": 1745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7635774115104024e-06,
+      "loss": 2.2405,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7629019184004324e-06,
+      "loss": 2.1495,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7622264252904619e-06,
+      "loss": 2.1073,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7615509321804918e-06,
+      "loss": 2.1346,
+      "step": 1765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7608754390705213e-06,
+      "loss": 2.0585,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7601999459605513e-06,
+      "loss": 2.1566,
+      "step": 1775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7595244528505808e-06,
+      "loss": 2.1382,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7588489597406107e-06,
+      "loss": 2.1801,
+      "step": 1785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7581734666306402e-06,
+      "loss": 2.1169,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7574979735206702e-06,
+      "loss": 2.1017,
+      "step": 1795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7568224804106997e-06,
+      "loss": 2.1697,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7561469873007296e-06,
+      "loss": 2.0574,
+      "step": 1805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7554714941907591e-06,
+      "loss": 2.1086,
+      "step": 1810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547960010807889e-06,
+      "loss": 2.1359,
+      "step": 1815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541205079708186e-06,
+      "loss": 2.178,
+      "step": 1820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534450148608483e-06,
+      "loss": 2.1483,
+      "step": 1825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752769521750878e-06,
+      "loss": 2.1726,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520940286409078e-06,
+      "loss": 2.0898,
+      "step": 1835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514185355309375e-06,
+      "loss": 2.0657,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507430424209672e-06,
+      "loss": 2.1823,
+      "step": 1845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750067549310997e-06,
+      "loss": 2.0883,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493920562010267e-06,
+      "loss": 2.1784,
+      "step": 1855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487165630910564e-06,
+      "loss": 2.0749,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480410699810861e-06,
+      "loss": 2.1461,
+      "step": 1865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473655768711158e-06,
+      "loss": 2.2457,
+      "step": 1870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466900837611456e-06,
+      "loss": 2.1692,
+      "step": 1875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746014590651175e-06,
+      "loss": 2.1989,
+      "step": 1880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.745339097541205e-06,
+      "loss": 2.194,
+      "step": 1885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446636044312345e-06,
+      "loss": 2.1568,
+      "step": 1890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439881113212645e-06,
+      "loss": 2.0265,
+      "step": 1895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433126182112942e-06,
+      "loss": 2.165,
+      "step": 1900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742637125101324e-06,
+      "loss": 2.1315,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419616319913537e-06,
+      "loss": 2.052,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412861388813834e-06,
+      "loss": 2.151,
+      "step": 1915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740610645771413e-06,
+      "loss": 2.1216,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399351526614428e-06,
+      "loss": 2.2173,
+      "step": 1925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392596595514726e-06,
+      "loss": 2.0417,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385841664415023e-06,
+      "loss": 2.1326,
+      "step": 1935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737908673331532e-06,
+      "loss": 2.0473,
+      "step": 1940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372331802215615e-06,
+      "loss": 2.1349,
+      "step": 1945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365576871115915e-06,
+      "loss": 2.0019,
+      "step": 1950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735882194001621e-06,
+      "loss": 2.1083,
+      "step": 1955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735206700891651e-06,
+      "loss": 2.0901,
+      "step": 1960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7345312077816804e-06,
+      "loss": 2.1184,
+      "step": 1965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7338557146717104e-06,
+      "loss": 2.121,
+      "step": 1970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7331802215617399e-06,
+      "loss": 1.9879,
+      "step": 1975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7325047284517698e-06,
+      "loss": 2.1164,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7318292353417993e-06,
+      "loss": 2.118,
+      "step": 1985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7311537422318293e-06,
+      "loss": 2.1429,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7304782491218588e-06,
+      "loss": 2.1996,
+      "step": 1995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7298027560118887e-06,
+      "loss": 2.1266,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 2.1100127696990967,
+      "eval_runtime": 165.3022,
+      "eval_samples_per_second": 3.606,
+      "eval_steps_per_second": 0.454,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7291272629019182e-06,
+      "loss": 2.2143,
+      "step": 2005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.728451769791948e-06,
+      "loss": 2.1678,
+      "step": 2010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7277762766819777e-06,
+      "loss": 2.1279,
+      "step": 2015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7271007835720074e-06,
+      "loss": 2.2299,
+      "step": 2020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7264252904620371e-06,
+      "loss": 2.1529,
+      "step": 2025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7257497973520669e-06,
+      "loss": 2.1269,
+      "step": 2030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7250743042420966e-06,
+      "loss": 2.1651,
+      "step": 2035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7243988111321263e-06,
+      "loss": 2.0818,
+      "step": 2040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7237233180221563e-06,
+      "loss": 2.0841,
+      "step": 2045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7230478249121858e-06,
+      "loss": 2.1054,
+      "step": 2050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7223723318022157e-06,
+      "loss": 2.1279,
+      "step": 2055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7216968386922452e-06,
+      "loss": 2.1461,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7210213455822752e-06,
+      "loss": 2.2039,
+      "step": 2065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7203458524723047e-06,
+      "loss": 2.2047,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7196703593623344e-06,
+      "loss": 2.11,
+      "step": 2075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7189948662523641e-06,
+      "loss": 2.1663,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7183193731423939e-06,
+      "loss": 2.0122,
+      "step": 2085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7176438800324236e-06,
+      "loss": 2.1258,
+      "step": 2090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7169683869224533e-06,
+      "loss": 2.103,
+      "step": 2095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.716292893812483e-06,
+      "loss": 2.181,
+      "step": 2100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7156174007025128e-06,
+      "loss": 2.1502,
+      "step": 2105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7149419075925425e-06,
+      "loss": 2.1905,
+      "step": 2110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7142664144825722e-06,
+      "loss": 2.1424,
+      "step": 2115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.713590921372602e-06,
+      "loss": 2.1408,
+      "step": 2120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7129154282626317e-06,
+      "loss": 2.0876,
+      "step": 2125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7122399351526614e-06,
+      "loss": 2.1308,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7115644420426911e-06,
+      "loss": 2.13,
+      "step": 2135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7108889489327206e-06,
+      "loss": 2.1297,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7102134558227506e-06,
+      "loss": 2.146,
+      "step": 2145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.70953796271278e-06,
+      "loss": 2.1063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.70886246960281e-06,
+      "loss": 2.1605,
+      "step": 2155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7081869764928395e-06,
+      "loss": 2.1093,
+      "step": 2160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7075114833828695e-06,
+      "loss": 2.1206,
+      "step": 2165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.706835990272899e-06,
+      "loss": 2.076,
+      "step": 2170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.706160497162929e-06,
+      "loss": 2.1505,
+      "step": 2175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7054850040529584e-06,
+      "loss": 2.1548,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7048095109429884e-06,
+      "loss": 2.098,
+      "step": 2185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7041340178330181e-06,
+      "loss": 2.1067,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7034585247230478e-06,
+      "loss": 2.0598,
+      "step": 2195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7027830316130776e-06,
+      "loss": 2.1151,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.702107538503107e-06,
+      "loss": 2.0832,
+      "step": 2205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.701432045393137e-06,
+      "loss": 2.1308,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7007565522831665e-06,
+      "loss": 2.1276,
+      "step": 2215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7000810591731965e-06,
+      "loss": 2.0819,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.699405566063226e-06,
+      "loss": 2.113,
+      "step": 2225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.698730072953256e-06,
+      "loss": 2.1185,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6980545798432854e-06,
+      "loss": 2.1705,
+      "step": 2235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6973790867333154e-06,
+      "loss": 2.1164,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6967035936233449e-06,
+      "loss": 2.0782,
+      "step": 2245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6960281005133748e-06,
+      "loss": 2.1195,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6953526074034043e-06,
+      "loss": 2.0733,
+      "step": 2255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6946771142934343e-06,
+      "loss": 2.1518,
+      "step": 2260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6940016211834638e-06,
+      "loss": 2.0086,
+      "step": 2265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6933261280734935e-06,
+      "loss": 2.0694,
+      "step": 2270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6926506349635232e-06,
+      "loss": 2.1871,
+      "step": 2275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.691975141853553e-06,
+      "loss": 2.1151,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6912996487435827e-06,
+      "loss": 2.1302,
+      "step": 2285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6906241556336124e-06,
+      "loss": 2.0942,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6899486625236422e-06,
+      "loss": 2.1311,
+      "step": 2295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6892731694136719e-06,
+      "loss": 2.0993,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6885976763037016e-06,
+      "loss": 2.1066,
+      "step": 2305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6879221831937313e-06,
+      "loss": 2.1216,
+      "step": 2310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.687246690083761e-06,
+      "loss": 2.0656,
+      "step": 2315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6865711969737908e-06,
+      "loss": 2.0993,
+      "step": 2320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6858957038638205e-06,
+      "loss": 2.0928,
+      "step": 2325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6852202107538502e-06,
+      "loss": 2.1319,
+      "step": 2330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.68454471764388e-06,
+      "loss": 2.1125,
+      "step": 2335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6838692245339097e-06,
+      "loss": 1.9319,
+      "step": 2340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6831937314239394e-06,
+      "loss": 2.1942,
+      "step": 2345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6825182383139691e-06,
+      "loss": 2.1001,
+      "step": 2350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6818427452039989e-06,
+      "loss": 2.1189,
+      "step": 2355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6811672520940286e-06,
+      "loss": 2.1774,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6804917589840583e-06,
+      "loss": 2.1566,
+      "step": 2365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.679816265874088e-06,
+      "loss": 2.1077,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6791407727641178e-06,
+      "loss": 1.994,
+      "step": 2375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6784652796541475e-06,
+      "loss": 2.1114,
+      "step": 2380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6777897865441772e-06,
+      "loss": 2.1286,
+      "step": 2385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.677114293434207e-06,
+      "loss": 2.1445,
+      "step": 2390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6764388003242367e-06,
+      "loss": 2.0908,
+      "step": 2395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6757633072142662e-06,
+      "loss": 2.0279,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 2.090486764907837,
+      "eval_runtime": 165.4313,
+      "eval_samples_per_second": 3.603,
+      "eval_steps_per_second": 0.453,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6750878141042961e-06,
+      "loss": 2.0959,
+      "step": 2405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6744123209943256e-06,
+      "loss": 2.1049,
+      "step": 2410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6737368278843556e-06,
+      "loss": 2.1089,
+      "step": 2415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.673061334774385e-06,
+      "loss": 2.1387,
+      "step": 2420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.672385841664415e-06,
+      "loss": 2.1712,
+      "step": 2425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6717103485544446e-06,
+      "loss": 2.0841,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6710348554444745e-06,
+      "loss": 2.1344,
+      "step": 2435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.670359362334504e-06,
+      "loss": 2.0958,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.669683869224534e-06,
+      "loss": 2.0917,
+      "step": 2445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6690083761145635e-06,
+      "loss": 2.0386,
+      "step": 2450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6683328830045934e-06,
+      "loss": 2.0025,
+      "step": 2455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.667657389894623e-06,
+      "loss": 2.1355,
+      "step": 2460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6669818967846526e-06,
+      "loss": 2.1008,
+      "step": 2465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6663064036746824e-06,
+      "loss": 2.0849,
+      "step": 2470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.665630910564712e-06,
+      "loss": 2.1699,
+      "step": 2475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.664955417454742e-06,
+      "loss": 2.0264,
+      "step": 2480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6642799243447715e-06,
+      "loss": 2.0421,
+      "step": 2485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6636044312348015e-06,
+      "loss": 2.1003,
+      "step": 2490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.662928938124831e-06,
+      "loss": 2.1109,
+      "step": 2495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.662253445014861e-06,
+      "loss": 2.1525,
+      "step": 2500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6615779519048904e-06,
+      "loss": 2.0433,
+      "step": 2505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6609024587949204e-06,
+      "loss": 2.0748,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.66022696568495e-06,
+      "loss": 2.1392,
+      "step": 2515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6595514725749798e-06,
+      "loss": 2.0972,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6588759794650093e-06,
+      "loss": 2.1862,
+      "step": 2525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.658200486355039e-06,
+      "loss": 2.1236,
+      "step": 2530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6575249932450688e-06,
+      "loss": 2.19,
+      "step": 2535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6568495001350985e-06,
+      "loss": 2.1086,
+      "step": 2540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6561740070251283e-06,
+      "loss": 2.133,
+      "step": 2545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.655498513915158e-06,
+      "loss": 2.0417,
+      "step": 2550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6548230208051877e-06,
+      "loss": 2.1212,
+      "step": 2555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6541475276952174e-06,
+      "loss": 2.0668,
+      "step": 2560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6534720345852472e-06,
+      "loss": 2.1242,
+      "step": 2565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6527965414752769e-06,
+      "loss": 2.0738,
+      "step": 2570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6521210483653066e-06,
+      "loss": 2.1083,
+      "step": 2575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6514455552553363e-06,
+      "loss": 2.0711,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.650770062145366e-06,
+      "loss": 2.0547,
+      "step": 2585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6500945690353958e-06,
+      "loss": 2.0984,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6494190759254253e-06,
+      "loss": 2.0657,
+      "step": 2595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6487435828154552e-06,
+      "loss": 2.1342,
+      "step": 2600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6480680897054848e-06,
+      "loss": 2.1238,
+      "step": 2605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6473925965955147e-06,
+      "loss": 2.0945,
+      "step": 2610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6467171034855442e-06,
+      "loss": 2.1569,
+      "step": 2615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6460416103755741e-06,
+      "loss": 2.055,
+      "step": 2620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6453661172656039e-06,
+      "loss": 2.1235,
+      "step": 2625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6446906241556336e-06,
+      "loss": 2.0776,
+      "step": 2630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6440151310456633e-06,
+      "loss": 2.1014,
+      "step": 2635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.643339637935693e-06,
+      "loss": 2.0771,
+      "step": 2640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6426641448257228e-06,
+      "loss": 2.0169,
+      "step": 2645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6419886517157525e-06,
+      "loss": 2.0718,
+      "step": 2650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6413131586057822e-06,
+      "loss": 2.1339,
+      "step": 2655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6406376654958117e-06,
+      "loss": 2.1413,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6399621723858417e-06,
+      "loss": 2.1967,
+      "step": 2665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6392866792758712e-06,
+      "loss": 2.0654,
+      "step": 2670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6386111861659011e-06,
+      "loss": 2.0067,
+      "step": 2675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6379356930559307e-06,
+      "loss": 1.9827,
+      "step": 2680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6372601999459606e-06,
+      "loss": 2.1348,
+      "step": 2685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.63658470683599e-06,
+      "loss": 2.0717,
+      "step": 2690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.63590921372602e-06,
+      "loss": 2.0616,
+      "step": 2695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6352337206160496e-06,
+      "loss": 2.1373,
+      "step": 2700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6345582275060795e-06,
+      "loss": 2.07,
+      "step": 2705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.633882734396109e-06,
+      "loss": 2.0335,
+      "step": 2710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.633207241286139e-06,
+      "loss": 2.119,
+      "step": 2715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6325317481761685e-06,
+      "loss": 2.0989,
+      "step": 2720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6318562550661982e-06,
+      "loss": 2.1273,
+      "step": 2725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.631180761956228e-06,
+      "loss": 2.0868,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6305052688462576e-06,
+      "loss": 2.0753,
+      "step": 2735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6298297757362874e-06,
+      "loss": 2.0779,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.629154282626317e-06,
+      "loss": 1.9635,
+      "step": 2745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6284787895163468e-06,
+      "loss": 2.0501,
+      "step": 2750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6278032964063765e-06,
+      "loss": 2.0689,
+      "step": 2755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6271278032964063e-06,
+      "loss": 2.0779,
+      "step": 2760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.626452310186436e-06,
+      "loss": 2.0272,
+      "step": 2765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.625776817076466e-06,
+      "loss": 2.1481,
+      "step": 2770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6251013239664955e-06,
+      "loss": 2.0285,
+      "step": 2775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6244258308565254e-06,
+      "loss": 2.013,
+      "step": 2780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.623750337746555e-06,
+      "loss": 2.1225,
+      "step": 2785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6230748446365846e-06,
+      "loss": 2.067,
+      "step": 2790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6223993515266144e-06,
+      "loss": 2.1046,
+      "step": 2795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.621723858416644e-06,
+      "loss": 2.0699,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 2.0739200115203857,
+      "eval_runtime": 165.5539,
+      "eval_samples_per_second": 3.6,
+      "eval_steps_per_second": 0.453,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6210483653066738e-06,
+      "loss": 2.119,
+      "step": 2805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6203728721967035e-06,
+      "loss": 2.0534,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6196973790867333e-06,
+      "loss": 2.1377,
+      "step": 2815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619021885976763e-06,
+      "loss": 2.0943,
+      "step": 2820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6183463928667927e-06,
+      "loss": 2.1415,
+      "step": 2825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6176708997568224e-06,
+      "loss": 2.0923,
+      "step": 2830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6169954066468522e-06,
+      "loss": 2.0528,
+      "step": 2835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6163199135368819e-06,
+      "loss": 2.0935,
+      "step": 2840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6156444204269116e-06,
+      "loss": 2.1159,
+      "step": 2845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6149689273169413e-06,
+      "loss": 2.0728,
+      "step": 2850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6142934342069709e-06,
+      "loss": 2.0722,
+      "step": 2855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6136179410970008e-06,
+      "loss": 2.0461,
+      "step": 2860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6129424479870303e-06,
+      "loss": 2.1287,
+      "step": 2865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6122669548770603e-06,
+      "loss": 2.0648,
+      "step": 2870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6115914617670898e-06,
+      "loss": 2.0533,
+      "step": 2875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6109159686571197e-06,
+      "loss": 2.0866,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6102404755471492e-06,
+      "loss": 2.1042,
+      "step": 2885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6095649824371792e-06,
+      "loss": 2.1186,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6088894893272087e-06,
+      "loss": 2.1474,
+      "step": 2895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6082139962172386e-06,
+      "loss": 2.1159,
+      "step": 2900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6075385031072681e-06,
+      "loss": 2.0443,
+      "step": 2905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606863009997298e-06,
+      "loss": 2.0029,
+      "step": 2910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6061875168873278e-06,
+      "loss": 2.0684,
+      "step": 2915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6055120237773573e-06,
+      "loss": 1.9991,
+      "step": 2920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6048365306673872e-06,
+      "loss": 2.0945,
+      "step": 2925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6041610375574168e-06,
+      "loss": 2.0717,
+      "step": 2930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6034855444474467e-06,
+      "loss": 2.0839,
+      "step": 2935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6028100513374762e-06,
+      "loss": 2.1329,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6021345582275061e-06,
+      "loss": 2.0077,
+      "step": 2945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6014590651175357e-06,
+      "loss": 2.0517,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6007835720075656e-06,
+      "loss": 2.0322,
+      "step": 2955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6001080788975951e-06,
+      "loss": 2.055,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.599432585787625e-06,
+      "loss": 2.1037,
+      "step": 2965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5987570926776546e-06,
+      "loss": 2.059,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5980815995676845e-06,
+      "loss": 2.0551,
+      "step": 2975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.597406106457714e-06,
+      "loss": 2.0848,
+      "step": 2980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5967306133477437e-06,
+      "loss": 2.0801,
+      "step": 2985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5960551202377735e-06,
+      "loss": 2.0898,
+      "step": 2990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5953796271278032e-06,
+      "loss": 2.0975,
+      "step": 2995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.594704134017833e-06,
+      "loss": 2.0278,
+      "step": 3000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5940286409078626e-06,
+      "loss": 2.0342,
+      "step": 3005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5933531477978924e-06,
+      "loss": 2.068,
+      "step": 3010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.592677654687922e-06,
+      "loss": 2.0488,
+      "step": 3015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5920021615779518e-06,
+      "loss": 2.0348,
+      "step": 3020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5913266684679816e-06,
+      "loss": 2.0646,
+      "step": 3025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5906511753580113e-06,
+      "loss": 2.0238,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.589975682248041e-06,
+      "loss": 2.1214,
+      "step": 3035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5893001891380705e-06,
+      "loss": 2.0474,
+      "step": 3040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5886246960281005e-06,
+      "loss": 2.0923,
+      "step": 3045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.58794920291813e-06,
+      "loss": 2.0824,
+      "step": 3050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.58727370980816e-06,
+      "loss": 2.0408,
+      "step": 3055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5865982166981896e-06,
+      "loss": 2.0531,
+      "step": 3060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5859227235882194e-06,
+      "loss": 2.0216,
+      "step": 3065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.585247230478249e-06,
+      "loss": 2.1115,
+      "step": 3070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5845717373682788e-06,
+      "loss": 2.0333,
+      "step": 3075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5838962442583085e-06,
+      "loss": 2.0941,
+      "step": 3080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5832207511483383e-06,
+      "loss": 2.0995,
+      "step": 3085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.582545258038368e-06,
+      "loss": 2.1471,
+      "step": 3090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5818697649283977e-06,
+      "loss": 1.9764,
+      "step": 3095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5811942718184274e-06,
+      "loss": 2.0089,
+      "step": 3100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.580518778708457e-06,
+      "loss": 2.1237,
+      "step": 3105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.579843285598487e-06,
+      "loss": 2.1105,
+      "step": 3110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5791677924885164e-06,
+      "loss": 2.1469,
+      "step": 3115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5784922993785464e-06,
+      "loss": 2.0442,
+      "step": 3120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5778168062685759e-06,
+      "loss": 2.194,
+      "step": 3125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5771413131586058e-06,
+      "loss": 2.0288,
+      "step": 3130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5764658200486353e-06,
+      "loss": 2.0379,
+      "step": 3135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5757903269386653e-06,
+      "loss": 2.07,
+      "step": 3140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5751148338286948e-06,
+      "loss": 2.0844,
+      "step": 3145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5744393407187247e-06,
+      "loss": 2.0518,
+      "step": 3150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5737638476087542e-06,
+      "loss": 2.0537,
+      "step": 3155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5730883544987842e-06,
+      "loss": 2.0722,
+      "step": 3160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5724128613888137e-06,
+      "loss": 2.1186,
+      "step": 3165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5717373682788436e-06,
+      "loss": 2.0531,
+      "step": 3170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5710618751688731e-06,
+      "loss": 2.0955,
+      "step": 3175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5703863820589029e-06,
+      "loss": 2.0703,
+      "step": 3180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5697108889489326e-06,
+      "loss": 2.1432,
+      "step": 3185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5690353958389623e-06,
+      "loss": 2.0921,
+      "step": 3190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.568359902728992e-06,
+      "loss": 2.0203,
+      "step": 3195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5676844096190218e-06,
+      "loss": 2.048,
+      "step": 3200
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 2.060811758041382,
+      "eval_runtime": 165.4321,
+      "eval_samples_per_second": 3.603,
+      "eval_steps_per_second": 0.453,
+      "step": 3200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5670089165090515e-06,
+      "loss": 2.0813,
+      "step": 3205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5663334233990812e-06,
+      "loss": 2.1681,
+      "step": 3210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5656579302891112e-06,
+      "loss": 2.009,
+      "step": 3215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5649824371791407e-06,
+      "loss": 2.1227,
+      "step": 3220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5643069440691706e-06,
+      "loss": 1.9925,
+      "step": 3225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5636314509592001e-06,
+      "loss": 2.0469,
+      "step": 3230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.56295595784923e-06,
+      "loss": 2.0438,
+      "step": 3235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5622804647392596e-06,
+      "loss": 2.0806,
+      "step": 3240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5616049716292893e-06,
+      "loss": 2.1108,
+      "step": 3245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.560929478519319e-06,
+      "loss": 2.0461,
+      "step": 3250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5602539854093488e-06,
+      "loss": 2.13,
+      "step": 3255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5595784922993785e-06,
+      "loss": 2.1125,
+      "step": 3260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5589029991894082e-06,
+      "loss": 2.0172,
+      "step": 3265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.558227506079438e-06,
+      "loss": 2.1299,
+      "step": 3270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5575520129694677e-06,
+      "loss": 2.0406,
+      "step": 3275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5568765198594974e-06,
+      "loss": 2.0501,
+      "step": 3280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5562010267495271e-06,
+      "loss": 2.0377,
+      "step": 3285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5555255336395568e-06,
+      "loss": 2.0649,
+      "step": 3290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5548500405295866e-06,
+      "loss": 2.0212,
+      "step": 3295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.554174547419616e-06,
+      "loss": 2.0358,
+      "step": 3300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.553499054309646e-06,
+      "loss": 2.1017,
+      "step": 3305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5528235611996755e-06,
+      "loss": 1.9907,
+      "step": 3310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5521480680897055e-06,
+      "loss": 2.0097,
+      "step": 3315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.551472574979735e-06,
+      "loss": 2.0266,
+      "step": 3320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.550797081869765e-06,
+      "loss": 2.1016,
+      "step": 3325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5501215887597944e-06,
+      "loss": 2.162,
+      "step": 3330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5494460956498244e-06,
+      "loss": 2.0395,
+      "step": 3335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5487706025398539e-06,
+      "loss": 2.0559,
+      "step": 3340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5480951094298838e-06,
+      "loss": 2.1894,
+      "step": 3345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5474196163199133e-06,
+      "loss": 2.0609,
+      "step": 3350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5467441232099433e-06,
+      "loss": 2.1246,
+      "step": 3355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.546068630099973e-06,
+      "loss": 2.125,
+      "step": 3360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5453931369900025e-06,
+      "loss": 2.1067,
+      "step": 3365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5447176438800325e-06,
+      "loss": 1.9872,
+      "step": 3370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.544042150770062e-06,
+      "loss": 2.1292,
+      "step": 3375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.543366657660092e-06,
+      "loss": 2.1035,
+      "step": 3380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5426911645501214e-06,
+      "loss": 2.1686,
+      "step": 3385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5420156714401514e-06,
+      "loss": 2.0928,
+      "step": 3390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5413401783301809e-06,
+      "loss": 2.0584,
+      "step": 3395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5406646852202108e-06,
+      "loss": 2.0046,
+      "step": 3400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5399891921102403e-06,
+      "loss": 2.0863,
+      "step": 3405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5393136990002703e-06,
+      "loss": 2.0954,
+      "step": 3410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5386382058902998e-06,
+      "loss": 2.0727,
+      "step": 3415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5379627127803297e-06,
+      "loss": 2.1576,
+      "step": 3420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5372872196703592e-06,
+      "loss": 2.046,
+      "step": 3425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5366117265603892e-06,
+      "loss": 2.0489,
+      "step": 3430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5359362334504187e-06,
+      "loss": 2.0607,
+      "step": 3435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5352607403404484e-06,
+      "loss": 1.995,
+      "step": 3440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5345852472304781e-06,
+      "loss": 2.0571,
+      "step": 3445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5339097541205079e-06,
+      "loss": 2.0897,
+      "step": 3450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5332342610105376e-06,
+      "loss": 2.0928,
+      "step": 3455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5325587679005673e-06,
+      "loss": 2.0929,
+      "step": 3460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.531883274790597e-06,
+      "loss": 2.123,
+      "step": 3465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5312077816806268e-06,
+      "loss": 2.1284,
+      "step": 3470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5305322885706565e-06,
+      "loss": 2.0668,
+      "step": 3475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5298567954606862e-06,
+      "loss": 2.038,
+      "step": 3480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.529181302350716e-06,
+      "loss": 2.0609,
+      "step": 3485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5285058092407457e-06,
+      "loss": 2.1102,
+      "step": 3490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5278303161307752e-06,
+      "loss": 2.0901,
+      "step": 3495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5271548230208051e-06,
+      "loss": 2.0164,
+      "step": 3500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5264793299108349e-06,
+      "loss": 2.0328,
+      "step": 3505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5258038368008646e-06,
+      "loss": 2.131,
+      "step": 3510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5251283436908943e-06,
+      "loss": 2.1453,
+      "step": 3515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.524452850580924e-06,
+      "loss": 2.0078,
+      "step": 3520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5237773574709538e-06,
+      "loss": 2.0995,
+      "step": 3525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5231018643609835e-06,
+      "loss": 2.0006,
+      "step": 3530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5224263712510132e-06,
+      "loss": 2.1118,
+      "step": 3535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.521750878141043e-06,
+      "loss": 2.1054,
+      "step": 3540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5210753850310727e-06,
+      "loss": 2.0164,
+      "step": 3545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5203998919211024e-06,
+      "loss": 2.146,
+      "step": 3550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5197243988111321e-06,
+      "loss": 2.0721,
+      "step": 3555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5190489057011616e-06,
+      "loss": 2.0081,
+      "step": 3560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5183734125911916e-06,
+      "loss": 2.1044,
+      "step": 3565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.517697919481221e-06,
+      "loss": 2.097,
+      "step": 3570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.517022426371251e-06,
+      "loss": 2.0469,
+      "step": 3575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5163469332612805e-06,
+      "loss": 2.0239,
+      "step": 3580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5156714401513105e-06,
+      "loss": 2.0602,
+      "step": 3585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.51499594704134e-06,
+      "loss": 2.0721,
+      "step": 3590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.51432045393137e-06,
+      "loss": 2.0337,
+      "step": 3595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5136449608213994e-06,
+      "loss": 1.8914,
+      "step": 3600
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 2.049622058868408,
+      "eval_runtime": 165.1588,
+      "eval_samples_per_second": 3.609,
+      "eval_steps_per_second": 0.454,
+      "step": 3600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5129694677114294e-06,
+      "loss": 1.9725,
+      "step": 3605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5122939746014589e-06,
+      "loss": 1.9695,
+      "step": 3610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5116184814914888e-06,
+      "loss": 2.0596,
+      "step": 3615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5109429883815183e-06,
+      "loss": 1.993,
+      "step": 3620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.510267495271548e-06,
+      "loss": 2.0617,
+      "step": 3625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5095920021615778e-06,
+      "loss": 2.0145,
+      "step": 3630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5089165090516075e-06,
+      "loss": 2.1149,
+      "step": 3635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5082410159416373e-06,
+      "loss": 2.0825,
+      "step": 3640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.507565522831667e-06,
+      "loss": 2.0762,
+      "step": 3645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.506890029721697e-06,
+      "loss": 2.1251,
+      "step": 3650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5062145366117264e-06,
+      "loss": 2.0812,
+      "step": 3655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5055390435017564e-06,
+      "loss": 2.0488,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5048635503917859e-06,
+      "loss": 2.0906,
+      "step": 3665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5041880572818158e-06,
+      "loss": 2.0587,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5035125641718453e-06,
+      "loss": 2.1736,
+      "step": 3675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5028370710618753e-06,
+      "loss": 2.0042,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5021615779519048e-06,
+      "loss": 2.0003,
+      "step": 3685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5014860848419347e-06,
+      "loss": 2.0667,
+      "step": 3690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5008105917319642e-06,
+      "loss": 2.111,
+      "step": 3695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.500135098621994e-06,
+      "loss": 2.0688,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4994596055120237e-06,
+      "loss": 1.9983,
+      "step": 3705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4987841124020534e-06,
+      "loss": 2.0898,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4981086192920831e-06,
+      "loss": 2.0828,
+      "step": 3715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4974331261821129e-06,
+      "loss": 2.0781,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4967576330721426e-06,
+      "loss": 2.0078,
+      "step": 3725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4960821399621723e-06,
+      "loss": 2.0798,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.495406646852202e-06,
+      "loss": 2.0405,
+      "step": 3735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4947311537422318e-06,
+      "loss": 2.0513,
+      "step": 3740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4940556606322615e-06,
+      "loss": 2.0381,
+      "step": 3745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4933801675222912e-06,
+      "loss": 2.0066,
+      "step": 3750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4927046744123207e-06,
+      "loss": 2.0412,
+      "step": 3755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4920291813023507e-06,
+      "loss": 2.0252,
+      "step": 3760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4913536881923802e-06,
+      "loss": 2.1475,
+      "step": 3765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4906781950824101e-06,
+      "loss": 2.0673,
+      "step": 3770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4900027019724396e-06,
+      "loss": 2.0293,
+      "step": 3775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4893272088624696e-06,
+      "loss": 2.0817,
+      "step": 3780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.488651715752499e-06,
+      "loss": 2.08,
+      "step": 3785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.487976222642529e-06,
+      "loss": 2.044,
+      "step": 3790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4873007295325588e-06,
+      "loss": 2.0545,
+      "step": 3795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4866252364225885e-06,
+      "loss": 2.0712,
+      "step": 3800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4859497433126182e-06,
+      "loss": 2.0178,
+      "step": 3805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.485274250202648e-06,
+      "loss": 2.0222,
+      "step": 3810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4845987570926777e-06,
+      "loss": 2.0417,
+      "step": 3815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4839232639827072e-06,
+      "loss": 2.0045,
+      "step": 3820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4832477708727371e-06,
+      "loss": 2.048,
+      "step": 3825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4825722777627666e-06,
+      "loss": 2.1043,
+      "step": 3830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4818967846527966e-06,
+      "loss": 2.0242,
+      "step": 3835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.481221291542826e-06,
+      "loss": 2.1304,
+      "step": 3840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.480545798432856e-06,
+      "loss": 2.0436,
+      "step": 3845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4798703053228855e-06,
+      "loss": 2.0306,
+      "step": 3850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4791948122129155e-06,
+      "loss": 1.9944,
+      "step": 3855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.478519319102945e-06,
+      "loss": 2.0746,
+      "step": 3860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.477843825992975e-06,
+      "loss": 2.0468,
+      "step": 3865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4771683328830044e-06,
+      "loss": 2.0807,
+      "step": 3870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4764928397730344e-06,
+      "loss": 2.1531,
+      "step": 3875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.475817346663064e-06,
+      "loss": 2.0949,
+      "step": 3880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4751418535530936e-06,
+      "loss": 1.9704,
+      "step": 3885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4744663604431234e-06,
+      "loss": 1.9872,
+      "step": 3890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.473790867333153e-06,
+      "loss": 1.997,
+      "step": 3895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4731153742231828e-06,
+      "loss": 2.0673,
+      "step": 3900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4724398811132125e-06,
+      "loss": 2.0324,
+      "step": 3905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4717643880032423e-06,
+      "loss": 2.0306,
+      "step": 3910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.471088894893272e-06,
+      "loss": 2.0718,
+      "step": 3915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4704134017833017e-06,
+      "loss": 2.1204,
+      "step": 3920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4697379086733314e-06,
+      "loss": 1.9834,
+      "step": 3925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4690624155633612e-06,
+      "loss": 2.0658,
+      "step": 3930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4683869224533909e-06,
+      "loss": 2.0699,
+      "step": 3935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4677114293434208e-06,
+      "loss": 2.0704,
+      "step": 3940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4670359362334503e-06,
+      "loss": 2.0563,
+      "step": 3945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.46636044312348e-06,
+      "loss": 2.0176,
+      "step": 3950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4656849500135098e-06,
+      "loss": 2.067,
+      "step": 3955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4650094569035395e-06,
+      "loss": 2.1244,
+      "step": 3960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4643339637935692e-06,
+      "loss": 2.0181,
+      "step": 3965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.463658470683599e-06,
+      "loss": 2.0596,
+      "step": 3970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4629829775736287e-06,
+      "loss": 2.0484,
+      "step": 3975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4623074844636584e-06,
+      "loss": 2.0901,
+      "step": 3980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4616319913536882e-06,
+      "loss": 1.9073,
+      "step": 3985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4609564982437179e-06,
+      "loss": 2.0358,
+      "step": 3990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4602810051337476e-06,
+      "loss": 2.0291,
+      "step": 3995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4596055120237773e-06,
+      "loss": 2.0276,
+      "step": 4000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 2.039377212524414,
+      "eval_runtime": 165.26,
+      "eval_samples_per_second": 3.606,
+      "eval_steps_per_second": 0.454,
+      "step": 4000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.458930018913807e-06,
+      "loss": 2.1331,
+      "step": 4005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4582545258038368e-06,
+      "loss": 2.0162,
+      "step": 4010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4575790326938663e-06,
+      "loss": 1.9811,
+      "step": 4015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4569035395838962e-06,
+      "loss": 1.9574,
+      "step": 4020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4562280464739258e-06,
+      "loss": 1.9547,
+      "step": 4025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4555525533639557e-06,
+      "loss": 2.0814,
+      "step": 4030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4548770602539852e-06,
+      "loss": 2.0517,
+      "step": 4035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4542015671440151e-06,
+      "loss": 2.0873,
+      "step": 4040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4535260740340447e-06,
+      "loss": 2.0353,
+      "step": 4045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4528505809240746e-06,
+      "loss": 2.0967,
+      "step": 4050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4521750878141041e-06,
+      "loss": 1.9676,
+      "step": 4055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.451499594704134e-06,
+      "loss": 2.0532,
+      "step": 4060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4508241015941636e-06,
+      "loss": 2.0872,
+      "step": 4065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4501486084841935e-06,
+      "loss": 2.092,
+      "step": 4070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.449473115374223e-06,
+      "loss": 1.9767,
+      "step": 4075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4487976222642527e-06,
+      "loss": 2.0484,
+      "step": 4080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4481221291542827e-06,
+      "loss": 2.0448,
+      "step": 4085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4474466360443122e-06,
+      "loss": 2.0531,
+      "step": 4090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4467711429343421e-06,
+      "loss": 2.014,
+      "step": 4095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4460956498243716e-06,
+      "loss": 2.0123,
+      "step": 4100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4454201567144016e-06,
+      "loss": 2.0659,
+      "step": 4105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.444744663604431e-06,
+      "loss": 2.0284,
+      "step": 4110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.444069170494461e-06,
+      "loss": 2.1546,
+      "step": 4115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4433936773844905e-06,
+      "loss": 2.0555,
+      "step": 4120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4427181842745205e-06,
+      "loss": 2.0946,
+      "step": 4125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.44204269116455e-06,
+      "loss": 2.037,
+      "step": 4130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.44136719805458e-06,
+      "loss": 2.0163,
+      "step": 4135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4406917049446095e-06,
+      "loss": 2.0224,
+      "step": 4140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4400162118346392e-06,
+      "loss": 2.0542,
+      "step": 4145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.439340718724669e-06,
+      "loss": 2.0628,
+      "step": 4150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4386652256146986e-06,
+      "loss": 2.0527,
+      "step": 4155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4379897325047284e-06,
+      "loss": 1.9956,
+      "step": 4160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.437314239394758e-06,
+      "loss": 1.9983,
+      "step": 4165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4366387462847878e-06,
+      "loss": 2.0351,
+      "step": 4170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4359632531748175e-06,
+      "loss": 1.9065,
+      "step": 4175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4352877600648473e-06,
+      "loss": 2.0242,
+      "step": 4180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.434612266954877e-06,
+      "loss": 2.0076,
+      "step": 4185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4339367738449067e-06,
+      "loss": 2.0707,
+      "step": 4190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4332612807349364e-06,
+      "loss": 2.0493,
+      "step": 4195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4325857876249662e-06,
+      "loss": 2.0358,
+      "step": 4200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.431910294514996e-06,
+      "loss": 1.9821,
+      "step": 4205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4312348014050254e-06,
+      "loss": 2.0288,
+      "step": 4210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4305593082950553e-06,
+      "loss": 2.0614,
+      "step": 4215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4298838151850849e-06,
+      "loss": 2.1316,
+      "step": 4220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4292083220751148e-06,
+      "loss": 1.9865,
+      "step": 4225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4285328289651445e-06,
+      "loss": 2.1231,
+      "step": 4230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4278573358551743e-06,
+      "loss": 2.0609,
+      "step": 4235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.427181842745204e-06,
+      "loss": 1.995,
+      "step": 4240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4265063496352337e-06,
+      "loss": 2.0347,
+      "step": 4245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4258308565252634e-06,
+      "loss": 2.0714,
+      "step": 4250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4251553634152932e-06,
+      "loss": 1.9985,
+      "step": 4255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4244798703053229e-06,
+      "loss": 2.0541,
+      "step": 4260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4238043771953526e-06,
+      "loss": 2.0663,
+      "step": 4265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4231288840853823e-06,
+      "loss": 2.0048,
+      "step": 4270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4224533909754119e-06,
+      "loss": 2.0742,
+      "step": 4275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4217778978654418e-06,
+      "loss": 2.0085,
+      "step": 4280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4211024047554713e-06,
+      "loss": 1.9951,
+      "step": 4285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4204269116455012e-06,
+      "loss": 2.0308,
+      "step": 4290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4197514185355308e-06,
+      "loss": 2.1408,
+      "step": 4295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4190759254255607e-06,
+      "loss": 1.9765,
+      "step": 4300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4184004323155902e-06,
+      "loss": 1.9744,
+      "step": 4305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4177249392056201e-06,
+      "loss": 2.0597,
+      "step": 4310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4170494460956497e-06,
+      "loss": 2.0787,
+      "step": 4315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4163739529856796e-06,
+      "loss": 1.9937,
+      "step": 4320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4156984598757091e-06,
+      "loss": 2.0415,
+      "step": 4325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.415022966765739e-06,
+      "loss": 1.9378,
+      "step": 4330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4143474736557686e-06,
+      "loss": 2.0868,
+      "step": 4335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4136719805457983e-06,
+      "loss": 2.0547,
+      "step": 4340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.412996487435828e-06,
+      "loss": 2.0683,
+      "step": 4345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4123209943258577e-06,
+      "loss": 2.1531,
+      "step": 4350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4116455012158875e-06,
+      "loss": 2.0331,
+      "step": 4355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4109700081059172e-06,
+      "loss": 2.0184,
+      "step": 4360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.410294514995947e-06,
+      "loss": 2.0473,
+      "step": 4365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4096190218859767e-06,
+      "loss": 1.9568,
+      "step": 4370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4089435287760066e-06,
+      "loss": 2.031,
+      "step": 4375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.408268035666036e-06,
+      "loss": 1.9944,
+      "step": 4380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.407592542556066e-06,
+      "loss": 2.1136,
+      "step": 4385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4069170494460956e-06,
+      "loss": 2.0197,
+      "step": 4390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4062415563361255e-06,
+      "loss": 2.1447,
+      "step": 4395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.405566063226155e-06,
+      "loss": 1.9879,
+      "step": 4400
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 2.0300517082214355,
+      "eval_runtime": 165.4272,
+      "eval_samples_per_second": 3.603,
+      "eval_steps_per_second": 0.453,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4048905701161847e-06,
+      "loss": 2.1061,
+      "step": 4405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4042150770062145e-06,
+      "loss": 2.0397,
+      "step": 4410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4035395838962442e-06,
+      "loss": 1.943,
+      "step": 4415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.402864090786274e-06,
+      "loss": 2.0362,
+      "step": 4420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4021885976763036e-06,
+      "loss": 2.0861,
+      "step": 4425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4015131045663334e-06,
+      "loss": 2.0146,
+      "step": 4430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.400837611456363e-06,
+      "loss": 2.0927,
+      "step": 4435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4001621183463928e-06,
+      "loss": 2.0839,
+      "step": 4440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3994866252364225e-06,
+      "loss": 2.0464,
+      "step": 4445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3988111321264523e-06,
+      "loss": 2.0034,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.398135639016482e-06,
+      "loss": 2.0405,
+      "step": 4455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3974601459065117e-06,
+      "loss": 1.9895,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3967846527965415e-06,
+      "loss": 2.058,
+      "step": 4465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.396109159686571e-06,
+      "loss": 2.0722,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.395433666576601e-06,
+      "loss": 1.9765,
+      "step": 4475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3947581734666304e-06,
+      "loss": 1.9971,
+      "step": 4480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3940826803566604e-06,
+      "loss": 2.0599,
+      "step": 4485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3934071872466899e-06,
+      "loss": 2.0489,
+      "step": 4490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3927316941367198e-06,
+      "loss": 1.9994,
+      "step": 4495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3920562010267493e-06,
+      "loss": 2.0765,
+      "step": 4500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3913807079167793e-06,
+      "loss": 2.1492,
+      "step": 4505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3907052148068088e-06,
+      "loss": 2.0749,
+      "step": 4510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3900297216968387e-06,
+      "loss": 1.9798,
+      "step": 4515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3893542285868684e-06,
+      "loss": 2.0888,
+      "step": 4520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3886787354768982e-06,
+      "loss": 1.9906,
+      "step": 4525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3880032423669279e-06,
+      "loss": 2.0952,
+      "step": 4530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3873277492569574e-06,
+      "loss": 2.0239,
+      "step": 4535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3866522561469873e-06,
+      "loss": 2.0986,
+      "step": 4540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3859767630370169e-06,
+      "loss": 2.1158,
+      "step": 4545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3853012699270468e-06,
+      "loss": 1.9589,
+      "step": 4550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3846257768170763e-06,
+      "loss": 2.0019,
+      "step": 4555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3839502837071062e-06,
+      "loss": 2.0844,
+      "step": 4560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3832747905971358e-06,
+      "loss": 2.0934,
+      "step": 4565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3825992974871657e-06,
+      "loss": 2.007,
+      "step": 4570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3819238043771952e-06,
+      "loss": 1.9538,
+      "step": 4575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3812483112672252e-06,
+      "loss": 2.0167,
+      "step": 4580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3805728181572547e-06,
+      "loss": 2.0112,
+      "step": 4585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3798973250472846e-06,
+      "loss": 2.0415,
+      "step": 4590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3792218319373141e-06,
+      "loss": 2.0616,
+      "step": 4595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3785463388273438e-06,
+      "loss": 2.0704,
+      "step": 4600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3778708457173736e-06,
+      "loss": 2.1166,
+      "step": 4605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3771953526074033e-06,
+      "loss": 2.0665,
+      "step": 4610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.376519859497433e-06,
+      "loss": 2.0999,
+      "step": 4615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3758443663874628e-06,
+      "loss": 2.0465,
+      "step": 4620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3751688732774925e-06,
+      "loss": 2.0234,
+      "step": 4625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3744933801675222e-06,
+      "loss": 2.0602,
+      "step": 4630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.373817887057552e-06,
+      "loss": 2.0471,
+      "step": 4635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3731423939475817e-06,
+      "loss": 2.0579,
+      "step": 4640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3724669008376114e-06,
+      "loss": 2.0118,
+      "step": 4645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3717914077276411e-06,
+      "loss": 1.9907,
+      "step": 4650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3711159146176708e-06,
+      "loss": 1.9865,
+      "step": 4655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3704404215077006e-06,
+      "loss": 2.0415,
+      "step": 4660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3697649283977303e-06,
+      "loss": 2.0445,
+      "step": 4665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.36908943528776e-06,
+      "loss": 1.9549,
+      "step": 4670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3684139421777897e-06,
+      "loss": 1.9776,
+      "step": 4675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3677384490678195e-06,
+      "loss": 2.0224,
+      "step": 4680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3670629559578492e-06,
+      "loss": 1.9824,
+      "step": 4685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.366387462847879e-06,
+      "loss": 1.9642,
+      "step": 4690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3657119697379086e-06,
+      "loss": 2.0882,
+      "step": 4695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3650364766279384e-06,
+      "loss": 2.098,
+      "step": 4700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.364360983517968e-06,
+      "loss": 2.0735,
+      "step": 4705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3636854904079978e-06,
+      "loss": 2.011,
+      "step": 4710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3630099972980276e-06,
+      "loss": 1.9934,
+      "step": 4715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3623345041880573e-06,
+      "loss": 2.0129,
+      "step": 4720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.361659011078087e-06,
+      "loss": 2.0225,
+      "step": 4725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3609835179681165e-06,
+      "loss": 2.1099,
+      "step": 4730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3603080248581465e-06,
+      "loss": 1.9598,
+      "step": 4735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.359632531748176e-06,
+      "loss": 1.9245,
+      "step": 4740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.358957038638206e-06,
+      "loss": 2.0642,
+      "step": 4745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3582815455282354e-06,
+      "loss": 2.1169,
+      "step": 4750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3576060524182654e-06,
+      "loss": 2.0505,
+      "step": 4755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3569305593082949e-06,
+      "loss": 2.0364,
+      "step": 4760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3562550661983248e-06,
+      "loss": 2.1029,
+      "step": 4765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3555795730883543e-06,
+      "loss": 2.136,
+      "step": 4770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3549040799783843e-06,
+      "loss": 1.9615,
+      "step": 4775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3542285868684138e-06,
+      "loss": 1.9855,
+      "step": 4780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3535530937584437e-06,
+      "loss": 1.9529,
+      "step": 4785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3528776006484732e-06,
+      "loss": 2.0281,
+      "step": 4790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.352202107538503e-06,
+      "loss": 1.9702,
+      "step": 4795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3515266144285327e-06,
+      "loss": 2.0656,
+      "step": 4800
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 2.021768808364868,
+      "eval_runtime": 165.3452,
+      "eval_samples_per_second": 3.605,
+      "eval_steps_per_second": 0.454,
+      "step": 4800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3508511213185624e-06,
+      "loss": 1.9761,
+      "step": 4805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3501756282085924e-06,
+      "loss": 2.043,
+      "step": 4810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3495001350986219e-06,
+      "loss": 2.0883,
+      "step": 4815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3488246419886518e-06,
+      "loss": 2.0304,
+      "step": 4820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3481491488786813e-06,
+      "loss": 2.0323,
+      "step": 4825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3474736557687113e-06,
+      "loss": 2.032,
+      "step": 4830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3467981626587408e-06,
+      "loss": 2.0289,
+      "step": 4835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3461226695487707e-06,
+      "loss": 1.9711,
+      "step": 4840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3454471764388002e-06,
+      "loss": 2.0748,
+      "step": 4845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3447716833288302e-06,
+      "loss": 2.0305,
+      "step": 4850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3440961902188597e-06,
+      "loss": 2.1066,
+      "step": 4855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3434206971088894e-06,
+      "loss": 2.009,
+      "step": 4860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3427452039989191e-06,
+      "loss": 1.9678,
+      "step": 4865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3420697108889489e-06,
+      "loss": 2.1183,
+      "step": 4870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3413942177789786e-06,
+      "loss": 2.0144,
+      "step": 4875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3407187246690083e-06,
+      "loss": 2.1348,
+      "step": 4880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.340043231559038e-06,
+      "loss": 1.955,
+      "step": 4885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3393677384490678e-06,
+      "loss": 1.9506,
+      "step": 4890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3386922453390975e-06,
+      "loss": 2.0105,
+      "step": 4895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3380167522291272e-06,
+      "loss": 1.9965,
+      "step": 4900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.337341259119157e-06,
+      "loss": 1.9777,
+      "step": 4905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3366657660091867e-06,
+      "loss": 2.0761,
+      "step": 4910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3359902728992164e-06,
+      "loss": 2.0911,
+      "step": 4915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3353147797892461e-06,
+      "loss": 1.9092,
+      "step": 4920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3346392866792756e-06,
+      "loss": 2.006,
+      "step": 4925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3339637935693056e-06,
+      "loss": 1.9245,
+      "step": 4930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.333288300459335e-06,
+      "loss": 2.0058,
+      "step": 4935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.332612807349365e-06,
+      "loss": 2.0073,
+      "step": 4940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3319373142393945e-06,
+      "loss": 1.9605,
+      "step": 4945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3312618211294245e-06,
+      "loss": 1.9436,
+      "step": 4950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3305863280194542e-06,
+      "loss": 2.0926,
+      "step": 4955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.329910834909484e-06,
+      "loss": 2.0783,
+      "step": 4960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3292353417995137e-06,
+      "loss": 1.9943,
+      "step": 4965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3285598486895434e-06,
+      "loss": 2.0203,
+      "step": 4970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3278843555795731e-06,
+      "loss": 2.0214,
+      "step": 4975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3272088624696028e-06,
+      "loss": 2.1185,
+      "step": 4980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3265333693596326e-06,
+      "loss": 2.0785,
+      "step": 4985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.325857876249662e-06,
+      "loss": 1.9431,
+      "step": 4990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.325182383139692e-06,
+      "loss": 2.0834,
+      "step": 4995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3245068900297215e-06,
+      "loss": 1.8992,
+      "step": 5000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3238313969197515e-06,
+      "loss": 2.0508,
+      "step": 5005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.323155903809781e-06,
+      "loss": 2.0358,
+      "step": 5010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.322480410699811e-06,
+      "loss": 2.0482,
+      "step": 5015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3218049175898404e-06,
+      "loss": 1.9552,
+      "step": 5020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3211294244798704e-06,
+      "loss": 1.9477,
+      "step": 5025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3204539313698999e-06,
+      "loss": 2.0353,
+      "step": 5030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3197784382599298e-06,
+      "loss": 1.9951,
+      "step": 5035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3191029451499593e-06,
+      "loss": 2.0656,
+      "step": 5040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3184274520399893e-06,
+      "loss": 2.0767,
+      "step": 5045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3177519589300188e-06,
+      "loss": 2.0943,
+      "step": 5050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3170764658200485e-06,
+      "loss": 2.0578,
+      "step": 5055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3164009727100782e-06,
+      "loss": 2.0385,
+      "step": 5060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.315725479600108e-06,
+      "loss": 1.9904,
+      "step": 5065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3150499864901377e-06,
+      "loss": 1.9545,
+      "step": 5070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3143744933801674e-06,
+      "loss": 2.0289,
+      "step": 5075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3136990002701971e-06,
+      "loss": 1.9972,
+      "step": 5080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3130235071602269e-06,
+      "loss": 1.9992,
+      "step": 5085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3123480140502566e-06,
+      "loss": 2.0014,
+      "step": 5090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3116725209402863e-06,
+      "loss": 2.0328,
+      "step": 5095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3109970278303163e-06,
+      "loss": 2.0877,
+      "step": 5100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3103215347203458e-06,
+      "loss": 2.0689,
+      "step": 5105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3096460416103757e-06,
+      "loss": 1.9704,
+      "step": 5110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3089705485004052e-06,
+      "loss": 1.9833,
+      "step": 5115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.308295055390435e-06,
+      "loss": 2.0617,
+      "step": 5120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3076195622804647e-06,
+      "loss": 1.995,
+      "step": 5125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3069440691704944e-06,
+      "loss": 1.9955,
+      "step": 5130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3062685760605241e-06,
+      "loss": 2.0641,
+      "step": 5135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3055930829505539e-06,
+      "loss": 2.0867,
+      "step": 5140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3049175898405836e-06,
+      "loss": 2.017,
+      "step": 5145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3042420967306133e-06,
+      "loss": 2.114,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.303566603620643e-06,
+      "loss": 2.0112,
+      "step": 5155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3028911105106728e-06,
+      "loss": 2.037,
+      "step": 5160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3022156174007025e-06,
+      "loss": 2.0592,
+      "step": 5165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3015401242907322e-06,
+      "loss": 2.0331,
+      "step": 5170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3008646311807617e-06,
+      "loss": 2.0045,
+      "step": 5175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3001891380707917e-06,
+      "loss": 2.0046,
+      "step": 5180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2995136449608212e-06,
+      "loss": 2.0658,
+      "step": 5185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2988381518508511e-06,
+      "loss": 2.038,
+      "step": 5190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2981626587408806e-06,
+      "loss": 1.9584,
+      "step": 5195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2974871656309106e-06,
+      "loss": 2.1201,
+      "step": 5200
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 2.0141761302948,
+      "eval_runtime": 165.1921,
+      "eval_samples_per_second": 3.608,
+      "eval_steps_per_second": 0.454,
+      "step": 5200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.29681167252094e-06,
+      "loss": 1.9123,
+      "step": 5205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.29613617941097e-06,
+      "loss": 1.9701,
+      "step": 5210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2954606863009995e-06,
+      "loss": 2.0273,
+      "step": 5215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2947851931910295e-06,
+      "loss": 2.0312,
+      "step": 5220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.294109700081059e-06,
+      "loss": 2.0207,
+      "step": 5225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.293434206971089e-06,
+      "loss": 2.0797,
+      "step": 5230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2927587138611185e-06,
+      "loss": 1.9935,
+      "step": 5235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2920832207511484e-06,
+      "loss": 2.0066,
+      "step": 5240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.291407727641178e-06,
+      "loss": 2.0568,
+      "step": 5245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2907322345312076e-06,
+      "loss": 2.1398,
+      "step": 5250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2900567414212376e-06,
+      "loss": 1.9884,
+      "step": 5255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.289381248311267e-06,
+      "loss": 2.0353,
+      "step": 5260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.288705755201297e-06,
+      "loss": 1.9529,
+      "step": 5265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2880302620913265e-06,
+      "loss": 2.0423,
+      "step": 5270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2873547689813565e-06,
+      "loss": 2.0891,
+      "step": 5275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.286679275871386e-06,
+      "loss": 2.0047,
+      "step": 5280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.286003782761416e-06,
+      "loss": 2.0324,
+      "step": 5285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2853282896514454e-06,
+      "loss": 2.0379,
+      "step": 5290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2846527965414754e-06,
+      "loss": 2.0345,
+      "step": 5295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2839773034315049e-06,
+      "loss": 1.9509,
+      "step": 5300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2833018103215348e-06,
+      "loss": 2.076,
+      "step": 5305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2826263172115643e-06,
+      "loss": 2.0107,
+      "step": 5310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.281950824101594e-06,
+      "loss": 1.9542,
+      "step": 5315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2812753309916238e-06,
+      "loss": 1.9767,
+      "step": 5320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2805998378816535e-06,
+      "loss": 1.9998,
+      "step": 5325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2799243447716832e-06,
+      "loss": 1.9633,
+      "step": 5330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.279248851661713e-06,
+      "loss": 2.006,
+      "step": 5335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2785733585517427e-06,
+      "loss": 2.0871,
+      "step": 5340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2778978654417724e-06,
+      "loss": 2.0124,
+      "step": 5345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2772223723318022e-06,
+      "loss": 1.9771,
+      "step": 5350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2765468792218319e-06,
+      "loss": 2.0389,
+      "step": 5355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2758713861118616e-06,
+      "loss": 2.1118,
+      "step": 5360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2751958930018913e-06,
+      "loss": 2.0358,
+      "step": 5365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2745203998919208e-06,
+      "loss": 2.0398,
+      "step": 5370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2738449067819508e-06,
+      "loss": 1.9582,
+      "step": 5375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2731694136719803e-06,
+      "loss": 2.0692,
+      "step": 5380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2724939205620102e-06,
+      "loss": 2.0527,
+      "step": 5385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2718184274520398e-06,
+      "loss": 2.077,
+      "step": 5390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2711429343420697e-06,
+      "loss": 2.0804,
+      "step": 5395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2704674412320994e-06,
+      "loss": 1.9849,
+      "step": 5400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2697919481221291e-06,
+      "loss": 2.0286,
+      "step": 5405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2691164550121589e-06,
+      "loss": 2.0492,
+      "step": 5410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2684409619021886e-06,
+      "loss": 1.9443,
+      "step": 5415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2677654687922183e-06,
+      "loss": 1.9515,
+      "step": 5420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.267089975682248e-06,
+      "loss": 2.044,
+      "step": 5425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2664144825722778e-06,
+      "loss": 2.0567,
+      "step": 5430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2657389894623073e-06,
+      "loss": 1.9627,
+      "step": 5435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2650634963523372e-06,
+      "loss": 2.0867,
+      "step": 5440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2643880032423667e-06,
+      "loss": 1.9958,
+      "step": 5445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2637125101323967e-06,
+      "loss": 1.9795,
+      "step": 5450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2630370170224262e-06,
+      "loss": 2.1329,
+      "step": 5455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2623615239124561e-06,
+      "loss": 1.9491,
+      "step": 5460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2616860308024856e-06,
+      "loss": 1.9966,
+      "step": 5465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2610105376925156e-06,
+      "loss": 2.0714,
+      "step": 5470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.260335044582545e-06,
+      "loss": 2.047,
+      "step": 5475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.259659551472575e-06,
+      "loss": 1.9376,
+      "step": 5480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2589840583626046e-06,
+      "loss": 2.012,
+      "step": 5485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2583085652526345e-06,
+      "loss": 2.0617,
+      "step": 5490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.257633072142664e-06,
+      "loss": 1.9983,
+      "step": 5495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.256957579032694e-06,
+      "loss": 2.0385,
+      "step": 5500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2562820859227235e-06,
+      "loss": 1.9452,
+      "step": 5505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2556065928127532e-06,
+      "loss": 2.0028,
+      "step": 5510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.254931099702783e-06,
+      "loss": 2.0326,
+      "step": 5515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2542556065928126e-06,
+      "loss": 1.9992,
+      "step": 5520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2535801134828424e-06,
+      "loss": 1.9506,
+      "step": 5525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.252904620372872e-06,
+      "loss": 2.039,
+      "step": 5530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2522291272629018e-06,
+      "loss": 2.072,
+      "step": 5535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2515536341529315e-06,
+      "loss": 2.0191,
+      "step": 5540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2508781410429615e-06,
+      "loss": 2.0545,
+      "step": 5545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.250202647932991e-06,
+      "loss": 1.9955,
+      "step": 5550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.249527154823021e-06,
+      "loss": 1.9908,
+      "step": 5555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2488516617130504e-06,
+      "loss": 1.9835,
+      "step": 5560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2481761686030804e-06,
+      "loss": 2.054,
+      "step": 5565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.24750067549311e-06,
+      "loss": 2.0544,
+      "step": 5570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2468251823831396e-06,
+      "loss": 2.0558,
+      "step": 5575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2461496892731694e-06,
+      "loss": 1.9398,
+      "step": 5580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.245474196163199e-06,
+      "loss": 2.1002,
+      "step": 5585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2447987030532288e-06,
+      "loss": 2.1542,
+      "step": 5590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2441232099432585e-06,
+      "loss": 2.0554,
+      "step": 5595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2434477168332883e-06,
+      "loss": 2.0143,
+      "step": 5600
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 2.007206916809082,
+      "eval_runtime": 165.3065,
+      "eval_samples_per_second": 3.605,
+      "eval_steps_per_second": 0.454,
+      "step": 5600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.242772223723318e-06,
+      "loss": 2.093,
+      "step": 5605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2420967306133477e-06,
+      "loss": 1.9739,
+      "step": 5610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2414212375033774e-06,
+      "loss": 2.1094,
+      "step": 5615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2407457443934072e-06,
+      "loss": 2.1694,
+      "step": 5620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2400702512834369e-06,
+      "loss": 2.0056,
+      "step": 5625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2393947581734664e-06,
+      "loss": 1.9875,
+      "step": 5630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2387192650634963e-06,
+      "loss": 2.0466,
+      "step": 5635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2380437719535259e-06,
+      "loss": 1.9781,
+      "step": 5640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2373682788435558e-06,
+      "loss": 2.044,
+      "step": 5645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2366927857335853e-06,
+      "loss": 1.9391,
+      "step": 5650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2360172926236152e-06,
+      "loss": 1.9503,
+      "step": 5655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2353417995136448e-06,
+      "loss": 2.1022,
+      "step": 5660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2346663064036747e-06,
+      "loss": 2.1214,
+      "step": 5665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2339908132937042e-06,
+      "loss": 2.0368,
+      "step": 5670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2333153201837342e-06,
+      "loss": 2.0938,
+      "step": 5675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2326398270737637e-06,
+      "loss": 1.9393,
+      "step": 5680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2319643339637936e-06,
+      "loss": 1.9626,
+      "step": 5685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2312888408538233e-06,
+      "loss": 2.0078,
+      "step": 5690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2306133477438528e-06,
+      "loss": 1.9386,
+      "step": 5695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2299378546338828e-06,
+      "loss": 2.0948,
+      "step": 5700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2292623615239123e-06,
+      "loss": 2.013,
+      "step": 5705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2285868684139422e-06,
+      "loss": 2.0048,
+      "step": 5710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2279113753039717e-06,
+      "loss": 2.0154,
+      "step": 5715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2272358821940017e-06,
+      "loss": 2.0012,
+      "step": 5720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2265603890840312e-06,
+      "loss": 1.9525,
+      "step": 5725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2258848959740611e-06,
+      "loss": 2.0631,
+      "step": 5730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2252094028640907e-06,
+      "loss": 2.0146,
+      "step": 5735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2245339097541206e-06,
+      "loss": 1.9761,
+      "step": 5740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2238584166441501e-06,
+      "loss": 2.0205,
+      "step": 5745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.22318292353418e-06,
+      "loss": 2.0304,
+      "step": 5750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2225074304242096e-06,
+      "loss": 2.0341,
+      "step": 5755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2218319373142395e-06,
+      "loss": 2.0402,
+      "step": 5760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.221156444204269e-06,
+      "loss": 2.0297,
+      "step": 5765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2204809510942987e-06,
+      "loss": 2.0116,
+      "step": 5770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2198054579843285e-06,
+      "loss": 2.0157,
+      "step": 5775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2191299648743582e-06,
+      "loss": 2.0291,
+      "step": 5780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.218454471764388e-06,
+      "loss": 2.0338,
+      "step": 5785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2177789786544176e-06,
+      "loss": 2.0364,
+      "step": 5790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2171034855444474e-06,
+      "loss": 1.9964,
+      "step": 5795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.216427992434477e-06,
+      "loss": 2.135,
+      "step": 5800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2157524993245068e-06,
+      "loss": 1.8928,
+      "step": 5805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2150770062145365e-06,
+      "loss": 1.9855,
+      "step": 5810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2144015131045663e-06,
+      "loss": 1.9905,
+      "step": 5815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.213726019994596e-06,
+      "loss": 1.9884,
+      "step": 5820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2130505268846255e-06,
+      "loss": 2.0023,
+      "step": 5825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2123750337746555e-06,
+      "loss": 2.0608,
+      "step": 5830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2116995406646852e-06,
+      "loss": 2.0036,
+      "step": 5835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.211024047554715e-06,
+      "loss": 2.0105,
+      "step": 5840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2103485544447446e-06,
+      "loss": 1.9843,
+      "step": 5845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2096730613347744e-06,
+      "loss": 1.9958,
+      "step": 5850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.208997568224804e-06,
+      "loss": 2.0423,
+      "step": 5855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2083220751148338e-06,
+      "loss": 1.9651,
+      "step": 5860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2076465820048635e-06,
+      "loss": 2.0644,
+      "step": 5865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2069710888948933e-06,
+      "loss": 1.9557,
+      "step": 5870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.206295595784923e-06,
+      "loss": 2.0243,
+      "step": 5875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2056201026749527e-06,
+      "loss": 1.9895,
+      "step": 5880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2049446095649824e-06,
+      "loss": 1.8585,
+      "step": 5885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.204269116455012e-06,
+      "loss": 2.0616,
+      "step": 5890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.203593623345042e-06,
+      "loss": 2.0316,
+      "step": 5895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2029181302350714e-06,
+      "loss": 2.0547,
+      "step": 5900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2022426371251013e-06,
+      "loss": 2.0177,
+      "step": 5905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2015671440151309e-06,
+      "loss": 1.9315,
+      "step": 5910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2008916509051608e-06,
+      "loss": 2.0497,
+      "step": 5915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2002161577951903e-06,
+      "loss": 2.0856,
+      "step": 5920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1995406646852203e-06,
+      "loss": 1.9031,
+      "step": 5925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1988651715752498e-06,
+      "loss": 2.0394,
+      "step": 5930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1981896784652797e-06,
+      "loss": 1.9831,
+      "step": 5935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1975141853553092e-06,
+      "loss": 1.9886,
+      "step": 5940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1968386922453392e-06,
+      "loss": 2.0037,
+      "step": 5945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1961631991353687e-06,
+      "loss": 2.018,
+      "step": 5950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1954877060253984e-06,
+      "loss": 2.1561,
+      "step": 5955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1948122129154281e-06,
+      "loss": 2.0146,
+      "step": 5960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1941367198054579e-06,
+      "loss": 2.0204,
+      "step": 5965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1934612266954876e-06,
+      "loss": 1.9847,
+      "step": 5970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1927857335855173e-06,
+      "loss": 1.9988,
+      "step": 5975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1921102404755472e-06,
+      "loss": 2.0426,
+      "step": 5980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1914347473655768e-06,
+      "loss": 1.9307,
+      "step": 5985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1907592542556067e-06,
+      "loss": 1.9322,
+      "step": 5990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1900837611456362e-06,
+      "loss": 2.0565,
+      "step": 5995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1894082680356661e-06,
+      "loss": 1.9926,
+      "step": 6000
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 2.0011448860168457,
+      "eval_runtime": 165.1917,
+      "eval_samples_per_second": 3.608,
+      "eval_steps_per_second": 0.454,
+      "step": 6000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1887327749256957e-06,
+      "loss": 2.0437,
+      "step": 6005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1880572818157256e-06,
+      "loss": 1.925,
+      "step": 6010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1873817887057551e-06,
+      "loss": 2.0602,
+      "step": 6015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1867062955957848e-06,
+      "loss": 2.025,
+      "step": 6020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1860308024858146e-06,
+      "loss": 1.9937,
+      "step": 6025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1853553093758443e-06,
+      "loss": 2.0074,
+      "step": 6030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.184679816265874e-06,
+      "loss": 2.0228,
+      "step": 6035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1840043231559037e-06,
+      "loss": 2.0472,
+      "step": 6040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1833288300459335e-06,
+      "loss": 2.0226,
+      "step": 6045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1826533369359632e-06,
+      "loss": 1.9945,
+      "step": 6050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.181977843825993e-06,
+      "loss": 1.9856,
+      "step": 6055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1813023507160227e-06,
+      "loss": 2.0037,
+      "step": 6060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1806268576060524e-06,
+      "loss": 1.9873,
+      "step": 6065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.179951364496082e-06,
+      "loss": 2.0066,
+      "step": 6070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1792758713861118e-06,
+      "loss": 2.0181,
+      "step": 6075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1786003782761416e-06,
+      "loss": 1.9531,
+      "step": 6080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.177924885166171e-06,
+      "loss": 2.0259,
+      "step": 6085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.177249392056201e-06,
+      "loss": 1.9526,
+      "step": 6090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1765738989462305e-06,
+      "loss": 1.9071,
+      "step": 6095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1758984058362605e-06,
+      "loss": 2.0754,
+      "step": 6100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.17522291272629e-06,
+      "loss": 2.0043,
+      "step": 6105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.17454741961632e-06,
+      "loss": 1.986,
+      "step": 6110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1738719265063494e-06,
+      "loss": 2.0672,
+      "step": 6115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1731964333963794e-06,
+      "loss": 2.0446,
+      "step": 6120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.172520940286409e-06,
+      "loss": 2.0913,
+      "step": 6125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1718454471764388e-06,
+      "loss": 2.0568,
+      "step": 6130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1711699540664685e-06,
+      "loss": 1.9685,
+      "step": 6135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1704944609564983e-06,
+      "loss": 2.033,
+      "step": 6140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.169818967846528e-06,
+      "loss": 1.9966,
+      "step": 6145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1691434747365575e-06,
+      "loss": 1.9999,
+      "step": 6150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1684679816265874e-06,
+      "loss": 2.0394,
+      "step": 6155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.167792488516617e-06,
+      "loss": 1.9596,
+      "step": 6160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.167116995406647e-06,
+      "loss": 2.0409,
+      "step": 6165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1664415022966764e-06,
+      "loss": 1.9961,
+      "step": 6170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1657660091867064e-06,
+      "loss": 2.025,
+      "step": 6175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1650905160767359e-06,
+      "loss": 2.0427,
+      "step": 6180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1644150229667658e-06,
+      "loss": 1.9939,
+      "step": 6185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1637395298567953e-06,
+      "loss": 2.061,
+      "step": 6190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1630640367468253e-06,
+      "loss": 2.0645,
+      "step": 6195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1623885436368548e-06,
+      "loss": 2.0212,
+      "step": 6200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1617130505268847e-06,
+      "loss": 2.0376,
+      "step": 6205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1610375574169142e-06,
+      "loss": 1.9781,
+      "step": 6210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.160362064306944e-06,
+      "loss": 2.0583,
+      "step": 6215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1596865711969737e-06,
+      "loss": 2.0362,
+      "step": 6220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1590110780870034e-06,
+      "loss": 2.0105,
+      "step": 6225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1583355849770331e-06,
+      "loss": 2.0859,
+      "step": 6230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1576600918670629e-06,
+      "loss": 1.9903,
+      "step": 6235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1569845987570926e-06,
+      "loss": 1.9948,
+      "step": 6240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1563091056471223e-06,
+      "loss": 1.9782,
+      "step": 6245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.155633612537152e-06,
+      "loss": 1.9493,
+      "step": 6250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1549581194271818e-06,
+      "loss": 2.0101,
+      "step": 6255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1542826263172115e-06,
+      "loss": 2.0458,
+      "step": 6260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1536071332072412e-06,
+      "loss": 1.8974,
+      "step": 6265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1529316400972712e-06,
+      "loss": 1.9465,
+      "step": 6270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1522561469873007e-06,
+      "loss": 1.9905,
+      "step": 6275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1515806538773304e-06,
+      "loss": 2.0912,
+      "step": 6280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1509051607673601e-06,
+      "loss": 1.9464,
+      "step": 6285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1502296676573898e-06,
+      "loss": 1.9497,
+      "step": 6290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1495541745474196e-06,
+      "loss": 2.0405,
+      "step": 6295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1488786814374493e-06,
+      "loss": 1.9306,
+      "step": 6300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148203188327479e-06,
+      "loss": 2.1404,
+      "step": 6305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1475276952175088e-06,
+      "loss": 1.906,
+      "step": 6310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1468522021075385e-06,
+      "loss": 2.0527,
+      "step": 6315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1461767089975682e-06,
+      "loss": 2.057,
+      "step": 6320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.145501215887598e-06,
+      "loss": 1.948,
+      "step": 6325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1448257227776277e-06,
+      "loss": 2.0469,
+      "step": 6330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1441502296676574e-06,
+      "loss": 2.0718,
+      "step": 6335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1434747365576871e-06,
+      "loss": 2.0162,
+      "step": 6340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1427992434477166e-06,
+      "loss": 2.0367,
+      "step": 6345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1421237503377466e-06,
+      "loss": 1.9805,
+      "step": 6350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.141448257227776e-06,
+      "loss": 1.975,
+      "step": 6355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.140772764117806e-06,
+      "loss": 2.1027,
+      "step": 6360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1400972710078355e-06,
+      "loss": 2.0534,
+      "step": 6365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1394217778978655e-06,
+      "loss": 2.0852,
+      "step": 6370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.138746284787895e-06,
+      "loss": 2.0087,
+      "step": 6375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.138070791677925e-06,
+      "loss": 1.9716,
+      "step": 6380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1373952985679544e-06,
+      "loss": 1.9873,
+      "step": 6385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1367198054579844e-06,
+      "loss": 2.0416,
+      "step": 6390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1360443123480139e-06,
+      "loss": 2.0387,
+      "step": 6395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1353688192380438e-06,
+      "loss": 1.9927,
+      "step": 6400
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 1.9952620267868042,
+      "eval_runtime": 165.2045,
+      "eval_samples_per_second": 3.608,
+      "eval_steps_per_second": 0.454,
+      "step": 6400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1346933261280733e-06,
+      "loss": 2.0118,
+      "step": 6405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.134017833018103e-06,
+      "loss": 1.9994,
+      "step": 6410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.133342339908133e-06,
+      "loss": 2.0647,
+      "step": 6415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1326668467981625e-06,
+      "loss": 2.0085,
+      "step": 6420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1319913536881925e-06,
+      "loss": 2.0304,
+      "step": 6425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.131315860578222e-06,
+      "loss": 2.0116,
+      "step": 6430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.130640367468252e-06,
+      "loss": 1.9938,
+      "step": 6435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1299648743582814e-06,
+      "loss": 1.9892,
+      "step": 6440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1292893812483114e-06,
+      "loss": 1.9668,
+      "step": 6445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1286138881383409e-06,
+      "loss": 1.9952,
+      "step": 6450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1279383950283708e-06,
+      "loss": 2.02,
+      "step": 6455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1272629019184003e-06,
+      "loss": 1.9425,
+      "step": 6460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1265874088084303e-06,
+      "loss": 2.1208,
+      "step": 6465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1259119156984598e-06,
+      "loss": 1.9683,
+      "step": 6470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1252364225884895e-06,
+      "loss": 1.9708,
+      "step": 6475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1245609294785192e-06,
+      "loss": 2.0613,
+      "step": 6480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.123885436368549e-06,
+      "loss": 1.9685,
+      "step": 6485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1232099432585787e-06,
+      "loss": 2.028,
+      "step": 6490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1225344501486084e-06,
+      "loss": 2.0237,
+      "step": 6495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1218589570386381e-06,
+      "loss": 2.007,
+      "step": 6500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1211834639286679e-06,
+      "loss": 2.1089,
+      "step": 6505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1205079708186976e-06,
+      "loss": 1.9852,
+      "step": 6510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1198324777087273e-06,
+      "loss": 2.005,
+      "step": 6515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.119156984598757e-06,
+      "loss": 1.8925,
+      "step": 6520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1184814914887868e-06,
+      "loss": 1.9801,
+      "step": 6525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1178059983788165e-06,
+      "loss": 1.9388,
+      "step": 6530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1171305052688462e-06,
+      "loss": 2.0295,
+      "step": 6535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1164550121588757e-06,
+      "loss": 1.9851,
+      "step": 6540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1157795190489057e-06,
+      "loss": 1.9958,
+      "step": 6545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1151040259389352e-06,
+      "loss": 2.0137,
+      "step": 6550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1144285328289651e-06,
+      "loss": 1.9999,
+      "step": 6555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1137530397189949e-06,
+      "loss": 2.0079,
+      "step": 6560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1130775466090246e-06,
+      "loss": 2.0772,
+      "step": 6565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1124020534990543e-06,
+      "loss": 1.9449,
+      "step": 6570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.111726560389084e-06,
+      "loss": 2.0685,
+      "step": 6575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1110510672791138e-06,
+      "loss": 2.0436,
+      "step": 6580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1103755741691435e-06,
+      "loss": 1.9999,
+      "step": 6585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1097000810591732e-06,
+      "loss": 1.9621,
+      "step": 6590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.109024587949203e-06,
+      "loss": 2.018,
+      "step": 6595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1083490948392327e-06,
+      "loss": 1.9571,
+      "step": 6600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1076736017292622e-06,
+      "loss": 1.9302,
+      "step": 6605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1069981086192921e-06,
+      "loss": 1.9668,
+      "step": 6610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1063226155093216e-06,
+      "loss": 2.0856,
+      "step": 6615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1056471223993516e-06,
+      "loss": 1.9749,
+      "step": 6620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.104971629289381e-06,
+      "loss": 1.9511,
+      "step": 6625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.104296136179411e-06,
+      "loss": 2.0511,
+      "step": 6630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1036206430694405e-06,
+      "loss": 2.0733,
+      "step": 6635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1029451499594705e-06,
+      "loss": 1.9882,
+      "step": 6640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1022696568495e-06,
+      "loss": 2.0366,
+      "step": 6645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.10159416373953e-06,
+      "loss": 1.9377,
+      "step": 6650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1009186706295594e-06,
+      "loss": 1.9914,
+      "step": 6655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1002431775195894e-06,
+      "loss": 2.035,
+      "step": 6660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.099567684409619e-06,
+      "loss": 2.0213,
+      "step": 6665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0988921912996486e-06,
+      "loss": 2.0916,
+      "step": 6670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0982166981896783e-06,
+      "loss": 1.9638,
+      "step": 6675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.097541205079708e-06,
+      "loss": 2.0284,
+      "step": 6680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0968657119697378e-06,
+      "loss": 1.9684,
+      "step": 6685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0961902188597675e-06,
+      "loss": 2.0286,
+      "step": 6690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0955147257497973e-06,
+      "loss": 1.94,
+      "step": 6695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.094839232639827e-06,
+      "loss": 2.0547,
+      "step": 6700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.094163739529857e-06,
+      "loss": 1.9955,
+      "step": 6705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0934882464198864e-06,
+      "loss": 2.0647,
+      "step": 6710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0928127533099164e-06,
+      "loss": 1.955,
+      "step": 6715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0921372601999459e-06,
+      "loss": 2.0865,
+      "step": 6720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0914617670899758e-06,
+      "loss": 1.9343,
+      "step": 6725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0907862739800053e-06,
+      "loss": 1.9694,
+      "step": 6730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.090110780870035e-06,
+      "loss": 2.071,
+      "step": 6735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0894352877600648e-06,
+      "loss": 2.0049,
+      "step": 6740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0887597946500945e-06,
+      "loss": 1.9945,
+      "step": 6745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0880843015401242e-06,
+      "loss": 2.0204,
+      "step": 6750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.087408808430154e-06,
+      "loss": 1.9284,
+      "step": 6755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0867333153201837e-06,
+      "loss": 1.9173,
+      "step": 6760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0860578222102134e-06,
+      "loss": 1.9321,
+      "step": 6765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0853823291002431e-06,
+      "loss": 2.0363,
+      "step": 6770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0847068359902729e-06,
+      "loss": 2.053,
+      "step": 6775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0840313428803026e-06,
+      "loss": 1.9384,
+      "step": 6780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0833558497703323e-06,
+      "loss": 1.9329,
+      "step": 6785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.082680356660362e-06,
+      "loss": 2.0216,
+      "step": 6790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0820048635503918e-06,
+      "loss": 2.0901,
+      "step": 6795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0813293704404213e-06,
+      "loss": 2.0983,
+      "step": 6800
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 1.989758014678955,
+      "eval_runtime": 165.4469,
+      "eval_samples_per_second": 3.602,
+      "eval_steps_per_second": 0.453,
+      "step": 6800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0806538773304512e-06,
+      "loss": 2.0018,
+      "step": 6805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0799783842204807e-06,
+      "loss": 1.9857,
+      "step": 6810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0793028911105107e-06,
+      "loss": 2.0529,
+      "step": 6815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0786273980005402e-06,
+      "loss": 1.9866,
+      "step": 6820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0779519048905701e-06,
+      "loss": 2.0546,
+      "step": 6825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0772764117805996e-06,
+      "loss": 1.9727,
+      "step": 6830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0766009186706296e-06,
+      "loss": 2.0949,
+      "step": 6835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.075925425560659e-06,
+      "loss": 2.049,
+      "step": 6840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.075249932450689e-06,
+      "loss": 2.0513,
+      "step": 6845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0745744393407188e-06,
+      "loss": 1.9751,
+      "step": 6850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0738989462307485e-06,
+      "loss": 2.0021,
+      "step": 6855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0732234531207782e-06,
+      "loss": 2.0116,
+      "step": 6860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0725479600108077e-06,
+      "loss": 1.9602,
+      "step": 6865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0718724669008377e-06,
+      "loss": 1.9296,
+      "step": 6870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0711969737908672e-06,
+      "loss": 1.9961,
+      "step": 6875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0705214806808971e-06,
+      "loss": 1.9395,
+      "step": 6880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0698459875709266e-06,
+      "loss": 1.9828,
+      "step": 6885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0691704944609566e-06,
+      "loss": 1.9577,
+      "step": 6890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.068495001350986e-06,
+      "loss": 1.9432,
+      "step": 6895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.067819508241016e-06,
+      "loss": 1.953,
+      "step": 6900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0671440151310455e-06,
+      "loss": 2.0855,
+      "step": 6905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0664685220210755e-06,
+      "loss": 2.021,
+      "step": 6910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.065793028911105e-06,
+      "loss": 2.1253,
+      "step": 6915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.065117535801135e-06,
+      "loss": 1.9398,
+      "step": 6920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0644420426911644e-06,
+      "loss": 1.8746,
+      "step": 6925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0637665495811942e-06,
+      "loss": 1.9618,
+      "step": 6930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.063091056471224e-06,
+      "loss": 1.9594,
+      "step": 6935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0624155633612536e-06,
+      "loss": 1.9582,
+      "step": 6940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0617400702512834e-06,
+      "loss": 2.0177,
+      "step": 6945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.061064577141313e-06,
+      "loss": 2.0466,
+      "step": 6950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0603890840313428e-06,
+      "loss": 1.9999,
+      "step": 6955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0597135909213725e-06,
+      "loss": 1.9628,
+      "step": 6960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0590380978114023e-06,
+      "loss": 2.0182,
+      "step": 6965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.058362604701432e-06,
+      "loss": 2.0239,
+      "step": 6970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0576871115914617e-06,
+      "loss": 1.915,
+      "step": 6975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0570116184814914e-06,
+      "loss": 2.0042,
+      "step": 6980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.056336125371521e-06,
+      "loss": 1.9469,
+      "step": 6985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0556606322615509e-06,
+      "loss": 2.1206,
+      "step": 6990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0549851391515806e-06,
+      "loss": 2.0603,
+      "step": 6995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0543096460416103e-06,
+      "loss": 1.9043,
+      "step": 7000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.05363415293164e-06,
+      "loss": 1.9525,
+      "step": 7005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0529586598216698e-06,
+      "loss": 2.0047,
+      "step": 7010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0522831667116995e-06,
+      "loss": 2.0078,
+      "step": 7015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0516076736017292e-06,
+      "loss": 2.0401,
+      "step": 7020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.050932180491759e-06,
+      "loss": 2.0235,
+      "step": 7025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0502566873817887e-06,
+      "loss": 1.8742,
+      "step": 7030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0495811942718184e-06,
+      "loss": 1.9213,
+      "step": 7035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0489057011618482e-06,
+      "loss": 2.0103,
+      "step": 7040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0482302080518779e-06,
+      "loss": 1.9676,
+      "step": 7045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0475547149419076e-06,
+      "loss": 2.0613,
+      "step": 7050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0468792218319373e-06,
+      "loss": 1.9115,
+      "step": 7055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0462037287219668e-06,
+      "loss": 2.1336,
+      "step": 7060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0455282356119968e-06,
+      "loss": 2.0362,
+      "step": 7065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0448527425020263e-06,
+      "loss": 1.9231,
+      "step": 7070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0441772493920562e-06,
+      "loss": 2.0279,
+      "step": 7075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0435017562820858e-06,
+      "loss": 1.9745,
+      "step": 7080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0428262631721157e-06,
+      "loss": 1.934,
+      "step": 7085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0421507700621452e-06,
+      "loss": 1.9617,
+      "step": 7090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0414752769521751e-06,
+      "loss": 1.9954,
+      "step": 7095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0407997838422047e-06,
+      "loss": 1.9847,
+      "step": 7100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0401242907322346e-06,
+      "loss": 2.0425,
+      "step": 7105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0394487976222641e-06,
+      "loss": 1.9804,
+      "step": 7110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.038773304512294e-06,
+      "loss": 1.9689,
+      "step": 7115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0380978114023236e-06,
+      "loss": 2.0032,
+      "step": 7120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0374223182923533e-06,
+      "loss": 2.0059,
+      "step": 7125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.036746825182383e-06,
+      "loss": 2.0135,
+      "step": 7130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0360713320724127e-06,
+      "loss": 1.94,
+      "step": 7135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0353958389624427e-06,
+      "loss": 1.9191,
+      "step": 7140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0347203458524722e-06,
+      "loss": 1.9087,
+      "step": 7145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0340448527425021e-06,
+      "loss": 1.9729,
+      "step": 7150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0333693596325316e-06,
+      "loss": 2.003,
+      "step": 7155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0326938665225616e-06,
+      "loss": 2.0825,
+      "step": 7160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.032018373412591e-06,
+      "loss": 1.9802,
+      "step": 7165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.031342880302621e-06,
+      "loss": 1.9544,
+      "step": 7170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0306673871926506e-06,
+      "loss": 1.9041,
+      "step": 7175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0299918940826805e-06,
+      "loss": 1.8864,
+      "step": 7180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.02931640097271e-06,
+      "loss": 2.0309,
+      "step": 7185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0286409078627397e-06,
+      "loss": 1.9711,
+      "step": 7190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0279654147527695e-06,
+      "loss": 1.9598,
+      "step": 7195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0272899216427992e-06,
+      "loss": 1.9517,
+      "step": 7200
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 1.985114336013794,
+      "eval_runtime": 165.362,
+      "eval_samples_per_second": 3.604,
+      "eval_steps_per_second": 0.454,
+      "step": 7200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.026614428532829e-06,
+      "loss": 1.8818,
+      "step": 7205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0259389354228586e-06,
+      "loss": 1.9379,
+      "step": 7210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0252634423128884e-06,
+      "loss": 1.9673,
+      "step": 7215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.024587949202918e-06,
+      "loss": 2.0222,
+      "step": 7220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0239124560929478e-06,
+      "loss": 1.9747,
+      "step": 7225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0232369629829775e-06,
+      "loss": 2.088,
+      "step": 7230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0225614698730073e-06,
+      "loss": 2.0251,
+      "step": 7235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.021885976763037e-06,
+      "loss": 2.1025,
+      "step": 7240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0212104836530665e-06,
+      "loss": 1.9793,
+      "step": 7245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0205349905430964e-06,
+      "loss": 2.0546,
+      "step": 7250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.019859497433126e-06,
+      "loss": 2.0635,
+      "step": 7255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.019184004323156e-06,
+      "loss": 1.9224,
+      "step": 7260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0185085112131854e-06,
+      "loss": 2.0158,
+      "step": 7265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0178330181032154e-06,
+      "loss": 2.0352,
+      "step": 7270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0171575249932449e-06,
+      "loss": 2.0249,
+      "step": 7275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0164820318832748e-06,
+      "loss": 1.9468,
+      "step": 7280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0158065387733043e-06,
+      "loss": 1.9318,
+      "step": 7285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0151310456633343e-06,
+      "loss": 2.0633,
+      "step": 7290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.014455552553364e-06,
+      "loss": 2.0459,
+      "step": 7295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0137800594433937e-06,
+      "loss": 2.0734,
+      "step": 7300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0131045663334234e-06,
+      "loss": 2.06,
+      "step": 7305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0124290732234532e-06,
+      "loss": 1.947,
+      "step": 7310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0117535801134829e-06,
+      "loss": 2.0348,
+      "step": 7315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0110780870035124e-06,
+      "loss": 2.0692,
+      "step": 7320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0104025938935423e-06,
+      "loss": 2.02,
+      "step": 7325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0097271007835719e-06,
+      "loss": 1.9659,
+      "step": 7330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0090516076736018e-06,
+      "loss": 1.9602,
+      "step": 7335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0083761145636313e-06,
+      "loss": 1.9835,
+      "step": 7340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0077006214536612e-06,
+      "loss": 1.9815,
+      "step": 7345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0070251283436908e-06,
+      "loss": 1.9859,
+      "step": 7350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0063496352337207e-06,
+      "loss": 2.1161,
+      "step": 7355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0056741421237502e-06,
+      "loss": 2.0387,
+      "step": 7360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0049986490137801e-06,
+      "loss": 2.0205,
+      "step": 7365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0043231559038097e-06,
+      "loss": 2.0478,
+      "step": 7370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0036476627938396e-06,
+      "loss": 2.023,
+      "step": 7375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0029721696838691e-06,
+      "loss": 2.0349,
+      "step": 7380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0022966765738988e-06,
+      "loss": 2.0054,
+      "step": 7385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0016211834639286e-06,
+      "loss": 2.1175,
+      "step": 7390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0009456903539583e-06,
+      "loss": 2.0055,
+      "step": 7395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.000270197243988e-06,
+      "loss": 2.0236,
+      "step": 7400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.995947041340177e-07,
+      "loss": 2.073,
+      "step": 7405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.989192110240475e-07,
+      "loss": 1.9406,
+      "step": 7410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.982437179140772e-07,
+      "loss": 2.0151,
+      "step": 7415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.97568224804107e-07,
+      "loss": 1.9841,
+      "step": 7420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.968927316941367e-07,
+      "loss": 1.8764,
+      "step": 7425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.962172385841664e-07,
+      "loss": 1.982,
+      "step": 7430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.95541745474196e-07,
+      "loss": 1.9922,
+      "step": 7435
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.948662523642258e-07,
+      "loss": 2.0233,
+      "step": 7440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.941907592542556e-07,
+      "loss": 1.9742,
+      "step": 7445
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.935152661442853e-07,
+      "loss": 2.048,
+      "step": 7450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.92839773034315e-07,
+      "loss": 1.9523,
+      "step": 7455
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.921642799243447e-07,
+      "loss": 2.0822,
+      "step": 7460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.914887868143745e-07,
+      "loss": 2.0152,
+      "step": 7465
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.908132937044042e-07,
+      "loss": 2.0208,
+      "step": 7470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.90137800594434e-07,
+      "loss": 1.9042,
+      "step": 7475
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.894623074844636e-07,
+      "loss": 2.0153,
+      "step": 7480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.887868143744934e-07,
+      "loss": 1.9946,
+      "step": 7485
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.88111321264523e-07,
+      "loss": 1.9934,
+      "step": 7490
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.874358281545528e-07,
+      "loss": 2.0187,
+      "step": 7495
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.867603350445825e-07,
+      "loss": 1.9518,
+      "step": 7500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.860848419346123e-07,
+      "loss": 2.0422,
+      "step": 7505
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.85409348824642e-07,
+      "loss": 2.0009,
+      "step": 7510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.847338557146717e-07,
+      "loss": 2.0596,
+      "step": 7515
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.840583626047015e-07,
+      "loss": 2.0612,
+      "step": 7520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.833828694947312e-07,
+      "loss": 1.9486,
+      "step": 7525
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.82707376384761e-07,
+      "loss": 2.0392,
+      "step": 7530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.820318832747906e-07,
+      "loss": 1.9368,
+      "step": 7535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.813563901648204e-07,
+      "loss": 2.0314,
+      "step": 7540
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.8068089705485e-07,
+      "loss": 1.9035,
+      "step": 7545
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.800054039448798e-07,
+      "loss": 2.0037,
+      "step": 7550
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.793299108349095e-07,
+      "loss": 2.0193,
+      "step": 7555
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.786544177249393e-07,
+      "loss": 2.0409,
+      "step": 7560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.77978924614969e-07,
+      "loss": 2.056,
+      "step": 7565
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.773034315049987e-07,
+      "loss": 1.9435,
+      "step": 7570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.766279383950282e-07,
+      "loss": 1.9287,
+      "step": 7575
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.75952445285058e-07,
+      "loss": 1.9647,
+      "step": 7580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.752769521750877e-07,
+      "loss": 1.8393,
+      "step": 7585
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.746014590651174e-07,
+      "loss": 1.9353,
+      "step": 7590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.739259659551471e-07,
+      "loss": 2.0181,
+      "step": 7595
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.732504728451769e-07,
+      "loss": 1.9621,
+      "step": 7600
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 1.9808038473129272,
+      "eval_runtime": 165.1869,
+      "eval_samples_per_second": 3.608,
+      "eval_steps_per_second": 0.454,
+      "step": 7600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.725749797352066e-07,
+      "loss": 2.0002,
+      "step": 7605
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.718994866252363e-07,
+      "loss": 1.9619,
+      "step": 7610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.71223993515266e-07,
+      "loss": 2.032,
+      "step": 7615
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.705485004052958e-07,
+      "loss": 1.8952,
+      "step": 7620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.698730072953255e-07,
+      "loss": 2.0198,
+      "step": 7625
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.691975141853552e-07,
+      "loss": 1.9973,
+      "step": 7630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.68522021075385e-07,
+      "loss": 2.0368,
+      "step": 7635
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.678465279654147e-07,
+      "loss": 1.913,
+      "step": 7640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.671710348554444e-07,
+      "loss": 1.9466,
+      "step": 7645
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.664955417454741e-07,
+      "loss": 1.9923,
+      "step": 7650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.658200486355039e-07,
+      "loss": 2.055,
+      "step": 7655
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.651445555255336e-07,
+      "loss": 1.9588,
+      "step": 7660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.644690624155633e-07,
+      "loss": 1.9861,
+      "step": 7665
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.63793569305593e-07,
+      "loss": 2.0062,
+      "step": 7670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.631180761956228e-07,
+      "loss": 1.992,
+      "step": 7675
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.624425830856525e-07,
+      "loss": 1.9221,
+      "step": 7680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.617670899756822e-07,
+      "loss": 1.9701,
+      "step": 7685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.61091596865712e-07,
+      "loss": 2.0102,
+      "step": 7690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.604161037557417e-07,
+      "loss": 1.9687,
+      "step": 7695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.597406106457714e-07,
+      "loss": 1.98,
+      "step": 7700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.590651175358011e-07,
+      "loss": 1.9699,
+      "step": 7705
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.583896244258308e-07,
+      "loss": 1.9465,
+      "step": 7710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.577141313158606e-07,
+      "loss": 2.04,
+      "step": 7715
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.570386382058903e-07,
+      "loss": 2.0439,
+      "step": 7720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.5636314509592e-07,
+      "loss": 1.9968,
+      "step": 7725
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.556876519859497e-07,
+      "loss": 2.0141,
+      "step": 7730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.550121588759795e-07,
+      "loss": 1.9623,
+      "step": 7735
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.543366657660092e-07,
+      "loss": 2.0191,
+      "step": 7740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.536611726560388e-07,
+      "loss": 1.9723,
+      "step": 7745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.529856795460685e-07,
+      "loss": 2.0218,
+      "step": 7750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.523101864360983e-07,
+      "loss": 1.8727,
+      "step": 7755
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.51634693326128e-07,
+      "loss": 1.8031,
+      "step": 7760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.509592002161577e-07,
+      "loss": 2.0312,
+      "step": 7765
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.502837071061874e-07,
+      "loss": 2.0651,
+      "step": 7770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.496082139962172e-07,
+      "loss": 2.0122,
+      "step": 7775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.489327208862469e-07,
+      "loss": 1.8839,
+      "step": 7780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.482572277762766e-07,
+      "loss": 2.0005,
+      "step": 7785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.475817346663064e-07,
+      "loss": 2.0028,
+      "step": 7790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.469062415563361e-07,
+      "loss": 1.9506,
+      "step": 7795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.462307484463658e-07,
+      "loss": 2.0455,
+      "step": 7800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.455552553363955e-07,
+      "loss": 1.9396,
+      "step": 7805
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.448797622264253e-07,
+      "loss": 2.0283,
+      "step": 7810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.44204269116455e-07,
+      "loss": 2.0555,
+      "step": 7815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.435287760064847e-07,
+      "loss": 2.0782,
+      "step": 7820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.428532828965144e-07,
+      "loss": 2.0381,
+      "step": 7825
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.421777897865442e-07,
+      "loss": 2.0217,
+      "step": 7830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.415022966765739e-07,
+      "loss": 1.9138,
+      "step": 7835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.408268035666036e-07,
+      "loss": 2.0361,
+      "step": 7840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.401513104566333e-07,
+      "loss": 1.9504,
+      "step": 7845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.394758173466631e-07,
+      "loss": 1.9862,
+      "step": 7850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.388003242366928e-07,
+      "loss": 1.923,
+      "step": 7855
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.381248311267225e-07,
+      "loss": 1.9324,
+      "step": 7860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.374493380167521e-07,
+      "loss": 2.098,
+      "step": 7865
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.367738449067819e-07,
+      "loss": 1.949,
+      "step": 7870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.360983517968116e-07,
+      "loss": 2.0113,
+      "step": 7875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.354228586868413e-07,
+      "loss": 1.9433,
+      "step": 7880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.34747365576871e-07,
+      "loss": 2.0162,
+      "step": 7885
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.340718724669008e-07,
+      "loss": 1.9516,
+      "step": 7890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.333963793569305e-07,
+      "loss": 1.9885,
+      "step": 7895
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.327208862469602e-07,
+      "loss": 1.9776,
+      "step": 7900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.3204539313699e-07,
+      "loss": 2.0417,
+      "step": 7905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.313699000270197e-07,
+      "loss": 2.0449,
+      "step": 7910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.306944069170494e-07,
+      "loss": 2.0715,
+      "step": 7915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.300189138070791e-07,
+      "loss": 1.9786,
+      "step": 7920
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.293434206971088e-07,
+      "loss": 1.9668,
+      "step": 7925
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.286679275871385e-07,
+      "loss": 2.0532,
+      "step": 7930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.279924344771682e-07,
+      "loss": 2.0171,
+      "step": 7935
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.273169413671979e-07,
+      "loss": 2.0196,
+      "step": 7940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.266414482572278e-07,
+      "loss": 1.9405,
+      "step": 7945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.259659551472575e-07,
+      "loss": 1.9522,
+      "step": 7950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.252904620372872e-07,
+      "loss": 1.9491,
+      "step": 7955
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.246149689273169e-07,
+      "loss": 1.9541,
+      "step": 7960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.239394758173467e-07,
+      "loss": 2.0164,
+      "step": 7965
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.232639827073764e-07,
+      "loss": 1.9922,
+      "step": 7970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.225884895974061e-07,
+      "loss": 1.9706,
+      "step": 7975
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.219129964874358e-07,
+      "loss": 1.9756,
+      "step": 7980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.212375033774656e-07,
+      "loss": 2.0676,
+      "step": 7985
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.205620102674953e-07,
+      "loss": 2.0105,
+      "step": 7990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.198865171575249e-07,
+      "loss": 1.9872,
+      "step": 7995
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.192110240475546e-07,
+      "loss": 1.9072,
+      "step": 8000
+    },
+    {
+      "epoch": 1.08,
+      "eval_loss": 1.9768513441085815,
+      "eval_runtime": 165.2166,
+      "eval_samples_per_second": 3.607,
+      "eval_steps_per_second": 0.454,
+      "step": 8000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.185355309375844e-07,
+      "loss": 2.021,
+      "step": 8005
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.178600378276141e-07,
+      "loss": 2.0312,
+      "step": 8010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.171845447176438e-07,
+      "loss": 2.0465,
+      "step": 8015
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.165090516076736e-07,
+      "loss": 2.0647,
+      "step": 8020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.158335584977033e-07,
+      "loss": 1.9135,
+      "step": 8025
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.15158065387733e-07,
+      "loss": 1.9873,
+      "step": 8030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.144825722777627e-07,
+      "loss": 2.0321,
+      "step": 8035
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.138070791677925e-07,
+      "loss": 1.8527,
+      "step": 8040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.131315860578222e-07,
+      "loss": 1.9411,
+      "step": 8045
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.124560929478519e-07,
+      "loss": 2.0264,
+      "step": 8050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.117805998378815e-07,
+      "loss": 2.0049,
+      "step": 8055
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.111051067279113e-07,
+      "loss": 1.9286,
+      "step": 8060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.10429613617941e-07,
+      "loss": 2.0032,
+      "step": 8065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.097541205079707e-07,
+      "loss": 1.9846,
+      "step": 8070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.090786273980004e-07,
+      "loss": 2.0288,
+      "step": 8075
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.084031342880302e-07,
+      "loss": 2.0061,
+      "step": 8080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.077276411780599e-07,
+      "loss": 1.9841,
+      "step": 8085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.070521480680897e-07,
+      "loss": 2.0113,
+      "step": 8090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.063766549581194e-07,
+      "loss": 1.9356,
+      "step": 8095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.057011618481492e-07,
+      "loss": 1.9294,
+      "step": 8100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.050256687381789e-07,
+      "loss": 1.9696,
+      "step": 8105
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.043501756282086e-07,
+      "loss": 1.9592,
+      "step": 8110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.036746825182383e-07,
+      "loss": 2.0304,
+      "step": 8115
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.029991894082681e-07,
+      "loss": 1.9916,
+      "step": 8120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.023236962982977e-07,
+      "loss": 2.0205,
+      "step": 8125
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.016482031883274e-07,
+      "loss": 1.9852,
+      "step": 8130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.009727100783571e-07,
+      "loss": 1.9217,
+      "step": 8135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.002972169683869e-07,
+      "loss": 1.929,
+      "step": 8140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.996217238584166e-07,
+      "loss": 1.9051,
+      "step": 8145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.989462307484463e-07,
+      "loss": 1.9446,
+      "step": 8150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.982707376384761e-07,
+      "loss": 1.9425,
+      "step": 8155
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.975952445285058e-07,
+      "loss": 1.9975,
+      "step": 8160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.969197514185355e-07,
+      "loss": 1.988,
+      "step": 8165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.962442583085652e-07,
+      "loss": 2.0169,
+      "step": 8170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.95568765198595e-07,
+      "loss": 2.0097,
+      "step": 8175
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.948932720886247e-07,
+      "loss": 2.0233,
+      "step": 8180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.942177789786543e-07,
+      "loss": 1.9843,
+      "step": 8185
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.93542285868684e-07,
+      "loss": 1.901,
+      "step": 8190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.928667927587138e-07,
+      "loss": 1.9648,
+      "step": 8195
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.921912996487435e-07,
+      "loss": 1.9876,
+      "step": 8200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.915158065387732e-07,
+      "loss": 1.9283,
+      "step": 8205
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.908403134288029e-07,
+      "loss": 1.9734,
+      "step": 8210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.901648203188327e-07,
+      "loss": 1.9392,
+      "step": 8215
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.894893272088624e-07,
+      "loss": 1.9981,
+      "step": 8220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.888138340988921e-07,
+      "loss": 1.9668,
+      "step": 8225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.881383409889218e-07,
+      "loss": 2.1008,
+      "step": 8230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.874628478789517e-07,
+      "loss": 1.8967,
+      "step": 8235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.867873547689814e-07,
+      "loss": 1.9444,
+      "step": 8240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.861118616590111e-07,
+      "loss": 1.9424,
+      "step": 8245
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.854363685490409e-07,
+      "loss": 2.0454,
+      "step": 8250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.847608754390705e-07,
+      "loss": 2.0193,
+      "step": 8255
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.840853823291002e-07,
+      "loss": 1.9897,
+      "step": 8260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.834098892191299e-07,
+      "loss": 1.9687,
+      "step": 8265
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.827343961091597e-07,
+      "loss": 2.0078,
+      "step": 8270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.820589029991894e-07,
+      "loss": 2.022,
+      "step": 8275
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.813834098892191e-07,
+      "loss": 1.9253,
+      "step": 8280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.807079167792488e-07,
+      "loss": 1.9977,
+      "step": 8285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.800324236692786e-07,
+      "loss": 1.9738,
+      "step": 8290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.793569305593083e-07,
+      "loss": 1.9692,
+      "step": 8295
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.78681437449338e-07,
+      "loss": 1.9502,
+      "step": 8300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.780059443393677e-07,
+      "loss": 1.9975,
+      "step": 8305
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.773304512293975e-07,
+      "loss": 2.0219,
+      "step": 8310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.766549581194271e-07,
+      "loss": 1.9972,
+      "step": 8315
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.759794650094568e-07,
+      "loss": 2.0072,
+      "step": 8320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.753039718994865e-07,
+      "loss": 1.9763,
+      "step": 8325
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.746284787895163e-07,
+      "loss": 2.0124,
+      "step": 8330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.73952985679546e-07,
+      "loss": 1.9558,
+      "step": 8335
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.732774925695757e-07,
+      "loss": 1.971,
+      "step": 8340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.726019994596054e-07,
+      "loss": 2.0614,
+      "step": 8345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.719265063496352e-07,
+      "loss": 2.0482,
+      "step": 8350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.712510132396649e-07,
+      "loss": 1.9451,
+      "step": 8355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.705755201296946e-07,
+      "loss": 1.9871,
+      "step": 8360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.699000270197243e-07,
+      "loss": 1.9311,
+      "step": 8365
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.692245339097541e-07,
+      "loss": 1.9456,
+      "step": 8370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.685490407997838e-07,
+      "loss": 2.0216,
+      "step": 8375
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.678735476898136e-07,
+      "loss": 1.9703,
+      "step": 8380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.671980545798433e-07,
+      "loss": 1.9589,
+      "step": 8385
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.66522561469873e-07,
+      "loss": 2.006,
+      "step": 8390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.658470683599027e-07,
+      "loss": 1.9825,
+      "step": 8395
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.651715752499324e-07,
+      "loss": 2.0099,
+      "step": 8400
+    },
+    {
+      "epoch": 1.13,
+      "eval_loss": 1.9731146097183228,
+      "eval_runtime": 165.3898,
+      "eval_samples_per_second": 3.604,
+      "eval_steps_per_second": 0.453,
+      "step": 8400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.644960821399622e-07,
+      "loss": 1.9268,
+      "step": 8405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.638205890299919e-07,
+      "loss": 2.0344,
+      "step": 8410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.631450959200216e-07,
+      "loss": 1.961,
+      "step": 8415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.624696028100513e-07,
+      "loss": 2.0161,
+      "step": 8420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.617941097000811e-07,
+      "loss": 2.0132,
+      "step": 8425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.611186165901108e-07,
+      "loss": 1.9307,
+      "step": 8430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.604431234801405e-07,
+      "loss": 2.0199,
+      "step": 8435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.597676303701702e-07,
+      "loss": 1.9297,
+      "step": 8440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.590921372601999e-07,
+      "loss": 1.9283,
+      "step": 8445
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.584166441502296e-07,
+      "loss": 2.0167,
+      "step": 8450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.577411510402593e-07,
+      "loss": 1.9519,
+      "step": 8455
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.57065657930289e-07,
+      "loss": 1.9769,
+      "step": 8460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.563901648203188e-07,
+      "loss": 1.9857,
+      "step": 8465
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.557146717103485e-07,
+      "loss": 1.9416,
+      "step": 8470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.550391786003782e-07,
+      "loss": 1.989,
+      "step": 8475
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.543636854904079e-07,
+      "loss": 1.9692,
+      "step": 8480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.536881923804377e-07,
+      "loss": 1.9521,
+      "step": 8485
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.530126992704674e-07,
+      "loss": 2.012,
+      "step": 8490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.523372061604971e-07,
+      "loss": 2.0318,
+      "step": 8495
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.516617130505268e-07,
+      "loss": 2.0502,
+      "step": 8500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.509862199405566e-07,
+      "loss": 2.0098,
+      "step": 8505
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.503107268305862e-07,
+      "loss": 1.9996,
+      "step": 8510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.496352337206159e-07,
+      "loss": 2.0079,
+      "step": 8515
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.489597406106456e-07,
+      "loss": 1.9978,
+      "step": 8520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.482842475006755e-07,
+      "loss": 1.9966,
+      "step": 8525
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.476087543907052e-07,
+      "loss": 1.9699,
+      "step": 8530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.469332612807349e-07,
+      "loss": 2.0915,
+      "step": 8535
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.462577681707647e-07,
+      "loss": 1.9294,
+      "step": 8540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.455822750607944e-07,
+      "loss": 1.9252,
+      "step": 8545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.449067819508241e-07,
+      "loss": 1.9872,
+      "step": 8550
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.442312888408538e-07,
+      "loss": 1.9805,
+      "step": 8555
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.435557957308836e-07,
+      "loss": 2.0113,
+      "step": 8560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.428803026209133e-07,
+      "loss": 1.9835,
+      "step": 8565
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.42204809510943e-07,
+      "loss": 2.0882,
+      "step": 8570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.415293164009726e-07,
+      "loss": 2.0025,
+      "step": 8575
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.408538232910024e-07,
+      "loss": 1.9061,
+      "step": 8580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.401783301810321e-07,
+      "loss": 1.9768,
+      "step": 8585
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.395028370710618e-07,
+      "loss": 2.023,
+      "step": 8590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.388273439610915e-07,
+      "loss": 2.0458,
+      "step": 8595
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.381518508511213e-07,
+      "loss": 2.0728,
+      "step": 8600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.37476357741151e-07,
+      "loss": 2.0385,
+      "step": 8605
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.368008646311807e-07,
+      "loss": 1.9309,
+      "step": 8610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.361253715212104e-07,
+      "loss": 1.9162,
+      "step": 8615
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.354498784112402e-07,
+      "loss": 1.9476,
+      "step": 8620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.347743853012699e-07,
+      "loss": 2.0305,
+      "step": 8625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.340988921912996e-07,
+      "loss": 2.0251,
+      "step": 8630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.334233990813294e-07,
+      "loss": 1.9413,
+      "step": 8635
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.32747905971359e-07,
+      "loss": 1.8424,
+      "step": 8640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.320724128613887e-07,
+      "loss": 1.9744,
+      "step": 8645
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.313969197514184e-07,
+      "loss": 1.9232,
+      "step": 8650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.307214266414482e-07,
+      "loss": 2.0065,
+      "step": 8655
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.300459335314779e-07,
+      "loss": 1.8987,
+      "step": 8660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.293704404215076e-07,
+      "loss": 1.9654,
+      "step": 8665
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.286949473115374e-07,
+      "loss": 2.0159,
+      "step": 8670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.280194542015672e-07,
+      "loss": 1.9236,
+      "step": 8675
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.273439610915969e-07,
+      "loss": 1.9523,
+      "step": 8680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.266684679816266e-07,
+      "loss": 1.9601,
+      "step": 8685
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.259929748716563e-07,
+      "loss": 1.8293,
+      "step": 8690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.253174817616861e-07,
+      "loss": 1.99,
+      "step": 8695
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.246419886517158e-07,
+      "loss": 1.9628,
+      "step": 8700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.239664955417454e-07,
+      "loss": 1.9766,
+      "step": 8705
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.232910024317751e-07,
+      "loss": 1.8718,
+      "step": 8710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.226155093218049e-07,
+      "loss": 1.9567,
+      "step": 8715
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.219400162118346e-07,
+      "loss": 1.8965,
+      "step": 8720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.212645231018643e-07,
+      "loss": 1.9987,
+      "step": 8725
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.20589029991894e-07,
+      "loss": 1.9691,
+      "step": 8730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.199135368819238e-07,
+      "loss": 1.9517,
+      "step": 8735
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.192380437719535e-07,
+      "loss": 2.0101,
+      "step": 8740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.185625506619832e-07,
+      "loss": 1.9377,
+      "step": 8745
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.17887057552013e-07,
+      "loss": 2.0027,
+      "step": 8750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.172115644420427e-07,
+      "loss": 1.939,
+      "step": 8755
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.165360713320724e-07,
+      "loss": 2.0467,
+      "step": 8760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.158605782221021e-07,
+      "loss": 1.9963,
+      "step": 8765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.151850851121318e-07,
+      "loss": 1.9886,
+      "step": 8770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.145095920021615e-07,
+      "loss": 1.944,
+      "step": 8775
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.138340988921912e-07,
+      "loss": 1.8703,
+      "step": 8780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.131586057822209e-07,
+      "loss": 1.886,
+      "step": 8785
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.124831126722507e-07,
+      "loss": 1.979,
+      "step": 8790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.118076195622804e-07,
+      "loss": 1.9356,
+      "step": 8795
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.111321264523101e-07,
+      "loss": 1.9588,
+      "step": 8800
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 1.9695961475372314,
+      "eval_runtime": 165.3874,
+      "eval_samples_per_second": 3.604,
+      "eval_steps_per_second": 0.453,
+      "step": 8800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.104566333423398e-07,
+      "loss": 1.9797,
+      "step": 8805
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.097811402323696e-07,
+      "loss": 1.9845,
+      "step": 8810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.091056471223993e-07,
+      "loss": 1.9642,
+      "step": 8815
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.084301540124291e-07,
+      "loss": 1.9069,
+      "step": 8820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.077546609024588e-07,
+      "loss": 2.1011,
+      "step": 8825
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.070791677924886e-07,
+      "loss": 2.0182,
+      "step": 8830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.064036746825182e-07,
+      "loss": 1.8241,
+      "step": 8835
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.057281815725479e-07,
+      "loss": 1.9258,
+      "step": 8840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.050526884625776e-07,
+      "loss": 1.9356,
+      "step": 8845
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.043771953526074e-07,
+      "loss": 1.959,
+      "step": 8850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.037017022426371e-07,
+      "loss": 1.9901,
+      "step": 8855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.030262091326668e-07,
+      "loss": 1.9471,
+      "step": 8860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.023507160226966e-07,
+      "loss": 2.0966,
+      "step": 8865
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.016752229127263e-07,
+      "loss": 1.9141,
+      "step": 8870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.00999729802756e-07,
+      "loss": 2.0125,
+      "step": 8875
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.003242366927857e-07,
+      "loss": 1.9031,
+      "step": 8880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.996487435828155e-07,
+      "loss": 2.0249,
+      "step": 8885
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.989732504728452e-07,
+      "loss": 1.9745,
+      "step": 8890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.982977573628749e-07,
+      "loss": 1.9759,
+      "step": 8895
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.976222642529045e-07,
+      "loss": 1.9258,
+      "step": 8900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.969467711429343e-07,
+      "loss": 2.013,
+      "step": 8905
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.96271278032964e-07,
+      "loss": 1.9858,
+      "step": 8910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.955957849229937e-07,
+      "loss": 1.9655,
+      "step": 8915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.949202918130234e-07,
+      "loss": 2.0186,
+      "step": 8920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.942447987030532e-07,
+      "loss": 1.9713,
+      "step": 8925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.935693055930829e-07,
+      "loss": 1.9275,
+      "step": 8930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.928938124831126e-07,
+      "loss": 1.8962,
+      "step": 8935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.922183193731423e-07,
+      "loss": 2.063,
+      "step": 8940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.915428262631721e-07,
+      "loss": 1.8896,
+      "step": 8945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.908673331532018e-07,
+      "loss": 1.966,
+      "step": 8950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.901918400432315e-07,
+      "loss": 1.9818,
+      "step": 8955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.895163469332611e-07,
+      "loss": 2.0381,
+      "step": 8960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.88840853823291e-07,
+      "loss": 1.9868,
+      "step": 8965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.881653607133207e-07,
+      "loss": 1.9937,
+      "step": 8970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.874898676033504e-07,
+      "loss": 1.9347,
+      "step": 8975
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.868143744933801e-07,
+      "loss": 1.9733,
+      "step": 8980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.861388813834099e-07,
+      "loss": 1.9135,
+      "step": 8985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.854633882734396e-07,
+      "loss": 1.9111,
+      "step": 8990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.847878951634693e-07,
+      "loss": 2.0278,
+      "step": 8995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.841124020534991e-07,
+      "loss": 2.0025,
+      "step": 9000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.834369089435288e-07,
+      "loss": 1.9607,
+      "step": 9005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.827614158335585e-07,
+      "loss": 1.966,
+      "step": 9010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.820859227235882e-07,
+      "loss": 2.0316,
+      "step": 9015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.81410429613618e-07,
+      "loss": 2.0147,
+      "step": 9020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.807349365036477e-07,
+      "loss": 1.9391,
+      "step": 9025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.800594433936773e-07,
+      "loss": 2.0185,
+      "step": 9030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.79383950283707e-07,
+      "loss": 1.9343,
+      "step": 9035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.787084571737368e-07,
+      "loss": 1.9053,
+      "step": 9040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.780329640637665e-07,
+      "loss": 1.9645,
+      "step": 9045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.773574709537962e-07,
+      "loss": 1.9119,
+      "step": 9050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.766819778438259e-07,
+      "loss": 1.9916,
+      "step": 9055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.760064847338557e-07,
+      "loss": 1.978,
+      "step": 9060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.753309916238854e-07,
+      "loss": 1.9838,
+      "step": 9065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.746554985139151e-07,
+      "loss": 2.0075,
+      "step": 9070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.739800054039448e-07,
+      "loss": 2.0059,
+      "step": 9075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.733045122939746e-07,
+      "loss": 1.9458,
+      "step": 9080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.726290191840043e-07,
+      "loss": 2.0189,
+      "step": 9085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.719535260740339e-07,
+      "loss": 1.907,
+      "step": 9090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.712780329640636e-07,
+      "loss": 2.0197,
+      "step": 9095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.706025398540934e-07,
+      "loss": 1.9569,
+      "step": 9100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.699270467441231e-07,
+      "loss": 2.0457,
+      "step": 9105
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.692515536341529e-07,
+      "loss": 1.9715,
+      "step": 9110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.685760605241827e-07,
+      "loss": 1.9375,
+      "step": 9115
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.679005674142124e-07,
+      "loss": 1.9681,
+      "step": 9120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.672250743042421e-07,
+      "loss": 2.0124,
+      "step": 9125
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.665495811942718e-07,
+      "loss": 1.9938,
+      "step": 9130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.658740880843016e-07,
+      "loss": 1.9986,
+      "step": 9135
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.651985949743313e-07,
+      "loss": 2.0119,
+      "step": 9140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.64523101864361e-07,
+      "loss": 1.989,
+      "step": 9145
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.638476087543907e-07,
+      "loss": 1.9794,
+      "step": 9150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.631721156444205e-07,
+      "loss": 1.9841,
+      "step": 9155
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.624966225344501e-07,
+      "loss": 2.0012,
+      "step": 9160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.618211294244798e-07,
+      "loss": 2.0727,
+      "step": 9165
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.611456363145095e-07,
+      "loss": 2.0328,
+      "step": 9170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.604701432045393e-07,
+      "loss": 1.9844,
+      "step": 9175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.59794650094569e-07,
+      "loss": 2.0104,
+      "step": 9180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.591191569845987e-07,
+      "loss": 1.9992,
+      "step": 9185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.584436638746284e-07,
+      "loss": 2.0314,
+      "step": 9190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.577681707646582e-07,
+      "loss": 1.9967,
+      "step": 9195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.570926776546879e-07,
+      "loss": 2.0098,
+      "step": 9200
+    },
+    {
+      "epoch": 1.24,
+      "eval_loss": 1.96656334400177,
+      "eval_runtime": 165.5096,
+      "eval_samples_per_second": 3.601,
+      "eval_steps_per_second": 0.453,
+      "step": 9200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.564171845447176e-07,
+      "loss": 1.9014,
+      "step": 9205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.557416914347473e-07,
+      "loss": 2.0112,
+      "step": 9210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.550661983247771e-07,
+      "loss": 1.981,
+      "step": 9215
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.543907052148067e-07,
+      "loss": 2.0282,
+      "step": 9220
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.537152121048364e-07,
+      "loss": 1.888,
+      "step": 9225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.530397189948661e-07,
+      "loss": 1.957,
+      "step": 9230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.523642258848959e-07,
+      "loss": 2.0638,
+      "step": 9235
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.516887327749256e-07,
+      "loss": 2.0383,
+      "step": 9240
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.510132396649553e-07,
+      "loss": 2.1005,
+      "step": 9245
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.50337746554985e-07,
+      "loss": 1.9186,
+      "step": 9250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.496622534450149e-07,
+      "loss": 1.9245,
+      "step": 9255
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.489867603350446e-07,
+      "loss": 1.9879,
+      "step": 9260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.483112672250743e-07,
+      "loss": 1.9247,
+      "step": 9265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.476357741151041e-07,
+      "loss": 1.9621,
+      "step": 9270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.469602810051338e-07,
+      "loss": 1.8959,
+      "step": 9275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.462847878951635e-07,
+      "loss": 2.0015,
+      "step": 9280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.456092947851932e-07,
+      "loss": 2.0016,
+      "step": 9285
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.449338016752229e-07,
+      "loss": 2.0414,
+      "step": 9290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.442583085652526e-07,
+      "loss": 1.8762,
+      "step": 9295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.435828154552823e-07,
+      "loss": 1.9855,
+      "step": 9300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.42907322345312e-07,
+      "loss": 1.8853,
+      "step": 9305
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.422318292353418e-07,
+      "loss": 1.9256,
+      "step": 9310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.415563361253715e-07,
+      "loss": 1.9988,
+      "step": 9315
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.408808430154012e-07,
+      "loss": 1.9554,
+      "step": 9320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.402053499054309e-07,
+      "loss": 1.9522,
+      "step": 9325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.395298567954607e-07,
+      "loss": 1.9611,
+      "step": 9330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.388543636854904e-07,
+      "loss": 1.9107,
+      "step": 9335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.381788705755201e-07,
+      "loss": 1.9087,
+      "step": 9340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.375033774655498e-07,
+      "loss": 1.8861,
+      "step": 9345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.368278843555795e-07,
+      "loss": 1.9844,
+      "step": 9350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.361523912456092e-07,
+      "loss": 1.9853,
+      "step": 9355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.354768981356389e-07,
+      "loss": 1.9048,
+      "step": 9360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.348014050256686e-07,
+      "loss": 1.9164,
+      "step": 9365
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.341259119156984e-07,
+      "loss": 1.9002,
+      "step": 9370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.334504188057281e-07,
+      "loss": 1.9329,
+      "step": 9375
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.327749256957578e-07,
+      "loss": 1.8667,
+      "step": 9380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.320994325857876e-07,
+      "loss": 1.9071,
+      "step": 9385
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.314239394758173e-07,
+      "loss": 1.9801,
+      "step": 9390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.30748446365847e-07,
+      "loss": 1.8903,
+      "step": 9395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.300729532558768e-07,
+      "loss": 2.0193,
+      "step": 9400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.293974601459066e-07,
+      "loss": 1.9583,
+      "step": 9405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.287219670359363e-07,
+      "loss": 1.9964,
+      "step": 9410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.28046473925966e-07,
+      "loss": 1.9577,
+      "step": 9415
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.273709808159956e-07,
+      "loss": 1.9443,
+      "step": 9420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.266954877060254e-07,
+      "loss": 2.057,
+      "step": 9425
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.260199945960551e-07,
+      "loss": 2.0193,
+      "step": 9430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.253445014860848e-07,
+      "loss": 2.0085,
+      "step": 9435
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.246690083761145e-07,
+      "loss": 1.9692,
+      "step": 9440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.239935152661443e-07,
+      "loss": 1.9368,
+      "step": 9445
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.23318022156174e-07,
+      "loss": 1.9555,
+      "step": 9450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.226425290462037e-07,
+      "loss": 1.9812,
+      "step": 9455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.219670359362334e-07,
+      "loss": 1.9442,
+      "step": 9460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.212915428262632e-07,
+      "loss": 1.9647,
+      "step": 9465
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.206160497162929e-07,
+      "loss": 2.0252,
+      "step": 9470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.199405566063226e-07,
+      "loss": 1.9675,
+      "step": 9475
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.192650634963522e-07,
+      "loss": 1.9984,
+      "step": 9480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.18589570386382e-07,
+      "loss": 2.0282,
+      "step": 9485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.179140772764117e-07,
+      "loss": 1.9355,
+      "step": 9490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.172385841664414e-07,
+      "loss": 2.0768,
+      "step": 9495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.165630910564712e-07,
+      "loss": 2.0453,
+      "step": 9500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.158875979465009e-07,
+      "loss": 1.9795,
+      "step": 9505
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.152121048365306e-07,
+      "loss": 1.9689,
+      "step": 9510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.145366117265603e-07,
+      "loss": 2.0186,
+      "step": 9515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.138611186165901e-07,
+      "loss": 2.0045,
+      "step": 9520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.131856255066198e-07,
+      "loss": 2.0813,
+      "step": 9525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.125101323966495e-07,
+      "loss": 1.9417,
+      "step": 9530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.118346392866792e-07,
+      "loss": 1.8859,
+      "step": 9535
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.11159146176709e-07,
+      "loss": 1.9949,
+      "step": 9540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.104836530667388e-07,
+      "loss": 2.0046,
+      "step": 9545
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.098081599567684e-07,
+      "loss": 1.9243,
+      "step": 9550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.091326668467981e-07,
+      "loss": 1.9855,
+      "step": 9555
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.084571737368279e-07,
+      "loss": 1.9326,
+      "step": 9560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.077816806268576e-07,
+      "loss": 2.0766,
+      "step": 9565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.071061875168873e-07,
+      "loss": 2.016,
+      "step": 9570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.06430694406917e-07,
+      "loss": 2.012,
+      "step": 9575
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.057552012969468e-07,
+      "loss": 1.9261,
+      "step": 9580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.050797081869765e-07,
+      "loss": 2.047,
+      "step": 9585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.044042150770062e-07,
+      "loss": 1.9532,
+      "step": 9590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.03728721967036e-07,
+      "loss": 2.0149,
+      "step": 9595
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.030532288570657e-07,
+      "loss": 1.9688,
+      "step": 9600
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 1.963936448097229,
+      "eval_runtime": 165.4614,
+      "eval_samples_per_second": 3.602,
+      "eval_steps_per_second": 0.453,
+      "step": 9600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.023777357470954e-07,
+      "loss": 1.9251,
+      "step": 9605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.01702242637125e-07,
+      "loss": 2.0004,
+      "step": 9610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.010267495271548e-07,
+      "loss": 1.903,
+      "step": 9615
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.003512564171845e-07,
+      "loss": 2.0285,
+      "step": 9620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.996757633072142e-07,
+      "loss": 1.9815,
+      "step": 9625
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.990002701972439e-07,
+      "loss": 1.8968,
+      "step": 9630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.983247770872737e-07,
+      "loss": 1.9809,
+      "step": 9635
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.976492839773034e-07,
+      "loss": 1.9713,
+      "step": 9640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.969737908673331e-07,
+      "loss": 1.9472,
+      "step": 9645
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.962982977573628e-07,
+      "loss": 1.9328,
+      "step": 9650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.956228046473926e-07,
+      "loss": 1.9631,
+      "step": 9655
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.949473115374223e-07,
+      "loss": 2.0051,
+      "step": 9660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.94271818427452e-07,
+      "loss": 1.9866,
+      "step": 9665
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.935963253174817e-07,
+      "loss": 1.9494,
+      "step": 9670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.929208322075114e-07,
+      "loss": 1.9731,
+      "step": 9675
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.922453390975411e-07,
+      "loss": 1.9718,
+      "step": 9680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.915698459875708e-07,
+      "loss": 1.9197,
+      "step": 9685
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.908943528776006e-07,
+      "loss": 2.0505,
+      "step": 9690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.902188597676304e-07,
+      "loss": 2.0586,
+      "step": 9695
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.895433666576601e-07,
+      "loss": 1.9232,
+      "step": 9700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.888678735476898e-07,
+      "loss": 1.9559,
+      "step": 9705
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.881923804377195e-07,
+      "loss": 2.0318,
+      "step": 9710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.875168873277493e-07,
+      "loss": 1.8752,
+      "step": 9715
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.86841394217779e-07,
+      "loss": 1.9969,
+      "step": 9720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.861659011078087e-07,
+      "loss": 1.9531,
+      "step": 9725
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.854904079978385e-07,
+      "loss": 2.0319,
+      "step": 9730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.848149148878682e-07,
+      "loss": 1.8927,
+      "step": 9735
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.841394217778978e-07,
+      "loss": 1.8673,
+      "step": 9740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.834639286679275e-07,
+      "loss": 2.0187,
+      "step": 9745
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.827884355579573e-07,
+      "loss": 2.0092,
+      "step": 9750
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.82112942447987e-07,
+      "loss": 2.0315,
+      "step": 9755
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.814374493380167e-07,
+      "loss": 1.9539,
+      "step": 9760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.807619562280464e-07,
+      "loss": 1.8357,
+      "step": 9765
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.800864631180762e-07,
+      "loss": 2.0681,
+      "step": 9770
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.794109700081059e-07,
+      "loss": 1.9327,
+      "step": 9775
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.787354768981356e-07,
+      "loss": 1.916,
+      "step": 9780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.780599837881653e-07,
+      "loss": 1.9925,
+      "step": 9785
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.773844906781951e-07,
+      "loss": 1.9407,
+      "step": 9790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.767089975682248e-07,
+      "loss": 1.9569,
+      "step": 9795
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.760335044582545e-07,
+      "loss": 1.9141,
+      "step": 9800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.753580113482841e-07,
+      "loss": 1.9845,
+      "step": 9805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.746825182383139e-07,
+      "loss": 2.0271,
+      "step": 9810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.740070251283436e-07,
+      "loss": 1.9355,
+      "step": 9815
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.733315320183733e-07,
+      "loss": 1.9558,
+      "step": 9820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.72656038908403e-07,
+      "loss": 1.9754,
+      "step": 9825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.719805457984328e-07,
+      "loss": 2.0408,
+      "step": 9830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.713050526884625e-07,
+      "loss": 1.9422,
+      "step": 9835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.706295595784923e-07,
+      "loss": 2.0819,
+      "step": 9840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.699540664685221e-07,
+      "loss": 1.9963,
+      "step": 9845
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.692785733585518e-07,
+      "loss": 1.9835,
+      "step": 9850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.686030802485815e-07,
+      "loss": 1.9566,
+      "step": 9855
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.679275871386112e-07,
+      "loss": 2.0353,
+      "step": 9860
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.67252094028641e-07,
+      "loss": 2.0683,
+      "step": 9865
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.665766009186706e-07,
+      "loss": 1.9563,
+      "step": 9870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.659011078087003e-07,
+      "loss": 1.8663,
+      "step": 9875
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.6522561469873e-07,
+      "loss": 1.9794,
+      "step": 9880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.645501215887598e-07,
+      "loss": 1.9824,
+      "step": 9885
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.638746284787895e-07,
+      "loss": 1.9124,
+      "step": 9890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.631991353688192e-07,
+      "loss": 1.9687,
+      "step": 9895
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.625236422588489e-07,
+      "loss": 2.0313,
+      "step": 9900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.618481491488787e-07,
+      "loss": 1.943,
+      "step": 9905
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.611726560389084e-07,
+      "loss": 1.9226,
+      "step": 9910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.604971629289381e-07,
+      "loss": 2.0105,
+      "step": 9915
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.598216698189678e-07,
+      "loss": 2.0327,
+      "step": 9920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.591461767089976e-07,
+      "loss": 1.9641,
+      "step": 9925
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.584706835990273e-07,
+      "loss": 1.9631,
+      "step": 9930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.577951904890569e-07,
+      "loss": 1.9812,
+      "step": 9935
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.571196973790866e-07,
+      "loss": 2.0004,
+      "step": 9940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.564442042691164e-07,
+      "loss": 1.8798,
+      "step": 9945
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.557687111591461e-07,
+      "loss": 2.0338,
+      "step": 9950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.550932180491758e-07,
+      "loss": 1.9378,
+      "step": 9955
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.544177249392055e-07,
+      "loss": 1.8363,
+      "step": 9960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.537422318292353e-07,
+      "loss": 1.9079,
+      "step": 9965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.53066738719265e-07,
+      "loss": 1.8645,
+      "step": 9970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.523912456092947e-07,
+      "loss": 1.9543,
+      "step": 9975
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.517157524993245e-07,
+      "loss": 1.942,
+      "step": 9980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.510402593893543e-07,
+      "loss": 1.9462,
+      "step": 9985
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.50364766279384e-07,
+      "loss": 2.1435,
+      "step": 9990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.496892731694137e-07,
+      "loss": 2.0368,
+      "step": 9995
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.490137800594434e-07,
+      "loss": 1.9291,
+      "step": 10000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 1.961105465888977,
+      "eval_runtime": 165.405,
+      "eval_samples_per_second": 3.603,
+      "eval_steps_per_second": 0.453,
+      "step": 10000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.483382869494731e-07,
+      "loss": 1.9377,
+      "step": 10005
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.476627938395028e-07,
+      "loss": 1.9865,
+      "step": 10010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.469873007295325e-07,
+      "loss": 1.9385,
+      "step": 10015
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.463118076195623e-07,
+      "loss": 1.9832,
+      "step": 10020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.45636314509592e-07,
+      "loss": 2.0501,
+      "step": 10025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.449608213996217e-07,
+      "loss": 1.931,
+      "step": 10030
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.442853282896514e-07,
+      "loss": 1.9716,
+      "step": 10035
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.436098351796812e-07,
+      "loss": 2.1156,
+      "step": 10040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.429343420697109e-07,
+      "loss": 1.9412,
+      "step": 10045
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.422588489597406e-07,
+      "loss": 1.9237,
+      "step": 10050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.415833558497703e-07,
+      "loss": 1.9539,
+      "step": 10055
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.409078627398001e-07,
+      "loss": 1.9547,
+      "step": 10060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.402323696298297e-07,
+      "loss": 1.9443,
+      "step": 10065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.395568765198594e-07,
+      "loss": 1.8834,
+      "step": 10070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.388813834098891e-07,
+      "loss": 1.996,
+      "step": 10075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.382058902999189e-07,
+      "loss": 1.9791,
+      "step": 10080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.375303971899486e-07,
+      "loss": 2.0237,
+      "step": 10085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.368549040799783e-07,
+      "loss": 1.955,
+      "step": 10090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.36179410970008e-07,
+      "loss": 2.0109,
+      "step": 10095
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.355039178600378e-07,
+      "loss": 1.9346,
+      "step": 10100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.348284247500675e-07,
+      "loss": 2.0298,
+      "step": 10105
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.341529316400972e-07,
+      "loss": 2.0031,
+      "step": 10110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.33477438530127e-07,
+      "loss": 1.887,
+      "step": 10115
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.328019454201567e-07,
+      "loss": 1.955,
+      "step": 10120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.321264523101863e-07,
+      "loss": 1.9275,
+      "step": 10125
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.314509592002161e-07,
+      "loss": 1.8876,
+      "step": 10130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.307754660902459e-07,
+      "loss": 1.9722,
+      "step": 10135
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.300999729802756e-07,
+      "loss": 1.9802,
+      "step": 10140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.294244798703053e-07,
+      "loss": 2.0026,
+      "step": 10145
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.28748986760335e-07,
+      "loss": 2.0283,
+      "step": 10150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.280734936503648e-07,
+      "loss": 1.9919,
+      "step": 10155
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.273980005403945e-07,
+      "loss": 1.915,
+      "step": 10160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.267225074304242e-07,
+      "loss": 1.9717,
+      "step": 10165
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.260470143204539e-07,
+      "loss": 1.9684,
+      "step": 10170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.253715212104837e-07,
+      "loss": 2.023,
+      "step": 10175
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.246960281005134e-07,
+      "loss": 1.8704,
+      "step": 10180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.240205349905431e-07,
+      "loss": 1.9476,
+      "step": 10185
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.233450418805728e-07,
+      "loss": 1.975,
+      "step": 10190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.226695487706025e-07,
+      "loss": 1.8549,
+      "step": 10195
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.219940556606322e-07,
+      "loss": 2.0041,
+      "step": 10200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.213185625506619e-07,
+      "loss": 1.9219,
+      "step": 10205
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.206430694406916e-07,
+      "loss": 2.0199,
+      "step": 10210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.199675763307214e-07,
+      "loss": 2.0977,
+      "step": 10215
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.192920832207511e-07,
+      "loss": 1.9502,
+      "step": 10220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.186165901107808e-07,
+      "loss": 2.0295,
+      "step": 10225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.179410970008106e-07,
+      "loss": 1.8866,
+      "step": 10230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.172656038908403e-07,
+      "loss": 1.9987,
+      "step": 10235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.1659011078087e-07,
+      "loss": 2.0693,
+      "step": 10240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.159146176708997e-07,
+      "loss": 1.9641,
+      "step": 10245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.152391245609295e-07,
+      "loss": 1.8745,
+      "step": 10250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.145636314509591e-07,
+      "loss": 1.9384,
+      "step": 10255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.138881383409888e-07,
+      "loss": 1.9278,
+      "step": 10260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.132126452310185e-07,
+      "loss": 2.0378,
+      "step": 10265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.125371521210483e-07,
+      "loss": 2.0384,
+      "step": 10270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.118616590110781e-07,
+      "loss": 1.9244,
+      "step": 10275
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.111861659011078e-07,
+      "loss": 1.9505,
+      "step": 10280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.105106727911375e-07,
+      "loss": 1.9794,
+      "step": 10285
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.098351796811673e-07,
+      "loss": 1.9922,
+      "step": 10290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.09159686571197e-07,
+      "loss": 1.9708,
+      "step": 10295
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.084841934612267e-07,
+      "loss": 2.0161,
+      "step": 10300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.078087003512564e-07,
+      "loss": 1.9192,
+      "step": 10305
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.071332072412862e-07,
+      "loss": 1.91,
+      "step": 10310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.064577141313159e-07,
+      "loss": 2.0866,
+      "step": 10315
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.057822210213456e-07,
+      "loss": 2.0044,
+      "step": 10320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.051067279113752e-07,
+      "loss": 2.1199,
+      "step": 10325
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.04431234801405e-07,
+      "loss": 1.996,
+      "step": 10330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.037557416914347e-07,
+      "loss": 1.9303,
+      "step": 10335
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.030802485814644e-07,
+      "loss": 1.9297,
+      "step": 10340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.024047554714942e-07,
+      "loss": 1.9069,
+      "step": 10345
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.017292623615239e-07,
+      "loss": 2.0062,
+      "step": 10350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.010537692515536e-07,
+      "loss": 2.0051,
+      "step": 10355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.003782761415833e-07,
+      "loss": 1.9996,
+      "step": 10360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.997027830316131e-07,
+      "loss": 1.9947,
+      "step": 10365
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.990272899216428e-07,
+      "loss": 1.984,
+      "step": 10370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.983517968116725e-07,
+      "loss": 1.9509,
+      "step": 10375
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.976763037017022e-07,
+      "loss": 1.9455,
+      "step": 10380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.970008105917319e-07,
+      "loss": 1.9383,
+      "step": 10385
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.963253174817616e-07,
+      "loss": 1.8293,
+      "step": 10390
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.956498243717913e-07,
+      "loss": 1.9228,
+      "step": 10395
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.94974331261821e-07,
+      "loss": 2.0348,
+      "step": 10400
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 1.9586541652679443,
+      "eval_runtime": 165.5067,
+      "eval_samples_per_second": 3.601,
+      "eval_steps_per_second": 0.453,
+      "step": 10400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.942988381518508e-07,
+      "loss": 1.9632,
+      "step": 10405
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.936233450418805e-07,
+      "loss": 1.8818,
+      "step": 10410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.929478519319102e-07,
+      "loss": 1.9375,
+      "step": 10415
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9227235882194e-07,
+      "loss": 1.9466,
+      "step": 10420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.915968657119698e-07,
+      "loss": 2.0166,
+      "step": 10425
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.909213726019995e-07,
+      "loss": 1.9846,
+      "step": 10430
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.902458794920292e-07,
+      "loss": 1.9175,
+      "step": 10435
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.89570386382059e-07,
+      "loss": 1.8928,
+      "step": 10440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.888948932720887e-07,
+      "loss": 1.9582,
+      "step": 10445
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.882194001621184e-07,
+      "loss": 1.9586,
+      "step": 10450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.87543907052148e-07,
+      "loss": 1.9719,
+      "step": 10455
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.868684139421778e-07,
+      "loss": 1.9996,
+      "step": 10460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.861929208322075e-07,
+      "loss": 1.935,
+      "step": 10465
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.855174277222372e-07,
+      "loss": 1.9258,
+      "step": 10470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.848419346122669e-07,
+      "loss": 2.0089,
+      "step": 10475
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.841664415022967e-07,
+      "loss": 1.9161,
+      "step": 10480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.834909483923264e-07,
+      "loss": 1.9469,
+      "step": 10485
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.828154552823561e-07,
+      "loss": 1.9688,
+      "step": 10490
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.821399621723858e-07,
+      "loss": 1.9424,
+      "step": 10495
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.814644690624156e-07,
+      "loss": 1.9893,
+      "step": 10500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.807889759524453e-07,
+      "loss": 1.9072,
+      "step": 10505
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.80113482842475e-07,
+      "loss": 2.0403,
+      "step": 10510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.794379897325046e-07,
+      "loss": 1.9779,
+      "step": 10515
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.787624966225344e-07,
+      "loss": 1.9289,
+      "step": 10520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.780870035125641e-07,
+      "loss": 2.047,
+      "step": 10525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.774115104025938e-07,
+      "loss": 2.0729,
+      "step": 10530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.767360172926235e-07,
+      "loss": 1.9825,
+      "step": 10535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.760605241826533e-07,
+      "loss": 1.9138,
+      "step": 10540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.75385031072683e-07,
+      "loss": 1.9638,
+      "step": 10545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.747095379627127e-07,
+      "loss": 2.0517,
+      "step": 10550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.740340448527424e-07,
+      "loss": 2.0771,
+      "step": 10555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.733585517427722e-07,
+      "loss": 1.9698,
+      "step": 10560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.72683058632802e-07,
+      "loss": 1.9299,
+      "step": 10565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.720075655228317e-07,
+      "loss": 2.0264,
+      "step": 10570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.713320724128615e-07,
+      "loss": 2.0212,
+      "step": 10575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.706565793028912e-07,
+      "loss": 1.9875,
+      "step": 10580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.699810861929208e-07,
+      "loss": 1.9172,
+      "step": 10585
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.693055930829505e-07,
+      "loss": 2.0435,
+      "step": 10590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.686300999729803e-07,
+      "loss": 2.0022,
+      "step": 10595
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.6795460686301e-07,
+      "loss": 2.0239,
+      "step": 10600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.672791137530397e-07,
+      "loss": 1.9575,
+      "step": 10605
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.666036206430694e-07,
+      "loss": 2.0316,
+      "step": 10610
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.659281275330992e-07,
+      "loss": 1.9751,
+      "step": 10615
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.652526344231289e-07,
+      "loss": 1.9895,
+      "step": 10620
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.645771413131586e-07,
+      "loss": 1.9672,
+      "step": 10625
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.639016482031883e-07,
+      "loss": 1.9807,
+      "step": 10630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.632261550932181e-07,
+      "loss": 1.9333,
+      "step": 10635
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.625506619832478e-07,
+      "loss": 1.835,
+      "step": 10640
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.618751688732774e-07,
+      "loss": 1.9753,
+      "step": 10645
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.611996757633071e-07,
+      "loss": 1.9376,
+      "step": 10650
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.605241826533369e-07,
+      "loss": 1.9732,
+      "step": 10655
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.598486895433666e-07,
+      "loss": 2.0749,
+      "step": 10660
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.591731964333963e-07,
+      "loss": 1.934,
+      "step": 10665
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.58497703323426e-07,
+      "loss": 1.9208,
+      "step": 10670
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.578222102134558e-07,
+      "loss": 1.9562,
+      "step": 10675
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.571467171034855e-07,
+      "loss": 2.082,
+      "step": 10680
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.564712239935152e-07,
+      "loss": 2.0432,
+      "step": 10685
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.557957308835449e-07,
+      "loss": 1.9236,
+      "step": 10690
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.551202377735747e-07,
+      "loss": 2.0632,
+      "step": 10695
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.544447446636044e-07,
+      "loss": 1.9083,
+      "step": 10700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.537692515536341e-07,
+      "loss": 2.0383,
+      "step": 10705
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.530937584436639e-07,
+      "loss": 1.9063,
+      "step": 10710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.524182653336936e-07,
+      "loss": 2.0845,
+      "step": 10715
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.517427722237233e-07,
+      "loss": 1.9743,
+      "step": 10720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.51067279113753e-07,
+      "loss": 2.0553,
+      "step": 10725
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.503917860037828e-07,
+      "loss": 2.0907,
+      "step": 10730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.497162928938125e-07,
+      "loss": 1.9165,
+      "step": 10735
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.490407997838422e-07,
+      "loss": 1.9457,
+      "step": 10740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.483653066738719e-07,
+      "loss": 1.9602,
+      "step": 10745
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.476898135639017e-07,
+      "loss": 1.9842,
+      "step": 10750
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.470143204539314e-07,
+      "loss": 1.9074,
+      "step": 10755
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.463388273439611e-07,
+      "loss": 1.9034,
+      "step": 10760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.456633342339908e-07,
+      "loss": 1.966,
+      "step": 10765
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.449878411240206e-07,
+      "loss": 2.0049,
+      "step": 10770
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.443123480140502e-07,
+      "loss": 1.9493,
+      "step": 10775
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.436368549040799e-07,
+      "loss": 1.9264,
+      "step": 10780
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.429613617941096e-07,
+      "loss": 1.9516,
+      "step": 10785
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.422858686841394e-07,
+      "loss": 2.0036,
+      "step": 10790
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.416103755741691e-07,
+      "loss": 1.9817,
+      "step": 10795
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.409348824641988e-07,
+      "loss": 1.9724,
+      "step": 10800
+    },
+    {
+      "epoch": 1.46,
+      "eval_loss": 1.9568144083023071,
+      "eval_runtime": 165.314,
+      "eval_samples_per_second": 3.605,
+      "eval_steps_per_second": 0.454,
+      "step": 10800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.402593893542285e-07,
+      "loss": 2.0044,
+      "step": 10805
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.395838962442583e-07,
+      "loss": 1.9504,
+      "step": 10810
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.38908403134288e-07,
+      "loss": 1.9536,
+      "step": 10815
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.382329100243177e-07,
+      "loss": 1.9608,
+      "step": 10820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.375574169143475e-07,
+      "loss": 2.0382,
+      "step": 10825
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.368819238043772e-07,
+      "loss": 1.9799,
+      "step": 10830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.362064306944069e-07,
+      "loss": 1.9551,
+      "step": 10835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.355309375844365e-07,
+      "loss": 1.9896,
+      "step": 10840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.348554444744662e-07,
+      "loss": 2.039,
+      "step": 10845
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.34179951364496e-07,
+      "loss": 1.9283,
+      "step": 10850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.335044582545257e-07,
+      "loss": 1.924,
+      "step": 10855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.328289651445555e-07,
+      "loss": 2.0038,
+      "step": 10860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.321534720345853e-07,
+      "loss": 1.9953,
+      "step": 10865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.31477978924615e-07,
+      "loss": 2.0623,
+      "step": 10870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.308024858146447e-07,
+      "loss": 2.0246,
+      "step": 10875
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.301269927046744e-07,
+      "loss": 1.9648,
+      "step": 10880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.294514995947042e-07,
+      "loss": 1.9202,
+      "step": 10885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.287760064847339e-07,
+      "loss": 2.0132,
+      "step": 10890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.281005133747636e-07,
+      "loss": 2.0106,
+      "step": 10895
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.274250202647933e-07,
+      "loss": 2.006,
+      "step": 10900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.26749527154823e-07,
+      "loss": 1.9925,
+      "step": 10905
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.260740340448527e-07,
+      "loss": 2.018,
+      "step": 10910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.253985409348824e-07,
+      "loss": 1.9322,
+      "step": 10915
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.247230478249121e-07,
+      "loss": 1.9453,
+      "step": 10920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.240475547149419e-07,
+      "loss": 1.965,
+      "step": 10925
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.233720616049716e-07,
+      "loss": 1.9446,
+      "step": 10930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.226965684950013e-07,
+      "loss": 1.9942,
+      "step": 10935
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.22021075385031e-07,
+      "loss": 1.9138,
+      "step": 10940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.213455822750608e-07,
+      "loss": 1.9764,
+      "step": 10945
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.206700891650905e-07,
+      "loss": 1.9884,
+      "step": 10950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.199945960551202e-07,
+      "loss": 1.9135,
+      "step": 10955
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.1931910294515e-07,
+      "loss": 2.0193,
+      "step": 10960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.186436098351797e-07,
+      "loss": 1.9446,
+      "step": 10965
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.179681167252093e-07,
+      "loss": 2.0596,
+      "step": 10970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.17292623615239e-07,
+      "loss": 1.9912,
+      "step": 10975
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.166171305052688e-07,
+      "loss": 1.9624,
+      "step": 10980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.159416373952985e-07,
+      "loss": 1.9297,
+      "step": 10985
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.152661442853282e-07,
+      "loss": 1.9808,
+      "step": 10990
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.145906511753579e-07,
+      "loss": 1.9933,
+      "step": 10995
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.139151580653877e-07,
+      "loss": 1.9767,
+      "step": 11000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.132396649554175e-07,
+      "loss": 1.9229,
+      "step": 11005
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.125641718454472e-07,
+      "loss": 1.9596,
+      "step": 11010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.118886787354769e-07,
+      "loss": 1.9631,
+      "step": 11015
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.112131856255067e-07,
+      "loss": 1.9485,
+      "step": 11020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.105376925155364e-07,
+      "loss": 1.9808,
+      "step": 11025
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.098621994055661e-07,
+      "loss": 1.925,
+      "step": 11030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.091867062955957e-07,
+      "loss": 1.9282,
+      "step": 11035
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.085112131856255e-07,
+      "loss": 1.9395,
+      "step": 11040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.078357200756552e-07,
+      "loss": 1.9219,
+      "step": 11045
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.071602269656849e-07,
+      "loss": 1.8854,
+      "step": 11050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.064847338557146e-07,
+      "loss": 1.9981,
+      "step": 11055
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.058092407457444e-07,
+      "loss": 1.9279,
+      "step": 11060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.051337476357741e-07,
+      "loss": 2.0484,
+      "step": 11065
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.044582545258038e-07,
+      "loss": 1.971,
+      "step": 11070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.037827614158336e-07,
+      "loss": 1.914,
+      "step": 11075
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.031072683058633e-07,
+      "loss": 1.9193,
+      "step": 11080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.02431775195893e-07,
+      "loss": 1.9318,
+      "step": 11085
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.017562820859227e-07,
+      "loss": 1.9414,
+      "step": 11090
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.010807889759525e-07,
+      "loss": 1.9664,
+      "step": 11095
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.004052958659821e-07,
+      "loss": 2.0118,
+      "step": 11100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.997298027560118e-07,
+      "loss": 1.9932,
+      "step": 11105
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.990543096460415e-07,
+      "loss": 1.9015,
+      "step": 11110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.983788165360714e-07,
+      "loss": 1.9539,
+      "step": 11115
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.977033234261011e-07,
+      "loss": 2.1365,
+      "step": 11120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.970278303161308e-07,
+      "loss": 1.966,
+      "step": 11125
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.963523372061604e-07,
+      "loss": 2.0074,
+      "step": 11130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.956768440961902e-07,
+      "loss": 2.0762,
+      "step": 11135
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.950013509862199e-07,
+      "loss": 1.8631,
+      "step": 11140
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.943258578762496e-07,
+      "loss": 1.9708,
+      "step": 11145
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.936503647662793e-07,
+      "loss": 1.9905,
+      "step": 11150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.929748716563091e-07,
+      "loss": 1.9543,
+      "step": 11155
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.922993785463388e-07,
+      "loss": 1.9418,
+      "step": 11160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.916238854363685e-07,
+      "loss": 1.9517,
+      "step": 11165
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.909483923263982e-07,
+      "loss": 1.8429,
+      "step": 11170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.90272899216428e-07,
+      "loss": 1.9754,
+      "step": 11175
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.895974061064577e-07,
+      "loss": 1.8974,
+      "step": 11180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.889219129964874e-07,
+      "loss": 1.9996,
+      "step": 11185
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.882464198865172e-07,
+      "loss": 2.0173,
+      "step": 11190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.875709267765469e-07,
+      "loss": 1.9683,
+      "step": 11195
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.868954336665766e-07,
+      "loss": 1.9497,
+      "step": 11200
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 1.955120325088501,
+      "eval_runtime": 165.2289,
+      "eval_samples_per_second": 3.607,
+      "eval_steps_per_second": 0.454,
+      "step": 11200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.862199405566063e-07,
+      "loss": 1.9811,
+      "step": 11205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.855444474466361e-07,
+      "loss": 1.9879,
+      "step": 11210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.848689543366658e-07,
+      "loss": 1.9957,
+      "step": 11215
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.841934612266955e-07,
+      "loss": 1.8997,
+      "step": 11220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.835179681167252e-07,
+      "loss": 1.9366,
+      "step": 11225
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.828424750067549e-07,
+      "loss": 1.9237,
+      "step": 11230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.821669818967846e-07,
+      "loss": 1.9348,
+      "step": 11235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.814914887868143e-07,
+      "loss": 1.8665,
+      "step": 11240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.80815995676844e-07,
+      "loss": 1.9615,
+      "step": 11245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.801405025668738e-07,
+      "loss": 2.0397,
+      "step": 11250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.794650094569035e-07,
+      "loss": 2.0333,
+      "step": 11255
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.787895163469332e-07,
+      "loss": 1.996,
+      "step": 11260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.781140232369629e-07,
+      "loss": 1.971,
+      "step": 11265
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.774385301269927e-07,
+      "loss": 2.0498,
+      "step": 11270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.767630370170224e-07,
+      "loss": 2.0136,
+      "step": 11275
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.760875439070521e-07,
+      "loss": 1.9929,
+      "step": 11280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7541205079708184e-07,
+      "loss": 1.9976,
+      "step": 11285
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7473655768711157e-07,
+      "loss": 1.8972,
+      "step": 11290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.740610645771413e-07,
+      "loss": 2.0062,
+      "step": 11295
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.73385571467171e-07,
+      "loss": 2.0605,
+      "step": 11300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7271007835720075e-07,
+      "loss": 1.9654,
+      "step": 11305
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.720345852472304e-07,
+      "loss": 2.0147,
+      "step": 11310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7135909213726015e-07,
+      "loss": 1.9944,
+      "step": 11315
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.706835990272899e-07,
+      "loss": 1.9651,
+      "step": 11320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.700081059173196e-07,
+      "loss": 1.9287,
+      "step": 11325
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.693326128073494e-07,
+      "loss": 2.0328,
+      "step": 11330
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.686571196973791e-07,
+      "loss": 2.0085,
+      "step": 11335
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.679816265874088e-07,
+      "loss": 1.969,
+      "step": 11340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.673061334774385e-07,
+      "loss": 1.8905,
+      "step": 11345
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.6663064036746823e-07,
+      "loss": 1.9451,
+      "step": 11350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.6595514725749796e-07,
+      "loss": 1.9741,
+      "step": 11355
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.652796541475277e-07,
+      "loss": 1.9929,
+      "step": 11360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.646041610375574e-07,
+      "loss": 1.9585,
+      "step": 11365
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6392866792758714e-07,
+      "loss": 1.9383,
+      "step": 11370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.632531748176168e-07,
+      "loss": 1.9451,
+      "step": 11375
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6257768170764654e-07,
+      "loss": 1.9353,
+      "step": 11380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6190218859767626e-07,
+      "loss": 1.9131,
+      "step": 11385
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.61226695487706e-07,
+      "loss": 2.0578,
+      "step": 11390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.605512023777357e-07,
+      "loss": 1.9363,
+      "step": 11395
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5987570926776544e-07,
+      "loss": 1.9979,
+      "step": 11400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5920021615779517e-07,
+      "loss": 1.9738,
+      "step": 11405
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.585247230478249e-07,
+      "loss": 1.9685,
+      "step": 11410
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.578492299378546e-07,
+      "loss": 1.9402,
+      "step": 11415
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5717373682788435e-07,
+      "loss": 1.9906,
+      "step": 11420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.564982437179141e-07,
+      "loss": 1.8393,
+      "step": 11425
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.558227506079438e-07,
+      "loss": 1.988,
+      "step": 11430
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5514725749797353e-07,
+      "loss": 1.9981,
+      "step": 11435
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.544717643880032e-07,
+      "loss": 1.9823,
+      "step": 11440
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.537962712780329e-07,
+      "loss": 1.9773,
+      "step": 11445
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5312077816806265e-07,
+      "loss": 1.9343,
+      "step": 11450
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.524452850580924e-07,
+      "loss": 1.949,
+      "step": 11455
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.517697919481221e-07,
+      "loss": 1.9503,
+      "step": 11460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5109429883815183e-07,
+      "loss": 1.9878,
+      "step": 11465
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.504188057281815e-07,
+      "loss": 2.0098,
+      "step": 11470
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.497433126182113e-07,
+      "loss": 1.9646,
+      "step": 11475
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.49067819508241e-07,
+      "loss": 2.0194,
+      "step": 11480
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.4839232639827074e-07,
+      "loss": 1.9737,
+      "step": 11485
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.4771683328830046e-07,
+      "loss": 1.9999,
+      "step": 11490
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.470413401783302e-07,
+      "loss": 1.9922,
+      "step": 11495
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.463658470683599e-07,
+      "loss": 1.9929,
+      "step": 11500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.456903539583896e-07,
+      "loss": 2.0408,
+      "step": 11505
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.450148608484193e-07,
+      "loss": 1.9746,
+      "step": 11510
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4433936773844904e-07,
+      "loss": 2.0071,
+      "step": 11515
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4366387462847877e-07,
+      "loss": 1.9578,
+      "step": 11520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.429883815185085e-07,
+      "loss": 1.9269,
+      "step": 11525
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.423128884085382e-07,
+      "loss": 1.9608,
+      "step": 11530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.416373952985679e-07,
+      "loss": 1.9224,
+      "step": 11535
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.409619021885976e-07,
+      "loss": 1.9858,
+      "step": 11540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4028640907862735e-07,
+      "loss": 1.9847,
+      "step": 11545
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.396109159686571e-07,
+      "loss": 2.0205,
+      "step": 11550
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.3893542285868685e-07,
+      "loss": 1.9723,
+      "step": 11555
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.382599297487166e-07,
+      "loss": 2.043,
+      "step": 11560
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.375844366387463e-07,
+      "loss": 1.9467,
+      "step": 11565
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.36908943528776e-07,
+      "loss": 1.9304,
+      "step": 11570
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.362334504188057e-07,
+      "loss": 2.0435,
+      "step": 11575
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.3555795730883543e-07,
+      "loss": 1.9946,
+      "step": 11580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3488246419886516e-07,
+      "loss": 1.9891,
+      "step": 11585
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.342069710888949e-07,
+      "loss": 1.9875,
+      "step": 11590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.335314779789246e-07,
+      "loss": 2.0281,
+      "step": 11595
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.328559848689543e-07,
+      "loss": 1.8857,
+      "step": 11600
+    },
+    {
+      "epoch": 1.57,
+      "eval_loss": 1.9535282850265503,
+      "eval_runtime": 165.4539,
+      "eval_samples_per_second": 3.602,
+      "eval_steps_per_second": 0.453,
+      "step": 11600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.32180491758984e-07,
+      "loss": 1.9551,
+      "step": 11605
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3150499864901373e-07,
+      "loss": 1.9194,
+      "step": 11610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3082950553904346e-07,
+      "loss": 1.9471,
+      "step": 11615
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.301540124290732e-07,
+      "loss": 1.9033,
+      "step": 11620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2947851931910297e-07,
+      "loss": 1.9445,
+      "step": 11625
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.288030262091327e-07,
+      "loss": 1.9713,
+      "step": 11630
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2812753309916237e-07,
+      "loss": 1.9252,
+      "step": 11635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.274520399891921e-07,
+      "loss": 1.88,
+      "step": 11640
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.267765468792218e-07,
+      "loss": 1.9462,
+      "step": 11645
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2610105376925154e-07,
+      "loss": 2.0004,
+      "step": 11650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2542556065928127e-07,
+      "loss": 1.9346,
+      "step": 11655
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.24750067549311e-07,
+      "loss": 2.0424,
+      "step": 11660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2407457443934067e-07,
+      "loss": 1.966,
+      "step": 11665
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.233990813293704e-07,
+      "loss": 2.016,
+      "step": 11670
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.227235882194001e-07,
+      "loss": 1.9669,
+      "step": 11675
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2204809510942985e-07,
+      "loss": 1.9423,
+      "step": 11680
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213726019994596e-07,
+      "loss": 1.9125,
+      "step": 11685
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.206971088894893e-07,
+      "loss": 1.8883,
+      "step": 11690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.200216157795191e-07,
+      "loss": 1.9714,
+      "step": 11695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1934612266954875e-07,
+      "loss": 1.9444,
+      "step": 11700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.186706295595785e-07,
+      "loss": 1.9369,
+      "step": 11705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.179951364496082e-07,
+      "loss": 1.9787,
+      "step": 11710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1731964333963793e-07,
+      "loss": 1.8535,
+      "step": 11715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1664415022966766e-07,
+      "loss": 1.9828,
+      "step": 11720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.159686571196974e-07,
+      "loss": 1.9836,
+      "step": 11725
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1529316400972706e-07,
+      "loss": 1.9408,
+      "step": 11730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.146176708997568e-07,
+      "loss": 1.9519,
+      "step": 11735
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.139421777897865e-07,
+      "loss": 2.0592,
+      "step": 11740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1326668467981624e-07,
+      "loss": 1.9832,
+      "step": 11745
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1259119156984596e-07,
+      "loss": 1.9843,
+      "step": 11750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.119156984598757e-07,
+      "loss": 1.947,
+      "step": 11755
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1124020534990536e-07,
+      "loss": 1.9285,
+      "step": 11760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.105647122399351e-07,
+      "loss": 1.8873,
+      "step": 11765
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0988921912996487e-07,
+      "loss": 1.9133,
+      "step": 11770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.092137260199946e-07,
+      "loss": 2.0237,
+      "step": 11775
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.085382329100243e-07,
+      "loss": 1.9161,
+      "step": 11780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0786273980005405e-07,
+      "loss": 1.8488,
+      "step": 11785
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.071872466900838e-07,
+      "loss": 2.0393,
+      "step": 11790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0651175358011345e-07,
+      "loss": 1.9536,
+      "step": 11795
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.058362604701432e-07,
+      "loss": 1.97,
+      "step": 11800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.051607673601729e-07,
+      "loss": 2.003,
+      "step": 11805
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0448527425020263e-07,
+      "loss": 1.9425,
+      "step": 11810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0380978114023235e-07,
+      "loss": 1.9305,
+      "step": 11815
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.031342880302621e-07,
+      "loss": 1.8838,
+      "step": 11820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0245879492029175e-07,
+      "loss": 2.0721,
+      "step": 11825
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.017833018103215e-07,
+      "loss": 1.8947,
+      "step": 11830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.011078087003512e-07,
+      "loss": 1.9371,
+      "step": 11835
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.00432315590381e-07,
+      "loss": 1.9521,
+      "step": 11840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.997568224804107e-07,
+      "loss": 1.9498,
+      "step": 11845
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9908132937044044e-07,
+      "loss": 1.9536,
+      "step": 11850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9840583626047016e-07,
+      "loss": 1.9916,
+      "step": 11855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9773034315049984e-07,
+      "loss": 1.9506,
+      "step": 11860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9705485004052956e-07,
+      "loss": 1.9623,
+      "step": 11865
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.963793569305593e-07,
+      "loss": 1.9731,
+      "step": 11870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.95703863820589e-07,
+      "loss": 1.9713,
+      "step": 11875
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9502837071061874e-07,
+      "loss": 2.0692,
+      "step": 11880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9435287760064847e-07,
+      "loss": 1.9771,
+      "step": 11885
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9367738449067814e-07,
+      "loss": 2.0068,
+      "step": 11890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9300189138070787e-07,
+      "loss": 1.8429,
+      "step": 11895
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.923263982707376e-07,
+      "loss": 2.0443,
+      "step": 11900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.916509051607673e-07,
+      "loss": 1.9551,
+      "step": 11905
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9097541205079705e-07,
+      "loss": 1.8964,
+      "step": 11910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.902999189408268e-07,
+      "loss": 1.9391,
+      "step": 11915
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8962442583085655e-07,
+      "loss": 2.0496,
+      "step": 11920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.889489327208862e-07,
+      "loss": 2.0249,
+      "step": 11925
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8827343961091595e-07,
+      "loss": 2.0152,
+      "step": 11930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.875979465009457e-07,
+      "loss": 1.9238,
+      "step": 11935
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.869224533909754e-07,
+      "loss": 2.0196,
+      "step": 11940
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8624696028100513e-07,
+      "loss": 1.9307,
+      "step": 11945
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8557146717103486e-07,
+      "loss": 1.9859,
+      "step": 11950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8489597406106453e-07,
+      "loss": 1.9542,
+      "step": 11955
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8422048095109426e-07,
+      "loss": 1.8884,
+      "step": 11960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.83544987841124e-07,
+      "loss": 2.0328,
+      "step": 11965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.828694947311537e-07,
+      "loss": 1.9707,
+      "step": 11970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8219400162118344e-07,
+      "loss": 1.9789,
+      "step": 11975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8151850851121316e-07,
+      "loss": 1.964,
+      "step": 11980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8084301540124294e-07,
+      "loss": 1.9321,
+      "step": 11985
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.801675222912726e-07,
+      "loss": 1.916,
+      "step": 11990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7949202918130234e-07,
+      "loss": 1.8828,
+      "step": 11995
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7881653607133207e-07,
+      "loss": 1.9949,
+      "step": 12000
+    },
+    {
+      "epoch": 1.62,
+      "eval_loss": 1.952235221862793,
+      "eval_runtime": 165.3673,
+      "eval_samples_per_second": 3.604,
+      "eval_steps_per_second": 0.454,
+      "step": 12000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.781410429613618e-07,
+      "loss": 1.8852,
+      "step": 12005
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.774655498513915e-07,
+      "loss": 1.9413,
+      "step": 12010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7679005674142125e-07,
+      "loss": 2.0267,
+      "step": 12015
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.761145636314509e-07,
+      "loss": 1.8799,
+      "step": 12020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7543907052148065e-07,
+      "loss": 1.9891,
+      "step": 12025
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7476357741151037e-07,
+      "loss": 1.9889,
+      "step": 12030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.740880843015401e-07,
+      "loss": 1.984,
+      "step": 12035
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.734125911915698e-07,
+      "loss": 1.8569,
+      "step": 12040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7273709808159955e-07,
+      "loss": 1.9308,
+      "step": 12045
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.720616049716293e-07,
+      "loss": 1.9315,
+      "step": 12050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7138611186165895e-07,
+      "loss": 1.9194,
+      "step": 12055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7071061875168873e-07,
+      "loss": 1.9769,
+      "step": 12060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7003512564171846e-07,
+      "loss": 1.9433,
+      "step": 12065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.693596325317482e-07,
+      "loss": 1.8922,
+      "step": 12070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.686841394217779e-07,
+      "loss": 2.0268,
+      "step": 12075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6800864631180763e-07,
+      "loss": 1.9862,
+      "step": 12080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.673331532018373e-07,
+      "loss": 2.0022,
+      "step": 12085
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6665766009186703e-07,
+      "loss": 1.9151,
+      "step": 12090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6598216698189676e-07,
+      "loss": 2.0128,
+      "step": 12095
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.653066738719265e-07,
+      "loss": 1.9876,
+      "step": 12100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.646311807619562e-07,
+      "loss": 2.0628,
+      "step": 12105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6395568765198594e-07,
+      "loss": 1.9863,
+      "step": 12110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6328019454201567e-07,
+      "loss": 1.9916,
+      "step": 12115
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6260470143204534e-07,
+      "loss": 2.0298,
+      "step": 12120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6192920832207506e-07,
+      "loss": 2.0356,
+      "step": 12125
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.612537152121048e-07,
+      "loss": 1.927,
+      "step": 12130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6057822210213457e-07,
+      "loss": 1.9696,
+      "step": 12135
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.599027289921643e-07,
+      "loss": 2.0689,
+      "step": 12140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.59227235882194e-07,
+      "loss": 1.9127,
+      "step": 12145
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.585517427722237e-07,
+      "loss": 1.9824,
+      "step": 12150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.578762496622534e-07,
+      "loss": 2.0156,
+      "step": 12155
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5720075655228315e-07,
+      "loss": 1.9376,
+      "step": 12160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.565252634423129e-07,
+      "loss": 1.8839,
+      "step": 12165
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.558497703323426e-07,
+      "loss": 1.986,
+      "step": 12170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5517427722237233e-07,
+      "loss": 1.9391,
+      "step": 12175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5449878411240205e-07,
+      "loss": 1.9136,
+      "step": 12180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5382329100243173e-07,
+      "loss": 1.9457,
+      "step": 12185
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5314779789246145e-07,
+      "loss": 2.0062,
+      "step": 12190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.524723047824912e-07,
+      "loss": 1.9796,
+      "step": 12195
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.517968116725209e-07,
+      "loss": 2.0649,
+      "step": 12200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.511213185625507e-07,
+      "loss": 2.0202,
+      "step": 12205
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.504458254525804e-07,
+      "loss": 1.9865,
+      "step": 12210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.497703323426101e-07,
+      "loss": 1.9317,
+      "step": 12215
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.490948392326398e-07,
+      "loss": 1.9266,
+      "step": 12220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4841934612266954e-07,
+      "loss": 1.9678,
+      "step": 12225
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4774385301269926e-07,
+      "loss": 2.0067,
+      "step": 12230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.47068359902729e-07,
+      "loss": 1.9964,
+      "step": 12235
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.463928667927587e-07,
+      "loss": 2.0902,
+      "step": 12240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.457173736827884e-07,
+      "loss": 1.9893,
+      "step": 12245
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.450418805728181e-07,
+      "loss": 1.9439,
+      "step": 12250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4436638746284784e-07,
+      "loss": 1.9843,
+      "step": 12255
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4369089435287757e-07,
+      "loss": 2.0811,
+      "step": 12260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.430154012429073e-07,
+      "loss": 1.8165,
+      "step": 12265
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.42339908132937e-07,
+      "loss": 2.0074,
+      "step": 12270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4166441502296675e-07,
+      "loss": 1.8722,
+      "step": 12275
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4098892191299647e-07,
+      "loss": 1.9534,
+      "step": 12280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.403134288030262e-07,
+      "loss": 2.0328,
+      "step": 12285
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.396379356930559e-07,
+      "loss": 2.0205,
+      "step": 12290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3896244258308565e-07,
+      "loss": 1.9499,
+      "step": 12295
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.382869494731154e-07,
+      "loss": 1.9113,
+      "step": 12300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.376114563631451e-07,
+      "loss": 1.9177,
+      "step": 12305
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.369359632531748e-07,
+      "loss": 1.8988,
+      "step": 12310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.362604701432045e-07,
+      "loss": 1.9675,
+      "step": 12315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3558497703323423e-07,
+      "loss": 1.9732,
+      "step": 12320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3490948392326396e-07,
+      "loss": 1.987,
+      "step": 12325
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.342339908132937e-07,
+      "loss": 1.9473,
+      "step": 12330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.335584977033234e-07,
+      "loss": 1.9229,
+      "step": 12335
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3288300459335314e-07,
+      "loss": 1.9963,
+      "step": 12340
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.322075114833828e-07,
+      "loss": 1.8348,
+      "step": 12345
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.315320183734126e-07,
+      "loss": 1.9457,
+      "step": 12350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.308565252634423e-07,
+      "loss": 1.9232,
+      "step": 12355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3018103215347204e-07,
+      "loss": 1.9743,
+      "step": 12360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2950553904350177e-07,
+      "loss": 1.9761,
+      "step": 12365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.288300459335315e-07,
+      "loss": 1.9756,
+      "step": 12370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2815455282356117e-07,
+      "loss": 1.8987,
+      "step": 12375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.274790597135909e-07,
+      "loss": 2.0811,
+      "step": 12380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.268035666036206e-07,
+      "loss": 1.9758,
+      "step": 12385
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2612807349365035e-07,
+      "loss": 1.9949,
+      "step": 12390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2545258038368007e-07,
+      "loss": 1.8838,
+      "step": 12395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.247770872737098e-07,
+      "loss": 1.9292,
+      "step": 12400
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 1.9511464834213257,
+      "eval_runtime": 165.4983,
+      "eval_samples_per_second": 3.601,
+      "eval_steps_per_second": 0.453,
+      "step": 12400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.241015941637395e-07,
+      "loss": 1.9073,
+      "step": 12405
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.234261010537692e-07,
+      "loss": 1.9725,
+      "step": 12410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.227506079437989e-07,
+      "loss": 1.8673,
+      "step": 12415
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2207511483382865e-07,
+      "loss": 2.0299,
+      "step": 12420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2139962172385843e-07,
+      "loss": 1.9586,
+      "step": 12425
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2072412861388816e-07,
+      "loss": 1.9822,
+      "step": 12430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.200486355039179e-07,
+      "loss": 2.0641,
+      "step": 12435
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1937314239394756e-07,
+      "loss": 2.0372,
+      "step": 12440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.186976492839773e-07,
+      "loss": 1.9135,
+      "step": 12445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.18022156174007e-07,
+      "loss": 1.9977,
+      "step": 12450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1734666306403673e-07,
+      "loss": 1.9981,
+      "step": 12455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1667116995406646e-07,
+      "loss": 2.0354,
+      "step": 12460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.159956768440962e-07,
+      "loss": 1.9714,
+      "step": 12465
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.153201837341259e-07,
+      "loss": 2.0264,
+      "step": 12470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.146446906241556e-07,
+      "loss": 1.9805,
+      "step": 12475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.139691975141853e-07,
+      "loss": 2.0613,
+      "step": 12480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1329370440421504e-07,
+      "loss": 1.8951,
+      "step": 12485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1261821129424477e-07,
+      "loss": 1.9624,
+      "step": 12490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1194271818427454e-07,
+      "loss": 1.9934,
+      "step": 12495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1126722507430427e-07,
+      "loss": 1.9149,
+      "step": 12500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1059173196433394e-07,
+      "loss": 1.9528,
+      "step": 12505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0991623885436367e-07,
+      "loss": 1.9816,
+      "step": 12510
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.092407457443934e-07,
+      "loss": 1.9338,
+      "step": 12515
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.085652526344231e-07,
+      "loss": 1.9294,
+      "step": 12520
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0788975952445285e-07,
+      "loss": 1.9416,
+      "step": 12525
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.072142664144826e-07,
+      "loss": 2.0216,
+      "step": 12530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.065387733045123e-07,
+      "loss": 1.9174,
+      "step": 12535
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.05863280194542e-07,
+      "loss": 1.8646,
+      "step": 12540
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.051877870845717e-07,
+      "loss": 2.0483,
+      "step": 12545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0451229397460143e-07,
+      "loss": 1.9749,
+      "step": 12550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0383680086463115e-07,
+      "loss": 2.0014,
+      "step": 12555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.031613077546609e-07,
+      "loss": 1.9313,
+      "step": 12560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.024858146446906e-07,
+      "loss": 1.8786,
+      "step": 12565
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0181032153472033e-07,
+      "loss": 1.8782,
+      "step": 12570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0113482842475006e-07,
+      "loss": 1.9038,
+      "step": 12575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.004593353147798e-07,
+      "loss": 1.9865,
+      "step": 12580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.997838422048095e-07,
+      "loss": 1.9386,
+      "step": 12585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9910834909483924e-07,
+      "loss": 1.884,
+      "step": 12590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9843285598486896e-07,
+      "loss": 1.9116,
+      "step": 12595
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.977573628748987e-07,
+      "loss": 1.9599,
+      "step": 12600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9708186976492836e-07,
+      "loss": 1.966,
+      "step": 12605
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.964063766549581e-07,
+      "loss": 1.9358,
+      "step": 12610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.957308835449878e-07,
+      "loss": 2.0271,
+      "step": 12615
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9505539043501754e-07,
+      "loss": 1.9708,
+      "step": 12620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9437989732504727e-07,
+      "loss": 1.8144,
+      "step": 12625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.93704404215077e-07,
+      "loss": 2.0278,
+      "step": 12630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9302891110510667e-07,
+      "loss": 1.9605,
+      "step": 12635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.923534179951364e-07,
+      "loss": 1.9958,
+      "step": 12640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.916779248851662e-07,
+      "loss": 1.9269,
+      "step": 12645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.910024317751959e-07,
+      "loss": 1.9597,
+      "step": 12650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.903269386652256e-07,
+      "loss": 1.9553,
+      "step": 12655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8965144555525535e-07,
+      "loss": 1.9797,
+      "step": 12660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.889759524452851e-07,
+      "loss": 1.9698,
+      "step": 12665
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8830045933531475e-07,
+      "loss": 2.0515,
+      "step": 12670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.876249662253445e-07,
+      "loss": 1.9132,
+      "step": 12675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.869494731153742e-07,
+      "loss": 1.9046,
+      "step": 12680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8627398000540393e-07,
+      "loss": 1.9502,
+      "step": 12685
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8559848689543366e-07,
+      "loss": 2.0462,
+      "step": 12690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.849229937854634e-07,
+      "loss": 2.0225,
+      "step": 12695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8424750067549306e-07,
+      "loss": 1.9613,
+      "step": 12700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.835720075655228e-07,
+      "loss": 1.9458,
+      "step": 12705
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.828965144555525e-07,
+      "loss": 1.9629,
+      "step": 12710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.822210213455823e-07,
+      "loss": 1.9982,
+      "step": 12715
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.81545528235612e-07,
+      "loss": 1.9519,
+      "step": 12720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8087003512564174e-07,
+      "loss": 2.0149,
+      "step": 12725
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8019454201567147e-07,
+      "loss": 1.9178,
+      "step": 12730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7951904890570114e-07,
+      "loss": 1.9672,
+      "step": 12735
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7884355579573087e-07,
+      "loss": 1.9545,
+      "step": 12740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.781680626857606e-07,
+      "loss": 1.9199,
+      "step": 12745
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.774925695757903e-07,
+      "loss": 1.962,
+      "step": 12750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7681707646582005e-07,
+      "loss": 1.9126,
+      "step": 12755
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7614158335584977e-07,
+      "loss": 1.9452,
+      "step": 12760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7546609024587945e-07,
+      "loss": 2.0155,
+      "step": 12765
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7479059713590917e-07,
+      "loss": 1.9335,
+      "step": 12770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.741151040259389e-07,
+      "loss": 1.8992,
+      "step": 12775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.734396109159686e-07,
+      "loss": 1.9687,
+      "step": 12780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7276411780599835e-07,
+      "loss": 1.9148,
+      "step": 12785
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7208862469602813e-07,
+      "loss": 2.0001,
+      "step": 12790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7141313158605786e-07,
+      "loss": 1.9786,
+      "step": 12795
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7073763847608753e-07,
+      "loss": 1.937,
+      "step": 12800
+    },
+    {
+      "epoch": 1.73,
+      "eval_loss": 1.95003080368042,
+      "eval_runtime": 165.3799,
+      "eval_samples_per_second": 3.604,
+      "eval_steps_per_second": 0.454,
+      "step": 12800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7006214536611726e-07,
+      "loss": 1.9034,
+      "step": 12805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.69386652256147e-07,
+      "loss": 2.009,
+      "step": 12810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.687111591461767e-07,
+      "loss": 1.957,
+      "step": 12815
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6803566603620643e-07,
+      "loss": 2.0366,
+      "step": 12820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6736017292623616e-07,
+      "loss": 1.9522,
+      "step": 12825
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6668467981626583e-07,
+      "loss": 1.9014,
+      "step": 12830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6600918670629556e-07,
+      "loss": 1.8543,
+      "step": 12835
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.653336935963253e-07,
+      "loss": 1.9508,
+      "step": 12840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.64658200486355e-07,
+      "loss": 1.9476,
+      "step": 12845
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6398270737638474e-07,
+      "loss": 1.9962,
+      "step": 12850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6330721426641447e-07,
+      "loss": 1.8131,
+      "step": 12855
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6263172115644424e-07,
+      "loss": 1.8921,
+      "step": 12860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.619562280464739e-07,
+      "loss": 1.9959,
+      "step": 12865
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6128073493650364e-07,
+      "loss": 1.9768,
+      "step": 12870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6060524182653337e-07,
+      "loss": 1.9248,
+      "step": 12875
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.599297487165631e-07,
+      "loss": 1.9049,
+      "step": 12880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.592542556065928e-07,
+      "loss": 1.9095,
+      "step": 12885
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5857876249662255e-07,
+      "loss": 1.8965,
+      "step": 12890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.579032693866522e-07,
+      "loss": 1.9665,
+      "step": 12895
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5722777627668195e-07,
+      "loss": 1.9266,
+      "step": 12900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.565522831667117e-07,
+      "loss": 2.0833,
+      "step": 12905
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.558767900567414e-07,
+      "loss": 1.9587,
+      "step": 12910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5520129694677113e-07,
+      "loss": 2.1045,
+      "step": 12915
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5452580383680085e-07,
+      "loss": 2.0159,
+      "step": 12920
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5385031072683053e-07,
+      "loss": 1.9911,
+      "step": 12925
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5317481761686025e-07,
+      "loss": 1.9373,
+      "step": 12930
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5249932450689003e-07,
+      "loss": 2.012,
+      "step": 12935
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5182383139691976e-07,
+      "loss": 1.9334,
+      "step": 12940
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.511483382869495e-07,
+      "loss": 2.0098,
+      "step": 12945
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.504728451769792e-07,
+      "loss": 1.8779,
+      "step": 12950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4979735206700894e-07,
+      "loss": 1.8818,
+      "step": 12955
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.491218589570386e-07,
+      "loss": 1.9791,
+      "step": 12960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4844636584706834e-07,
+      "loss": 1.9753,
+      "step": 12965
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4777087273709806e-07,
+      "loss": 1.9361,
+      "step": 12970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.470953796271278e-07,
+      "loss": 1.8509,
+      "step": 12975
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.464198865171575e-07,
+      "loss": 2.0183,
+      "step": 12980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4574439340718724e-07,
+      "loss": 1.9777,
+      "step": 12985
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4506890029721697e-07,
+      "loss": 1.9786,
+      "step": 12990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.443934071872467e-07,
+      "loss": 1.9021,
+      "step": 12995
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.437179140772764e-07,
+      "loss": 2.0193,
+      "step": 13000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4304242096730615e-07,
+      "loss": 1.8966,
+      "step": 13005
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.423669278573358e-07,
+      "loss": 2.0073,
+      "step": 13010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4169143474736555e-07,
+      "loss": 1.9552,
+      "step": 13015
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.410159416373953e-07,
+      "loss": 1.9251,
+      "step": 13020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.40340448527425e-07,
+      "loss": 2.0429,
+      "step": 13025
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3966495541745473e-07,
+      "loss": 2.0433,
+      "step": 13030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3898946230748445e-07,
+      "loss": 1.9169,
+      "step": 13035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3831396919751418e-07,
+      "loss": 1.9662,
+      "step": 13040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.376384760875439e-07,
+      "loss": 1.9404,
+      "step": 13045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.369629829775736e-07,
+      "loss": 1.9799,
+      "step": 13050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3628748986760333e-07,
+      "loss": 1.8214,
+      "step": 13055
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3561199675763308e-07,
+      "loss": 1.9932,
+      "step": 13060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3493650364766278e-07,
+      "loss": 1.9733,
+      "step": 13065
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.342610105376925e-07,
+      "loss": 1.9566,
+      "step": 13070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3358551742772224e-07,
+      "loss": 1.897,
+      "step": 13075
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3291002431775194e-07,
+      "loss": 1.9646,
+      "step": 13080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3223453120778166e-07,
+      "loss": 2.0013,
+      "step": 13085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.315590380978114e-07,
+      "loss": 1.8838,
+      "step": 13090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.308835449878411e-07,
+      "loss": 2.0112,
+      "step": 13095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3020805187787084e-07,
+      "loss": 1.9287,
+      "step": 13100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2953255876790057e-07,
+      "loss": 2.0699,
+      "step": 13105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.288570656579303e-07,
+      "loss": 1.8843,
+      "step": 13110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2818157254796e-07,
+      "loss": 1.9018,
+      "step": 13115
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2750607943798972e-07,
+      "loss": 1.9611,
+      "step": 13120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2683058632801945e-07,
+      "loss": 1.9831,
+      "step": 13125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2615509321804915e-07,
+      "loss": 1.9069,
+      "step": 13130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.254796001080789e-07,
+      "loss": 1.9747,
+      "step": 13135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2480410699810863e-07,
+      "loss": 1.9792,
+      "step": 13140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2412861388813833e-07,
+      "loss": 2.0183,
+      "step": 13145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2345312077816805e-07,
+      "loss": 1.9591,
+      "step": 13150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2277762766819778e-07,
+      "loss": 1.9085,
+      "step": 13155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2210213455822748e-07,
+      "loss": 2.0675,
+      "step": 13160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.214266414482572e-07,
+      "loss": 1.9204,
+      "step": 13165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2075114833828696e-07,
+      "loss": 2.035,
+      "step": 13170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2007565522831668e-07,
+      "loss": 2.0147,
+      "step": 13175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1940016211834638e-07,
+      "loss": 1.917,
+      "step": 13180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.187246690083761e-07,
+      "loss": 1.9055,
+      "step": 13185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1804917589840584e-07,
+      "loss": 1.9916,
+      "step": 13190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1737368278843553e-07,
+      "loss": 1.9845,
+      "step": 13195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1669818967846526e-07,
+      "loss": 2.0573,
+      "step": 13200
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 1.9494786262512207,
+      "eval_runtime": 165.1665,
+      "eval_samples_per_second": 3.608,
+      "eval_steps_per_second": 0.454,
+      "step": 13200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1602269656849501e-07,
+      "loss": 2.0237,
+      "step": 13205
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1534720345852471e-07,
+      "loss": 1.9262,
+      "step": 13210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1467171034855444e-07,
+      "loss": 1.9819,
+      "step": 13215
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1399621723858417e-07,
+      "loss": 1.9537,
+      "step": 13220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1332072412861387e-07,
+      "loss": 1.8926,
+      "step": 13225
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.126452310186436e-07,
+      "loss": 1.9364,
+      "step": 13230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1196973790867332e-07,
+      "loss": 1.9544,
+      "step": 13235
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1129424479870302e-07,
+      "loss": 1.9288,
+      "step": 13240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1061875168873277e-07,
+      "loss": 1.9206,
+      "step": 13245
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.099432585787625e-07,
+      "loss": 1.9185,
+      "step": 13250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0926776546879222e-07,
+      "loss": 1.9306,
+      "step": 13255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0859227235882192e-07,
+      "loss": 1.9818,
+      "step": 13260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0791677924885165e-07,
+      "loss": 1.9057,
+      "step": 13265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0724128613888138e-07,
+      "loss": 1.9167,
+      "step": 13270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0656579302891108e-07,
+      "loss": 1.9288,
+      "step": 13275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0589029991894083e-07,
+      "loss": 2.0249,
+      "step": 13280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0521480680897055e-07,
+      "loss": 2.0134,
+      "step": 13285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0453931369900025e-07,
+      "loss": 2.0235,
+      "step": 13290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0386382058902998e-07,
+      "loss": 1.8421,
+      "step": 13295
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.031883274790597e-07,
+      "loss": 1.9947,
+      "step": 13300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.025128343690894e-07,
+      "loss": 2.0185,
+      "step": 13305
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0183734125911913e-07,
+      "loss": 1.955,
+      "step": 13310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0116184814914889e-07,
+      "loss": 2.025,
+      "step": 13315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.004863550391786e-07,
+      "loss": 2.0244,
+      "step": 13320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.998108619292083e-07,
+      "loss": 1.9509,
+      "step": 13325
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9913536881923804e-07,
+      "loss": 2.064,
+      "step": 13330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9845987570926776e-07,
+      "loss": 1.8872,
+      "step": 13335
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9778438259929746e-07,
+      "loss": 1.9271,
+      "step": 13340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.971088894893272e-07,
+      "loss": 1.9738,
+      "step": 13345
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9643339637935692e-07,
+      "loss": 1.9089,
+      "step": 13350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9575790326938664e-07,
+      "loss": 1.9611,
+      "step": 13355
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9508241015941637e-07,
+      "loss": 1.8635,
+      "step": 13360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.944069170494461e-07,
+      "loss": 2.0379,
+      "step": 13365
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.937314239394758e-07,
+      "loss": 1.9776,
+      "step": 13370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9305593082950552e-07,
+      "loss": 1.8788,
+      "step": 13375
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9238043771953525e-07,
+      "loss": 1.9518,
+      "step": 13380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9170494460956497e-07,
+      "loss": 2.0224,
+      "step": 13385
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.910294514995947e-07,
+      "loss": 1.8859,
+      "step": 13390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9035395838962443e-07,
+      "loss": 1.9991,
+      "step": 13395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8967846527965415e-07,
+      "loss": 1.9029,
+      "step": 13400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8900297216968385e-07,
+      "loss": 1.9768,
+      "step": 13405
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8832747905971358e-07,
+      "loss": 1.8818,
+      "step": 13410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.876519859497433e-07,
+      "loss": 1.8601,
+      "step": 13415
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.86976492839773e-07,
+      "loss": 1.9527,
+      "step": 13420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8630099972980276e-07,
+      "loss": 1.9236,
+      "step": 13425
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8562550661983248e-07,
+      "loss": 1.8954,
+      "step": 13430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8495001350986218e-07,
+      "loss": 2.0441,
+      "step": 13435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.842745203998919e-07,
+      "loss": 1.9821,
+      "step": 13440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8359902728992164e-07,
+      "loss": 1.958,
+      "step": 13445
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8292353417995134e-07,
+      "loss": 1.8381,
+      "step": 13450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8224804106998106e-07,
+      "loss": 1.9425,
+      "step": 13455
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8157254796001082e-07,
+      "loss": 1.9096,
+      "step": 13460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8089705485004054e-07,
+      "loss": 1.9968,
+      "step": 13465
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8022156174007024e-07,
+      "loss": 1.8728,
+      "step": 13470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7954606863009997e-07,
+      "loss": 1.9997,
+      "step": 13475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.788705755201297e-07,
+      "loss": 1.931,
+      "step": 13480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.781950824101594e-07,
+      "loss": 1.9951,
+      "step": 13485
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7751958930018912e-07,
+      "loss": 1.9485,
+      "step": 13490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7684409619021885e-07,
+      "loss": 2.0155,
+      "step": 13495
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7616860308024857e-07,
+      "loss": 1.9448,
+      "step": 13500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.754931099702783e-07,
+      "loss": 2.0389,
+      "step": 13505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7481761686030803e-07,
+      "loss": 1.9617,
+      "step": 13510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7414212375033773e-07,
+      "loss": 2.0045,
+      "step": 13515
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7346663064036745e-07,
+      "loss": 1.8886,
+      "step": 13520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7279113753039718e-07,
+      "loss": 2.0266,
+      "step": 13525
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.721156444204269e-07,
+      "loss": 2.0236,
+      "step": 13530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7144015131045663e-07,
+      "loss": 1.9988,
+      "step": 13535
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7076465820048636e-07,
+      "loss": 1.8733,
+      "step": 13540
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7008916509051608e-07,
+      "loss": 1.886,
+      "step": 13545
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6941367198054578e-07,
+      "loss": 2.0076,
+      "step": 13550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.687381788705755e-07,
+      "loss": 1.9322,
+      "step": 13555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6806268576060524e-07,
+      "loss": 1.955,
+      "step": 13560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6738719265063494e-07,
+      "loss": 1.9884,
+      "step": 13565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.667116995406647e-07,
+      "loss": 1.9705,
+      "step": 13570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6603620643069441e-07,
+      "loss": 2.0089,
+      "step": 13575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6536071332072411e-07,
+      "loss": 2.0441,
+      "step": 13580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6468522021075384e-07,
+      "loss": 1.9654,
+      "step": 13585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6400972710078357e-07,
+      "loss": 1.919,
+      "step": 13590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.633342339908133e-07,
+      "loss": 2.0188,
+      "step": 13595
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.62658740880843e-07,
+      "loss": 2.0181,
+      "step": 13600
+    },
+    {
+      "epoch": 1.84,
+      "eval_loss": 1.9488131999969482,
+      "eval_runtime": 165.4599,
+      "eval_samples_per_second": 3.602,
+      "eval_steps_per_second": 0.453,
+      "step": 13600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6198324777087272e-07,
+      "loss": 1.8996,
+      "step": 13605
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6130775466090247e-07,
+      "loss": 1.9036,
+      "step": 13610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6063226155093217e-07,
+      "loss": 1.9854,
+      "step": 13615
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.599567684409619e-07,
+      "loss": 2.0518,
+      "step": 13620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5928127533099162e-07,
+      "loss": 1.9395,
+      "step": 13625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5860578222102132e-07,
+      "loss": 1.941,
+      "step": 13630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5793028911105105e-07,
+      "loss": 1.8729,
+      "step": 13635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5725479600108078e-07,
+      "loss": 2.0032,
+      "step": 13640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.565793028911105e-07,
+      "loss": 1.8852,
+      "step": 13645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5590380978114023e-07,
+      "loss": 2.0203,
+      "step": 13650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5522831667116996e-07,
+      "loss": 2.0542,
+      "step": 13655
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5455282356119968e-07,
+      "loss": 1.979,
+      "step": 13660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5387733045122938e-07,
+      "loss": 1.9459,
+      "step": 13665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.532018373412591e-07,
+      "loss": 1.9584,
+      "step": 13670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5252634423128883e-07,
+      "loss": 2.0211,
+      "step": 13675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5185085112131856e-07,
+      "loss": 1.9053,
+      "step": 13680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5117535801134829e-07,
+      "loss": 1.9061,
+      "step": 13685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.50499864901378e-07,
+      "loss": 1.963,
+      "step": 13690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.498243717914077e-07,
+      "loss": 1.9455,
+      "step": 13695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4914887868143744e-07,
+      "loss": 1.9384,
+      "step": 13700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4847338557146717e-07,
+      "loss": 1.974,
+      "step": 13705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4779789246149686e-07,
+      "loss": 2.0507,
+      "step": 13710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4712239935152662e-07,
+      "loss": 1.8949,
+      "step": 13715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4644690624155634e-07,
+      "loss": 1.9455,
+      "step": 13720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4577141313158607e-07,
+      "loss": 1.8905,
+      "step": 13725
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509592002161577e-07,
+      "loss": 1.9834,
+      "step": 13730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444204269116455e-07,
+      "loss": 1.9444,
+      "step": 13735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4374493380167522e-07,
+      "loss": 1.8684,
+      "step": 13740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4306944069170492e-07,
+      "loss": 1.9758,
+      "step": 13745
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4239394758173465e-07,
+      "loss": 1.9717,
+      "step": 13750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.417184544717644e-07,
+      "loss": 2.0405,
+      "step": 13755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.410429613617941e-07,
+      "loss": 1.8387,
+      "step": 13760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4036746825182383e-07,
+      "loss": 1.8584,
+      "step": 13765
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3969197514185355e-07,
+      "loss": 2.015,
+      "step": 13770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3901648203188325e-07,
+      "loss": 1.9601,
+      "step": 13775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3834098892191298e-07,
+      "loss": 2.0155,
+      "step": 13780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.376654958119427e-07,
+      "loss": 1.9741,
+      "step": 13785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3699000270197246e-07,
+      "loss": 1.9371,
+      "step": 13790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3631450959200216e-07,
+      "loss": 2.0218,
+      "step": 13795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3563901648203188e-07,
+      "loss": 2.0401,
+      "step": 13800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.349635233720616e-07,
+      "loss": 2.0355,
+      "step": 13805
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.342880302620913e-07,
+      "loss": 2.0323,
+      "step": 13810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3361253715212104e-07,
+      "loss": 1.9805,
+      "step": 13815
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3293704404215076e-07,
+      "loss": 1.9251,
+      "step": 13820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.322615509321805e-07,
+      "loss": 1.9105,
+      "step": 13825
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3158605782221022e-07,
+      "loss": 1.9058,
+      "step": 13830
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3091056471223994e-07,
+      "loss": 1.99,
+      "step": 13835
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3023507160226964e-07,
+      "loss": 2.0117,
+      "step": 13840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2955957849229937e-07,
+      "loss": 1.9481,
+      "step": 13845
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.288840853823291e-07,
+      "loss": 1.9639,
+      "step": 13850
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.282085922723588e-07,
+      "loss": 1.8786,
+      "step": 13855
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2753309916238852e-07,
+      "loss": 1.9065,
+      "step": 13860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2685760605241827e-07,
+      "loss": 2.0212,
+      "step": 13865
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.26182112942448e-07,
+      "loss": 2.0138,
+      "step": 13870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.255066198324777e-07,
+      "loss": 1.8938,
+      "step": 13875
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2483112672250743e-07,
+      "loss": 2.0057,
+      "step": 13880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2415563361253715e-07,
+      "loss": 1.9609,
+      "step": 13885
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2348014050256688e-07,
+      "loss": 1.9497,
+      "step": 13890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2280464739259658e-07,
+      "loss": 1.9263,
+      "step": 13895
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.221291542826263e-07,
+      "loss": 1.9301,
+      "step": 13900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2145366117265603e-07,
+      "loss": 1.9621,
+      "step": 13905
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2077816806268576e-07,
+      "loss": 1.9314,
+      "step": 13910
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2010267495271548e-07,
+      "loss": 1.9993,
+      "step": 13915
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1942718184274518e-07,
+      "loss": 1.9101,
+      "step": 13920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1875168873277492e-07,
+      "loss": 1.9138,
+      "step": 13925
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1807619562280465e-07,
+      "loss": 2.0387,
+      "step": 13930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1740070251283436e-07,
+      "loss": 1.9822,
+      "step": 13935
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1672520940286408e-07,
+      "loss": 1.8946,
+      "step": 13940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1604971629289381e-07,
+      "loss": 1.9925,
+      "step": 13945
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1537422318292353e-07,
+      "loss": 1.9311,
+      "step": 13950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1469873007295325e-07,
+      "loss": 1.9261,
+      "step": 13955
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1402323696298297e-07,
+      "loss": 2.0045,
+      "step": 13960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1334774385301269e-07,
+      "loss": 1.932,
+      "step": 13965
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1267225074304242e-07,
+      "loss": 1.9065,
+      "step": 13970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1199675763307213e-07,
+      "loss": 1.8788,
+      "step": 13975
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1132126452310187e-07,
+      "loss": 1.9045,
+      "step": 13980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1064577141313159e-07,
+      "loss": 1.9846,
+      "step": 13985
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.099702783031613e-07,
+      "loss": 1.9391,
+      "step": 13990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0929478519319102e-07,
+      "loss": 2.0232,
+      "step": 13995
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0861929208322075e-07,
+      "loss": 1.9595,
+      "step": 14000
+    },
+    {
+      "epoch": 1.89,
+      "eval_loss": 1.9482014179229736,
+      "eval_runtime": 165.3869,
+      "eval_samples_per_second": 3.604,
+      "eval_steps_per_second": 0.453,
+      "step": 14000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0794379897325046e-07,
+      "loss": 1.995,
+      "step": 14005
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0726830586328019e-07,
+      "loss": 1.9835,
+      "step": 14010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.065928127533099e-07,
+      "loss": 1.9466,
+      "step": 14015
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0591731964333964e-07,
+      "loss": 2.0073,
+      "step": 14020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0524182653336936e-07,
+      "loss": 2.0322,
+      "step": 14025
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0456633342339907e-07,
+      "loss": 1.9171,
+      "step": 14030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0389084031342881e-07,
+      "loss": 1.9881,
+      "step": 14035
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0321534720345852e-07,
+      "loss": 2.0138,
+      "step": 14040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0253985409348823e-07,
+      "loss": 2.0584,
+      "step": 14045
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0186436098351796e-07,
+      "loss": 1.926,
+      "step": 14050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0118886787354769e-07,
+      "loss": 1.9551,
+      "step": 14055
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0051337476357741e-07,
+      "loss": 1.9745,
+      "step": 14060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.983788165360713e-08,
+      "loss": 2.0285,
+      "step": 14065
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.916238854363685e-08,
+      "loss": 2.0206,
+      "step": 14070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.848689543366658e-08,
+      "loss": 1.9034,
+      "step": 14075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.781140232369629e-08,
+      "loss": 1.9332,
+      "step": 14080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.7135909213726e-08,
+      "loss": 1.8562,
+      "step": 14085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.646041610375574e-08,
+      "loss": 1.9742,
+      "step": 14090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.578492299378546e-08,
+      "loss": 1.9395,
+      "step": 14095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.510942988381518e-08,
+      "loss": 2.0896,
+      "step": 14100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.44339367738449e-08,
+      "loss": 1.9833,
+      "step": 14105
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.375844366387462e-08,
+      "loss": 1.909,
+      "step": 14110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.308295055390435e-08,
+      "loss": 1.8967,
+      "step": 14115
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.240745744393406e-08,
+      "loss": 1.9244,
+      "step": 14120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.17319643339638e-08,
+      "loss": 1.8847,
+      "step": 14125
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.105647122399351e-08,
+      "loss": 1.9759,
+      "step": 14130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.038097811402323e-08,
+      "loss": 2.0605,
+      "step": 14135
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.970548500405295e-08,
+      "loss": 2.0337,
+      "step": 14140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.902999189408268e-08,
+      "loss": 2.0754,
+      "step": 14145
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.835449878411239e-08,
+      "loss": 2.1058,
+      "step": 14150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.767900567414212e-08,
+      "loss": 1.9074,
+      "step": 14155
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.700351256417183e-08,
+      "loss": 1.9853,
+      "step": 14160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.632801945420157e-08,
+      "loss": 1.9416,
+      "step": 14165
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.565252634423129e-08,
+      "loss": 1.9635,
+      "step": 14170
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.4977033234261e-08,
+      "loss": 1.9167,
+      "step": 14175
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.430154012429074e-08,
+      "loss": 2.0947,
+      "step": 14180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.362604701432045e-08,
+      "loss": 1.9955,
+      "step": 14185
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.295055390435018e-08,
+      "loss": 1.9231,
+      "step": 14190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.227506079437989e-08,
+      "loss": 1.9107,
+      "step": 14195
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.159956768440962e-08,
+      "loss": 1.9193,
+      "step": 14200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.092407457443934e-08,
+      "loss": 1.955,
+      "step": 14205
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.024858146446906e-08,
+      "loss": 2.0003,
+      "step": 14210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.957308835449877e-08,
+      "loss": 1.993,
+      "step": 14215
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.889759524452851e-08,
+      "loss": 1.9449,
+      "step": 14220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.822210213455822e-08,
+      "loss": 2.0275,
+      "step": 14225
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.754660902458795e-08,
+      "loss": 1.955,
+      "step": 14230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.687111591461767e-08,
+      "loss": 1.9618,
+      "step": 14235
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.619562280464739e-08,
+      "loss": 1.9257,
+      "step": 14240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.552012969467711e-08,
+      "loss": 1.9296,
+      "step": 14245
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.484463658470683e-08,
+      "loss": 2.0142,
+      "step": 14250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.416914347473655e-08,
+      "loss": 1.9446,
+      "step": 14255
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.349365036476628e-08,
+      "loss": 1.9638,
+      "step": 14260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.281815725479599e-08,
+      "loss": 1.9541,
+      "step": 14265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.214266414482572e-08,
+      "loss": 1.9324,
+      "step": 14270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.146717103485544e-08,
+      "loss": 1.9464,
+      "step": 14275
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.079167792488516e-08,
+      "loss": 1.8618,
+      "step": 14280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.011618481491488e-08,
+      "loss": 1.8554,
+      "step": 14285
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.944069170494461e-08,
+      "loss": 1.8643,
+      "step": 14290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.876519859497434e-08,
+      "loss": 1.8951,
+      "step": 14295
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.808970548500405e-08,
+      "loss": 2.0413,
+      "step": 14300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.741421237503376e-08,
+      "loss": 2.0529,
+      "step": 14305
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.67387192650635e-08,
+      "loss": 1.9858,
+      "step": 14310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.606322615509321e-08,
+      "loss": 1.9525,
+      "step": 14315
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.538773304512293e-08,
+      "loss": 1.9431,
+      "step": 14320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.471223993515267e-08,
+      "loss": 2.0377,
+      "step": 14325
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.403674682518238e-08,
+      "loss": 1.804,
+      "step": 14330
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.336125371521211e-08,
+      "loss": 2.0369,
+      "step": 14335
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.268576060524182e-08,
+      "loss": 1.9369,
+      "step": 14340
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.201026749527155e-08,
+      "loss": 1.9237,
+      "step": 14345
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.133477438530127e-08,
+      "loss": 1.9972,
+      "step": 14350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.065928127533099e-08,
+      "loss": 1.9778,
+      "step": 14355
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.998378816536071e-08,
+      "loss": 1.9137,
+      "step": 14360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.930829505539043e-08,
+      "loss": 1.9997,
+      "step": 14365
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.863280194542016e-08,
+      "loss": 1.9587,
+      "step": 14370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.795730883544987e-08,
+      "loss": 2.0108,
+      "step": 14375
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.72818157254796e-08,
+      "loss": 1.8876,
+      "step": 14380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.660632261550932e-08,
+      "loss": 2.042,
+      "step": 14385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.593082950553904e-08,
+      "loss": 1.9656,
+      "step": 14390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.525533639556876e-08,
+      "loss": 1.9729,
+      "step": 14395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.457984328559848e-08,
+      "loss": 2.0976,
+      "step": 14400
+    },
+    {
+      "epoch": 1.95,
+      "eval_loss": 1.9481086730957031,
+      "eval_runtime": 165.5103,
+      "eval_samples_per_second": 3.601,
+      "eval_steps_per_second": 0.453,
+      "step": 14400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.390435017562821e-08,
+      "loss": 1.8936,
+      "step": 14405
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.322885706565793e-08,
+      "loss": 1.9298,
+      "step": 14410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.255336395568765e-08,
+      "loss": 1.9587,
+      "step": 14415
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.187787084571737e-08,
+      "loss": 1.8686,
+      "step": 14420
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1202377735747094e-08,
+      "loss": 1.92,
+      "step": 14425
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.052688462577682e-08,
+      "loss": 1.9898,
+      "step": 14430
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.985139151580653e-08,
+      "loss": 1.9962,
+      "step": 14435
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.917589840583626e-08,
+      "loss": 1.9581,
+      "step": 14440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.850040529586598e-08,
+      "loss": 2.0457,
+      "step": 14445
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7824912185895705e-08,
+      "loss": 2.0811,
+      "step": 14450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.714941907592542e-08,
+      "loss": 1.9705,
+      "step": 14455
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6473925965955145e-08,
+      "loss": 1.8892,
+      "step": 14460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5798432855984864e-08,
+      "loss": 1.9186,
+      "step": 14465
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.512293974601459e-08,
+      "loss": 2.0552,
+      "step": 14470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.444744663604431e-08,
+      "loss": 1.9579,
+      "step": 14475
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.377195352607403e-08,
+      "loss": 1.9762,
+      "step": 14480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3096460416103756e-08,
+      "loss": 2.0308,
+      "step": 14485
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2420967306133476e-08,
+      "loss": 1.9251,
+      "step": 14490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.17454741961632e-08,
+      "loss": 1.9119,
+      "step": 14495
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.1069981086192915e-08,
+      "loss": 1.9872,
+      "step": 14500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.039448797622264e-08,
+      "loss": 1.9612,
+      "step": 14505
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.971899486625236e-08,
+      "loss": 1.8368,
+      "step": 14510
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.904350175628209e-08,
+      "loss": 2.0556,
+      "step": 14515
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.83680086463118e-08,
+      "loss": 1.9518,
+      "step": 14520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7692515536341527e-08,
+      "loss": 1.9561,
+      "step": 14525
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.701702242637125e-08,
+      "loss": 2.0511,
+      "step": 14530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.634152931640097e-08,
+      "loss": 2.0264,
+      "step": 14535
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.56660362064307e-08,
+      "loss": 1.9518,
+      "step": 14540
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.499054309646041e-08,
+      "loss": 1.958,
+      "step": 14545
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.431504998649014e-08,
+      "loss": 1.8629,
+      "step": 14550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.363955687651986e-08,
+      "loss": 1.9151,
+      "step": 14555
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.2964063766549584e-08,
+      "loss": 2.046,
+      "step": 14560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.22885706565793e-08,
+      "loss": 1.9737,
+      "step": 14565
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1613077546609023e-08,
+      "loss": 1.9286,
+      "step": 14570
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.093758443663874e-08,
+      "loss": 1.8991,
+      "step": 14575
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.026209132666847e-08,
+      "loss": 1.9493,
+      "step": 14580
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.958659821669819e-08,
+      "loss": 1.9246,
+      "step": 14585
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.891110510672791e-08,
+      "loss": 2.0081,
+      "step": 14590
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.823561199675763e-08,
+      "loss": 1.8692,
+      "step": 14595
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.756011888678735e-08,
+      "loss": 1.9802,
+      "step": 14600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6884625776817074e-08,
+      "loss": 1.9681,
+      "step": 14605
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6209132666846797e-08,
+      "loss": 1.9555,
+      "step": 14610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.553363955687652e-08,
+      "loss": 2.0044,
+      "step": 14615
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.485814644690624e-08,
+      "loss": 1.9247,
+      "step": 14620
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4182653336935963e-08,
+      "loss": 1.9702,
+      "step": 14625
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3507160226965682e-08,
+      "loss": 1.9554,
+      "step": 14630
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2831667116995405e-08,
+      "loss": 2.0136,
+      "step": 14635
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.215617400702513e-08,
+      "loss": 1.9539,
+      "step": 14640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1480680897054848e-08,
+      "loss": 2.0292,
+      "step": 14645
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.080518778708457e-08,
+      "loss": 1.9981,
+      "step": 14650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.012969467711429e-08,
+      "loss": 2.0681,
+      "step": 14655
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9454201567144017e-08,
+      "loss": 2.0592,
+      "step": 14660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8778708457173737e-08,
+      "loss": 1.9643,
+      "step": 14665
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.810321534720346e-08,
+      "loss": 2.0253,
+      "step": 14670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.742772223723318e-08,
+      "loss": 2.0889,
+      "step": 14675
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6752229127262902e-08,
+      "loss": 1.9485,
+      "step": 14680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6076736017292622e-08,
+      "loss": 1.8839,
+      "step": 14685
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5401242907322345e-08,
+      "loss": 1.9396,
+      "step": 14690
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4725749797352068e-08,
+      "loss": 1.9704,
+      "step": 14695
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4050256687381789e-08,
+      "loss": 1.9368,
+      "step": 14700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.337476357741151e-08,
+      "loss": 1.9722,
+      "step": 14705
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2699270467441232e-08,
+      "loss": 2.0104,
+      "step": 14710
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2023777357470953e-08,
+      "loss": 1.9681,
+      "step": 14715
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1348284247500676e-08,
+      "loss": 1.9765,
+      "step": 14720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0672791137530397e-08,
+      "loss": 1.9395,
+      "step": 14725
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.997298027560119e-09,
+      "loss": 1.9876,
+      "step": 14730
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.32180491758984e-09,
+      "loss": 1.9795,
+      "step": 14735
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.646311807619561e-09,
+      "loss": 1.9714,
+      "step": 14740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.970818697649283e-09,
+      "loss": 1.9901,
+      "step": 14745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.295325587679005e-09,
+      "loss": 2.0808,
+      "step": 14750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.619832477708728e-09,
+      "loss": 1.9317,
+      "step": 14755
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.944339367738449e-09,
+      "loss": 1.9596,
+      "step": 14760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.268846257768171e-09,
+      "loss": 1.9245,
+      "step": 14765
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.5933531477978924e-09,
+      "loss": 2.0051,
+      "step": 14770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.917860037827614e-09,
+      "loss": 1.9284,
+      "step": 14775
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.2423669278573355e-09,
+      "loss": 1.9367,
+      "step": 14780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.566873817887057e-09,
+      "loss": 2.0314,
+      "step": 14785
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8913807079167793e-09,
+      "loss": 1.8842,
+      "step": 14790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.215887597946501e-09,
+      "loss": 1.9025,
+      "step": 14795
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.403944879762226e-10,
+      "loss": 1.9204,
+      "step": 14800
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.9478894472122192,
+      "eval_runtime": 165.3905,
+      "eval_samples_per_second": 3.604,
+      "eval_steps_per_second": 0.453,
+      "step": 14800
+    },
+    {
+      "epoch": 2.0,
+      "step": 14804,
+      "total_flos": 8.826710223564841e+18,
+      "train_loss": 2.036540643423514,
+      "train_runtime": 233902.6153,
+      "train_samples_per_second": 1.013,
+      "train_steps_per_second": 0.063
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 14804,
+  "num_train_epochs": 2,
+  "save_steps": 2000,
+  "total_flos": 8.826710223564841e+18,
+  "trial_name": null,
+  "trial_params": null
+}