diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5236 @@
+{
+  "best_metric": 0.07686587423086166,
+  "best_model_checkpoint": "/teamspace/studios/this_studio/save/LLama_End/checkpoint-3500",
+  "epoch": 1.9991833401388321,
+  "eval_steps": 500,
+  "global_step": 3672,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0027221995372260785,
+      "grad_norm": 0.3611927926540375,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.7652,
+      "step": 5
+    },
+    {
+      "epoch": 0.005444399074452157,
+      "grad_norm": 0.3633479177951813,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7798,
+      "step": 10
+    },
+    {
+      "epoch": 0.008166598611678236,
+      "grad_norm": 0.31540995836257935,
+      "learning_rate": 1.5e-06,
+      "loss": 0.7635,
+      "step": 15
+    },
+    {
+      "epoch": 0.010888798148904314,
+      "grad_norm": 0.3314383924007416,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7475,
+      "step": 20
+    },
+    {
+      "epoch": 0.013610997686130393,
+      "grad_norm": 0.391197144985199,
+      "learning_rate": 2.5e-06,
+      "loss": 0.7827,
+      "step": 25
+    },
+    {
+      "epoch": 0.016333197223356473,
+      "grad_norm": 0.39311718940734863,
+      "learning_rate": 3e-06,
+      "loss": 0.7985,
+      "step": 30
+    },
+    {
+      "epoch": 0.01905539676058255,
+      "grad_norm": 0.34564024209976196,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 0.7608,
+      "step": 35
+    },
+    {
+      "epoch": 0.021777596297808628,
+      "grad_norm": 0.43378740549087524,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7566,
+      "step": 40
+    },
+    {
+      "epoch": 0.02449979583503471,
+      "grad_norm": 0.38521063327789307,
+      "learning_rate": 4.5e-06,
+      "loss": 0.75,
+      "step": 45
+    },
+    {
+      "epoch": 0.027221995372260787,
+      "grad_norm": 0.4021587371826172,
+      "learning_rate": 5e-06,
+      "loss": 0.7492,
+      "step": 50
+    },
+    {
+      "epoch": 0.029944194909486865,
+      "grad_norm": 0.44062522053718567,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.7208,
+      "step": 55
+    },
+    {
+      "epoch": 0.032666394446712946,
+      "grad_norm": 0.37620142102241516,
+      "learning_rate": 6e-06,
+      "loss": 0.6911,
+      "step": 60
+    },
+    {
+      "epoch": 0.03538859398393902,
+      "grad_norm": 0.27264639735221863,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.6624,
+      "step": 65
+    },
+    {
+      "epoch": 0.0381107935211651,
+      "grad_norm": 0.2566782832145691,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6326,
+      "step": 70
+    },
+    {
+      "epoch": 0.04083299305839118,
+      "grad_norm": 0.27515003085136414,
+      "learning_rate": 7.5e-06,
+      "loss": 0.6135,
+      "step": 75
+    },
+    {
+      "epoch": 0.043555192595617256,
+      "grad_norm": 0.2592063248157501,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.5585,
+      "step": 80
+    },
+    {
+      "epoch": 0.046277392132843334,
+      "grad_norm": 0.22615216672420502,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 0.5482,
+      "step": 85
+    },
+    {
+      "epoch": 0.04899959167006942,
+      "grad_norm": 0.21914538741111755,
+      "learning_rate": 9e-06,
+      "loss": 0.5488,
+      "step": 90
+    },
+    {
+      "epoch": 0.051721791207295496,
+      "grad_norm": 0.23039759695529938,
+      "learning_rate": 9.5e-06,
+      "loss": 0.4816,
+      "step": 95
+    },
+    {
+      "epoch": 0.054443990744521574,
+      "grad_norm": 0.21080121397972107,
+      "learning_rate": 1e-05,
+      "loss": 0.4686,
+      "step": 100
+    },
+    {
+      "epoch": 0.05716619028174765,
+      "grad_norm": 0.19633150100708008,
+      "learning_rate": 1.05e-05,
+      "loss": 0.4426,
+      "step": 105
+    },
+    {
+      "epoch": 0.05988838981897373,
+      "grad_norm": 0.20176054537296295,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.4131,
+      "step": 110
+    },
+    {
+      "epoch": 0.06261058935619981,
+      "grad_norm": 0.1965443640947342,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.394,
+      "step": 115
+    },
+    {
+      "epoch": 0.06533278889342589,
+      "grad_norm": 0.18852515518665314,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3621,
+      "step": 120
+    },
+    {
+      "epoch": 0.06805498843065197,
+      "grad_norm": 0.1972607970237732,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3436,
+      "step": 125
+    },
+    {
+      "epoch": 0.07077718796787805,
+      "grad_norm": 0.1883157640695572,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.3235,
+      "step": 130
+    },
+    {
+      "epoch": 0.07349938750510412,
+      "grad_norm": 0.20938365161418915,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.3093,
+      "step": 135
+    },
+    {
+      "epoch": 0.0762215870423302,
+      "grad_norm": 0.2136233150959015,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.281,
+      "step": 140
+    },
+    {
+      "epoch": 0.07894378657955628,
+      "grad_norm": 0.23373591899871826,
+      "learning_rate": 1.45e-05,
+      "loss": 0.2645,
+      "step": 145
+    },
+    {
+      "epoch": 0.08166598611678236,
+      "grad_norm": 0.21917635202407837,
+      "learning_rate": 1.5e-05,
+      "loss": 0.2399,
+      "step": 150
+    },
+    {
+      "epoch": 0.08438818565400844,
+      "grad_norm": 0.2619529068470001,
+      "learning_rate": 1.55e-05,
+      "loss": 0.2381,
+      "step": 155
+    },
+    {
+      "epoch": 0.08711038519123451,
+      "grad_norm": 0.2083514928817749,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2139,
+      "step": 160
+    },
+    {
+      "epoch": 0.08983258472846059,
+      "grad_norm": 0.22957220673561096,
+      "learning_rate": 1.65e-05,
+      "loss": 0.1956,
+      "step": 165
+    },
+    {
+      "epoch": 0.09255478426568667,
+      "grad_norm": 0.270378977060318,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.1906,
+      "step": 170
+    },
+    {
+      "epoch": 0.09527698380291276,
+      "grad_norm": 0.24123062193393707,
+      "learning_rate": 1.75e-05,
+      "loss": 0.1811,
+      "step": 175
+    },
+    {
+      "epoch": 0.09799918334013884,
+      "grad_norm": 0.2726229429244995,
+      "learning_rate": 1.8e-05,
+      "loss": 0.1756,
+      "step": 180
+    },
+    {
+      "epoch": 0.10072138287736492,
+      "grad_norm": 0.24878880381584167,
+      "learning_rate": 1.85e-05,
+      "loss": 0.1745,
+      "step": 185
+    },
+    {
+      "epoch": 0.10344358241459099,
+      "grad_norm": 0.22229672968387604,
+      "learning_rate": 1.9e-05,
+      "loss": 0.1641,
+      "step": 190
+    },
+    {
+      "epoch": 0.10616578195181707,
+      "grad_norm": 0.25277239084243774,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.1478,
+      "step": 195
+    },
+    {
+      "epoch": 0.10888798148904315,
+      "grad_norm": 0.23793649673461914,
+      "learning_rate": 2e-05,
+      "loss": 0.1621,
+      "step": 200
+    },
+    {
+      "epoch": 0.11161018102626923,
+      "grad_norm": 0.22567716240882874,
+      "learning_rate": 2.05e-05,
+      "loss": 0.1416,
+      "step": 205
+    },
+    {
+      "epoch": 0.1143323805634953,
+      "grad_norm": 0.24615788459777832,
+      "learning_rate": 2.1e-05,
+      "loss": 0.1418,
+      "step": 210
+    },
+    {
+      "epoch": 0.11705458010072138,
+      "grad_norm": 0.23229427635669708,
+      "learning_rate": 2.15e-05,
+      "loss": 0.137,
+      "step": 215
+    },
+    {
+      "epoch": 0.11977677963794746,
+      "grad_norm": 0.2679113447666168,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.1186,
+      "step": 220
+    },
+    {
+      "epoch": 0.12249897917517354,
+      "grad_norm": 0.24723300337791443,
+      "learning_rate": 2.25e-05,
+      "loss": 0.1243,
+      "step": 225
+    },
+    {
+      "epoch": 0.12522117871239963,
+      "grad_norm": 0.24604055285453796,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.1472,
+      "step": 230
+    },
+    {
+      "epoch": 0.1279433782496257,
+      "grad_norm": 0.2207736223936081,
+      "learning_rate": 2.35e-05,
+      "loss": 0.1258,
+      "step": 235
+    },
+    {
+      "epoch": 0.13066557778685178,
+      "grad_norm": 0.2685292065143585,
+      "learning_rate": 2.4e-05,
+      "loss": 0.1407,
+      "step": 240
+    },
+    {
+      "epoch": 0.13338777732407786,
+      "grad_norm": 0.21564778685569763,
+      "learning_rate": 2.45e-05,
+      "loss": 0.1355,
+      "step": 245
+    },
+    {
+      "epoch": 0.13610997686130394,
+      "grad_norm": 0.22922654449939728,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1298,
+      "step": 250
+    },
+    {
+      "epoch": 0.13883217639853002,
+      "grad_norm": 0.21827565133571625,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 0.1256,
+      "step": 255
+    },
+    {
+      "epoch": 0.1415543759357561,
+      "grad_norm": 0.2129678726196289,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.1435,
+      "step": 260
+    },
+    {
+      "epoch": 0.14427657547298217,
+      "grad_norm": 0.196010559797287,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 0.1353,
+      "step": 265
+    },
+    {
+      "epoch": 0.14699877501020825,
+      "grad_norm": 0.22913923859596252,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.1187,
+      "step": 270
+    },
+    {
+      "epoch": 0.14972097454743433,
+      "grad_norm": 0.26281264424324036,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.1697,
+      "step": 275
+    },
+    {
+      "epoch": 0.1524431740846604,
+      "grad_norm": 0.28295812010765076,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.1229,
+      "step": 280
+    },
+    {
+      "epoch": 0.15516537362188648,
+      "grad_norm": 0.23690921068191528,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 0.1086,
+      "step": 285
+    },
+    {
+      "epoch": 0.15788757315911256,
+      "grad_norm": 0.22863948345184326,
+      "learning_rate": 2.9e-05,
+      "loss": 0.1064,
+      "step": 290
+    },
+    {
+      "epoch": 0.16060977269633864,
+      "grad_norm": 0.1989349126815796,
+      "learning_rate": 2.95e-05,
+      "loss": 0.1145,
+      "step": 295
+    },
+    {
+      "epoch": 0.16333197223356472,
+      "grad_norm": 0.2325228750705719,
+      "learning_rate": 3e-05,
+      "loss": 0.1181,
+      "step": 300
+    },
+    {
+      "epoch": 0.1660541717707908,
+      "grad_norm": 0.2611011266708374,
+      "learning_rate": 3.05e-05,
+      "loss": 0.1093,
+      "step": 305
+    },
+    {
+      "epoch": 0.16877637130801687,
+      "grad_norm": 0.23333916068077087,
+      "learning_rate": 3.1e-05,
+      "loss": 0.1119,
+      "step": 310
+    },
+    {
+      "epoch": 0.17149857084524295,
+      "grad_norm": 0.22769390046596527,
+      "learning_rate": 3.15e-05,
+      "loss": 0.1079,
+      "step": 315
+    },
+    {
+      "epoch": 0.17422077038246903,
+      "grad_norm": 0.23360049724578857,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.1035,
+      "step": 320
+    },
+    {
+      "epoch": 0.1769429699196951,
+      "grad_norm": 0.2405271828174591,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.1208,
+      "step": 325
+    },
+    {
+      "epoch": 0.17966516945692118,
+      "grad_norm": 0.2959926128387451,
+      "learning_rate": 3.3e-05,
+      "loss": 0.113,
+      "step": 330
+    },
+    {
+      "epoch": 0.18238736899414726,
+      "grad_norm": 0.2246614694595337,
+      "learning_rate": 3.35e-05,
+      "loss": 0.1082,
+      "step": 335
+    },
+    {
+      "epoch": 0.18510956853137334,
+      "grad_norm": 0.23407892882823944,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.103,
+      "step": 340
+    },
+    {
+      "epoch": 0.18783176806859944,
+      "grad_norm": 0.21018989384174347,
+      "learning_rate": 3.45e-05,
+      "loss": 0.1084,
+      "step": 345
+    },
+    {
+      "epoch": 0.19055396760582552,
+      "grad_norm": 0.1964080035686493,
+      "learning_rate": 3.5e-05,
+      "loss": 0.1021,
+      "step": 350
+    },
+    {
+      "epoch": 0.1932761671430516,
+      "grad_norm": 0.2037973254919052,
+      "learning_rate": 3.55e-05,
+      "loss": 0.1088,
+      "step": 355
+    },
+    {
+      "epoch": 0.19599836668027767,
+      "grad_norm": 0.2156139761209488,
+      "learning_rate": 3.6e-05,
+      "loss": 0.1104,
+      "step": 360
+    },
+    {
+      "epoch": 0.19872056621750375,
+      "grad_norm": 0.21694843471050262,
+      "learning_rate": 3.65e-05,
+      "loss": 0.1073,
+      "step": 365
+    },
+    {
+      "epoch": 0.20144276575472983,
+      "grad_norm": 0.21144092082977295,
+      "learning_rate": 3.7e-05,
+      "loss": 0.1365,
+      "step": 370
+    },
+    {
+      "epoch": 0.2041649652919559,
+      "grad_norm": 0.24178194999694824,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.0984,
+      "step": 375
+    },
+    {
+      "epoch": 0.20688716482918199,
+      "grad_norm": 0.20173633098602295,
+      "learning_rate": 3.8e-05,
+      "loss": 0.1289,
+      "step": 380
+    },
+    {
+      "epoch": 0.20960936436640806,
+      "grad_norm": 0.2253100723028183,
+      "learning_rate": 3.85e-05,
+      "loss": 0.1055,
+      "step": 385
+    },
+    {
+      "epoch": 0.21233156390363414,
+      "grad_norm": 0.2524998188018799,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.1203,
+      "step": 390
+    },
+    {
+      "epoch": 0.21505376344086022,
+      "grad_norm": 0.21095526218414307,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 0.0965,
+      "step": 395
+    },
+    {
+      "epoch": 0.2177759629780863,
+      "grad_norm": 0.1919548511505127,
+      "learning_rate": 4e-05,
+      "loss": 0.0961,
+      "step": 400
+    },
+    {
+      "epoch": 0.22049816251531237,
+      "grad_norm": 0.2088468074798584,
+      "learning_rate": 4.05e-05,
+      "loss": 0.0945,
+      "step": 405
+    },
+    {
+      "epoch": 0.22322036205253845,
+      "grad_norm": 0.22967375814914703,
+      "learning_rate": 4.1e-05,
+      "loss": 0.0991,
+      "step": 410
+    },
+    {
+      "epoch": 0.22594256158976453,
+      "grad_norm": 0.19900205731391907,
+      "learning_rate": 4.15e-05,
+      "loss": 0.1069,
+      "step": 415
+    },
+    {
+      "epoch": 0.2286647611269906,
+      "grad_norm": 0.22003936767578125,
+      "learning_rate": 4.2e-05,
+      "loss": 0.1037,
+      "step": 420
+    },
+    {
+      "epoch": 0.23138696066421668,
+      "grad_norm": 0.21360959112644196,
+      "learning_rate": 4.25e-05,
+      "loss": 0.0952,
+      "step": 425
+    },
+    {
+      "epoch": 0.23410916020144276,
+      "grad_norm": 0.1812208890914917,
+      "learning_rate": 4.3e-05,
+      "loss": 0.0962,
+      "step": 430
+    },
+    {
+      "epoch": 0.23683135973866884,
+      "grad_norm": 0.19043520092964172,
+      "learning_rate": 4.35e-05,
+      "loss": 0.1075,
+      "step": 435
+    },
+    {
+      "epoch": 0.23955355927589492,
+      "grad_norm": 0.2013886421918869,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.104,
+      "step": 440
+    },
+    {
+      "epoch": 0.242275758813121,
+      "grad_norm": 0.1831788420677185,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 0.1012,
+      "step": 445
+    },
+    {
+      "epoch": 0.24499795835034707,
+      "grad_norm": 0.1884724646806717,
+      "learning_rate": 4.5e-05,
+      "loss": 0.1115,
+      "step": 450
+    },
+    {
+      "epoch": 0.24772015788757315,
+      "grad_norm": 0.1994744837284088,
+      "learning_rate": 4.55e-05,
+      "loss": 0.1006,
+      "step": 455
+    },
+    {
+      "epoch": 0.25044235742479926,
+      "grad_norm": 0.1783529371023178,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.0902,
+      "step": 460
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 0.180820032954216,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 0.0937,
+      "step": 465
+    },
+    {
+      "epoch": 0.2558867564992514,
+      "grad_norm": 0.19279232621192932,
+      "learning_rate": 4.7e-05,
+      "loss": 0.0944,
+      "step": 470
+    },
+    {
+      "epoch": 0.2586089560364775,
+      "grad_norm": 0.1725597232580185,
+      "learning_rate": 4.75e-05,
+      "loss": 0.104,
+      "step": 475
+    },
+    {
+      "epoch": 0.26133115557370357,
+      "grad_norm": 0.17777124047279358,
+      "learning_rate": 4.8e-05,
+      "loss": 0.0889,
+      "step": 480
+    },
+    {
+      "epoch": 0.26405335511092964,
+      "grad_norm": 0.18942312896251678,
+      "learning_rate": 4.85e-05,
+      "loss": 0.0964,
+      "step": 485
+    },
+    {
+      "epoch": 0.2667755546481557,
+      "grad_norm": 0.17807991802692413,
+      "learning_rate": 4.9e-05,
+      "loss": 0.1038,
+      "step": 490
+    },
+    {
+      "epoch": 0.2694977541853818,
+      "grad_norm": 0.21202729642391205,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 0.0941,
+      "step": 495
+    },
+    {
+      "epoch": 0.2722199537226079,
+      "grad_norm": 0.18360304832458496,
+      "learning_rate": 5e-05,
+      "loss": 0.1013,
+      "step": 500
+    },
+    {
+      "epoch": 0.2722199537226079,
+      "eval_loss": 0.09772992134094238,
+      "eval_runtime": 271.0226,
+      "eval_samples_per_second": 2.214,
+      "eval_steps_per_second": 0.553,
+      "step": 500
+    },
+    {
+      "epoch": 0.27494215325983395,
+      "grad_norm": 0.1730509251356125,
+      "learning_rate": 4.999969346326857e-05,
+      "loss": 0.098,
+      "step": 505
+    },
+    {
+      "epoch": 0.27766435279706003,
+      "grad_norm": 0.17720815539360046,
+      "learning_rate": 4.9998773860591444e-05,
+      "loss": 0.0912,
+      "step": 510
+    },
+    {
+      "epoch": 0.2803865523342861,
+      "grad_norm": 0.16912730038166046,
+      "learning_rate": 4.9997241214519986e-05,
+      "loss": 0.0885,
+      "step": 515
+    },
+    {
+      "epoch": 0.2831087518715122,
+      "grad_norm": 0.166640505194664,
+      "learning_rate": 4.999509556263919e-05,
+      "loss": 0.1013,
+      "step": 520
+    },
+    {
+      "epoch": 0.28583095140873827,
+      "grad_norm": 0.1893271952867508,
+      "learning_rate": 4.999233695756673e-05,
+      "loss": 0.1005,
+      "step": 525
+    },
+    {
+      "epoch": 0.28855315094596434,
+      "grad_norm": 0.1966632455587387,
+      "learning_rate": 4.998896546695172e-05,
+      "loss": 0.0958,
+      "step": 530
+    },
+    {
+      "epoch": 0.2912753504831904,
+      "grad_norm": 0.16383662819862366,
+      "learning_rate": 4.998498117347302e-05,
+      "loss": 0.0971,
+      "step": 535
+    },
+    {
+      "epoch": 0.2939975500204165,
+      "grad_norm": 0.21534255146980286,
+      "learning_rate": 4.998038417483721e-05,
+      "loss": 0.1057,
+      "step": 540
+    },
+    {
+      "epoch": 0.2967197495576426,
+      "grad_norm": 0.16162091493606567,
+      "learning_rate": 4.9975174583776196e-05,
+      "loss": 0.0943,
+      "step": 545
+    },
+    {
+      "epoch": 0.29944194909486865,
+      "grad_norm": 0.1837451457977295,
+      "learning_rate": 4.996935252804448e-05,
+      "loss": 0.0968,
+      "step": 550
+    },
+    {
+      "epoch": 0.30216414863209473,
+      "grad_norm": 0.15664127469062805,
+      "learning_rate": 4.9962918150415946e-05,
+      "loss": 0.0975,
+      "step": 555
+    },
+    {
+      "epoch": 0.3048863481693208,
+      "grad_norm": 0.16413035988807678,
+      "learning_rate": 4.995587160868047e-05,
+      "loss": 0.0925,
+      "step": 560
+    },
+    {
+      "epoch": 0.3076085477065469,
+      "grad_norm": 0.16700997948646545,
+      "learning_rate": 4.994821307563995e-05,
+      "loss": 0.097,
+      "step": 565
+    },
+    {
+      "epoch": 0.31033074724377296,
+      "grad_norm": 0.16121503710746765,
+      "learning_rate": 4.9939942739104105e-05,
+      "loss": 0.0996,
+      "step": 570
+    },
+    {
+      "epoch": 0.31305294678099904,
+      "grad_norm": 0.16212740540504456,
+      "learning_rate": 4.993106080188592e-05,
+      "loss": 0.0888,
+      "step": 575
+    },
+    {
+      "epoch": 0.3157751463182251,
+      "grad_norm": 0.1662980020046234,
+      "learning_rate": 4.9921567481796585e-05,
+      "loss": 0.0994,
+      "step": 580
+    },
+    {
+      "epoch": 0.3184973458554512,
+      "grad_norm": 0.1845940202474594,
+      "learning_rate": 4.9911463011640195e-05,
+      "loss": 0.0976,
+      "step": 585
+    },
+    {
+      "epoch": 0.3212195453926773,
+      "grad_norm": 0.17079226672649384,
+      "learning_rate": 4.9900747639208044e-05,
+      "loss": 0.0947,
+      "step": 590
+    },
+    {
+      "epoch": 0.32394174492990335,
+      "grad_norm": 0.1642676740884781,
+      "learning_rate": 4.9889421627272575e-05,
+      "loss": 0.1039,
+      "step": 595
+    },
+    {
+      "epoch": 0.32666394446712943,
+      "grad_norm": 0.20938318967819214,
+      "learning_rate": 4.987748525358087e-05,
+      "loss": 0.1005,
+      "step": 600
+    },
+    {
+      "epoch": 0.3293861440043555,
+      "grad_norm": 0.17017363011837006,
+      "learning_rate": 4.9864938810847884e-05,
+      "loss": 0.0941,
+      "step": 605
+    },
+    {
+      "epoch": 0.3321083435415816,
+      "grad_norm": 0.17874999344348907,
+      "learning_rate": 4.985178260674927e-05,
+      "loss": 0.1055,
+      "step": 610
+    },
+    {
+      "epoch": 0.33483054307880766,
+      "grad_norm": 0.16853144764900208,
+      "learning_rate": 4.98380169639138e-05,
+      "loss": 0.0941,
+      "step": 615
+    },
+    {
+      "epoch": 0.33755274261603374,
+      "grad_norm": 0.18321168422698975,
+      "learning_rate": 4.98236422199155e-05,
+      "loss": 0.0991,
+      "step": 620
+    },
+    {
+      "epoch": 0.3402749421532598,
+      "grad_norm": 0.16565477848052979,
+      "learning_rate": 4.980865872726532e-05,
+      "loss": 0.0979,
+      "step": 625
+    },
+    {
+      "epoch": 0.3429971416904859,
+      "grad_norm": 0.19292525947093964,
+      "learning_rate": 4.9793066853402536e-05,
+      "loss": 0.0957,
+      "step": 630
+    },
+    {
+      "epoch": 0.345719341227712,
+      "grad_norm": 0.1700926125049591,
+      "learning_rate": 4.977686698068572e-05,
+      "loss": 0.0889,
+      "step": 635
+    },
+    {
+      "epoch": 0.34844154076493805,
+      "grad_norm": 0.15266568958759308,
+      "learning_rate": 4.976005950638334e-05,
+      "loss": 0.0928,
+      "step": 640
+    },
+    {
+      "epoch": 0.35116374030216413,
+      "grad_norm": 0.15712082386016846,
+      "learning_rate": 4.974264484266406e-05,
+      "loss": 0.0927,
+      "step": 645
+    },
+    {
+      "epoch": 0.3538859398393902,
+      "grad_norm": 0.15186965465545654,
+      "learning_rate": 4.972462341658661e-05,
+      "loss": 0.0944,
+      "step": 650
+    },
+    {
+      "epoch": 0.3566081393766163,
+      "grad_norm": 0.18069899082183838,
+      "learning_rate": 4.970599567008931e-05,
+      "loss": 0.0998,
+      "step": 655
+    },
+    {
+      "epoch": 0.35933033891384236,
+      "grad_norm": 0.14533384144306183,
+      "learning_rate": 4.968676205997925e-05,
+      "loss": 0.0981,
+      "step": 660
+    },
+    {
+      "epoch": 0.36205253845106844,
+      "grad_norm": 0.16364873945713043,
+      "learning_rate": 4.966692305792106e-05,
+      "loss": 0.0932,
+      "step": 665
+    },
+    {
+      "epoch": 0.3647747379882945,
+      "grad_norm": 0.18472006916999817,
+      "learning_rate": 4.9646479150425376e-05,
+      "loss": 0.0866,
+      "step": 670
+    },
+    {
+      "epoch": 0.3674969375255206,
+      "grad_norm": 0.15847539901733398,
+      "learning_rate": 4.962543083883687e-05,
+      "loss": 0.0837,
+      "step": 675
+    },
+    {
+      "epoch": 0.3702191370627467,
+      "grad_norm": 0.155991330742836,
+      "learning_rate": 4.9603778639322004e-05,
+      "loss": 0.0906,
+      "step": 680
+    },
+    {
+      "epoch": 0.37294133659997275,
+      "grad_norm": 0.16835445165634155,
+      "learning_rate": 4.958152308285633e-05,
+      "loss": 0.0949,
+      "step": 685
+    },
+    {
+      "epoch": 0.3756635361371989,
+      "grad_norm": 0.18460558354854584,
+      "learning_rate": 4.95586647152115e-05,
+      "loss": 0.0931,
+      "step": 690
+    },
+    {
+      "epoch": 0.37838573567442496,
+      "grad_norm": 0.17255567014217377,
+      "learning_rate": 4.9535204096941854e-05,
+      "loss": 0.087,
+      "step": 695
+    },
+    {
+      "epoch": 0.38110793521165104,
+      "grad_norm": 0.15889868140220642,
+      "learning_rate": 4.951114180337069e-05,
+      "loss": 0.0965,
+      "step": 700
+    },
+    {
+      "epoch": 0.3838301347488771,
+      "grad_norm": 0.1403263658285141,
+      "learning_rate": 4.948647842457615e-05,
+      "loss": 0.0862,
+      "step": 705
+    },
+    {
+      "epoch": 0.3865523342861032,
+      "grad_norm": 0.16774417459964752,
+      "learning_rate": 4.946121456537676e-05,
+      "loss": 0.0911,
+      "step": 710
+    },
+    {
+      "epoch": 0.38927453382332927,
+      "grad_norm": 0.16498145461082458,
+      "learning_rate": 4.9435350845316575e-05,
+      "loss": 0.0851,
+      "step": 715
+    },
+    {
+      "epoch": 0.39199673336055535,
+      "grad_norm": 0.15889093279838562,
+      "learning_rate": 4.9408887898650036e-05,
+      "loss": 0.0836,
+      "step": 720
+    },
+    {
+      "epoch": 0.3947189328977814,
+      "grad_norm": 0.14469870924949646,
+      "learning_rate": 4.9381826374326336e-05,
+      "loss": 0.1033,
+      "step": 725
+    },
+    {
+      "epoch": 0.3974411324350075,
+      "grad_norm": 0.1625497043132782,
+      "learning_rate": 4.935416693597358e-05,
+      "loss": 0.0862,
+      "step": 730
+    },
+    {
+      "epoch": 0.4001633319722336,
+      "grad_norm": 0.15590594708919525,
+      "learning_rate": 4.932591026188247e-05,
+      "loss": 0.0946,
+      "step": 735
+    },
+    {
+      "epoch": 0.40288553150945966,
+      "grad_norm": 0.1706922948360443,
+      "learning_rate": 4.929705704498969e-05,
+      "loss": 0.0913,
+      "step": 740
+    },
+    {
+      "epoch": 0.40560773104668574,
+      "grad_norm": 0.15704989433288574,
+      "learning_rate": 4.9267607992860906e-05,
+      "loss": 0.0888,
+      "step": 745
+    },
+    {
+      "epoch": 0.4083299305839118,
+      "grad_norm": 0.14272324740886688,
+      "learning_rate": 4.9237563827673416e-05,
+      "loss": 0.0845,
+      "step": 750
+    },
+    {
+      "epoch": 0.4110521301211379,
+      "grad_norm": 0.1592000275850296,
+      "learning_rate": 4.9206925286198426e-05,
+      "loss": 0.09,
+      "step": 755
+    },
+    {
+      "epoch": 0.41377432965836397,
+      "grad_norm": 0.16091205179691315,
+      "learning_rate": 4.9175693119783013e-05,
+      "loss": 0.0855,
+      "step": 760
+    },
+    {
+      "epoch": 0.41649652919559005,
+      "grad_norm": 0.13851270079612732,
+      "learning_rate": 4.914386809433167e-05,
+      "loss": 0.0946,
+      "step": 765
+    },
+    {
+      "epoch": 0.4192187287328161,
+      "grad_norm": 0.15238268673419952,
+      "learning_rate": 4.911145099028753e-05,
+      "loss": 0.0861,
+      "step": 770
+    },
+    {
+      "epoch": 0.4219409282700422,
+      "grad_norm": 0.15689148008823395,
+      "learning_rate": 4.9078442602613265e-05,
+      "loss": 0.088,
+      "step": 775
+    },
+    {
+      "epoch": 0.4246631278072683,
+      "grad_norm": 0.17214879393577576,
+      "learning_rate": 4.9044843740771505e-05,
+      "loss": 0.0882,
+      "step": 780
+    },
+    {
+      "epoch": 0.42738532734449436,
+      "grad_norm": 0.15415912866592407,
+      "learning_rate": 4.901065522870511e-05,
+      "loss": 0.0978,
+      "step": 785
+    },
+    {
+      "epoch": 0.43010752688172044,
+      "grad_norm": 0.165154829621315,
+      "learning_rate": 4.897587790481683e-05,
+      "loss": 0.0833,
+      "step": 790
+    },
+    {
+      "epoch": 0.4328297264189465,
+      "grad_norm": 0.15241703391075134,
+      "learning_rate": 4.894051262194885e-05,
+      "loss": 0.0944,
+      "step": 795
+    },
+    {
+      "epoch": 0.4355519259561726,
+      "grad_norm": 0.13514836132526398,
+      "learning_rate": 4.8904560247361833e-05,
+      "loss": 0.1036,
+      "step": 800
+    },
+    {
+      "epoch": 0.43827412549339867,
+      "grad_norm": 0.14687085151672363,
+      "learning_rate": 4.886802166271364e-05,
+      "loss": 0.1042,
+      "step": 805
+    },
+    {
+      "epoch": 0.44099632503062475,
+      "grad_norm": 0.143589586019516,
+      "learning_rate": 4.8830897764037744e-05,
+      "loss": 0.0771,
+      "step": 810
+    },
+    {
+      "epoch": 0.4437185245678508,
+      "grad_norm": 0.16582971811294556,
+      "learning_rate": 4.879318946172123e-05,
+      "loss": 0.0992,
+      "step": 815
+    },
+    {
+      "epoch": 0.4464407241050769,
+      "grad_norm": 0.1646450310945511,
+      "learning_rate": 4.875489768048247e-05,
+      "loss": 0.0959,
+      "step": 820
+    },
+    {
+      "epoch": 0.449162923642303,
+      "grad_norm": 0.12446028739213943,
+      "learning_rate": 4.871602335934847e-05,
+      "loss": 0.0817,
+      "step": 825
+    },
+    {
+      "epoch": 0.45188512317952906,
+      "grad_norm": 0.15374915301799774,
+      "learning_rate": 4.867656745163182e-05,
+      "loss": 0.0763,
+      "step": 830
+    },
+    {
+      "epoch": 0.45460732271675514,
+      "grad_norm": 0.1652667224407196,
+      "learning_rate": 4.8636530924907296e-05,
+      "loss": 0.1026,
+      "step": 835
+    },
+    {
+      "epoch": 0.4573295222539812,
+      "grad_norm": 0.13503149151802063,
+      "learning_rate": 4.85959147609882e-05,
+      "loss": 0.0986,
+      "step": 840
+    },
+    {
+      "epoch": 0.4600517217912073,
+      "grad_norm": 0.16977562010288239,
+      "learning_rate": 4.855471995590222e-05,
+      "loss": 0.1099,
+      "step": 845
+    },
+    {
+      "epoch": 0.46277392132843337,
+      "grad_norm": 0.14991120994091034,
+      "learning_rate": 4.851294751986702e-05,
+      "loss": 0.0839,
+      "step": 850
+    },
+    {
+      "epoch": 0.46549612086565945,
+      "grad_norm": 0.15737441182136536,
+      "learning_rate": 4.84705984772655e-05,
+      "loss": 0.097,
+      "step": 855
+    },
+    {
+      "epoch": 0.4682183204028855,
+      "grad_norm": 0.14336347579956055,
+      "learning_rate": 4.8427673866620615e-05,
+      "loss": 0.0837,
+      "step": 860
+    },
+    {
+      "epoch": 0.4709405199401116,
+      "grad_norm": 0.12957650423049927,
+      "learning_rate": 4.8384174740569944e-05,
+      "loss": 0.0871,
+      "step": 865
+    },
+    {
+      "epoch": 0.4736627194773377,
+      "grad_norm": 0.15204955637454987,
+      "learning_rate": 4.83401021658399e-05,
+      "loss": 0.0905,
+      "step": 870
+    },
+    {
+      "epoch": 0.47638491901456376,
+      "grad_norm": 0.1311759650707245,
+      "learning_rate": 4.82954572232195e-05,
+      "loss": 0.1187,
+      "step": 875
+    },
+    {
+      "epoch": 0.47910711855178983,
+      "grad_norm": 0.14466483891010284,
+      "learning_rate": 4.825024100753395e-05,
+      "loss": 0.0892,
+      "step": 880
+    },
+    {
+      "epoch": 0.4818293180890159,
+      "grad_norm": 0.14845815300941467,
+      "learning_rate": 4.820445462761771e-05,
+      "loss": 0.0865,
+      "step": 885
+    },
+    {
+      "epoch": 0.484551517626242,
+      "grad_norm": 0.15656766295433044,
+      "learning_rate": 4.815809920628738e-05,
+      "loss": 0.0899,
+      "step": 890
+    },
+    {
+      "epoch": 0.48727371716346807,
+      "grad_norm": 0.11957818269729614,
+      "learning_rate": 4.8111175880314084e-05,
+      "loss": 0.0767,
+      "step": 895
+    },
+    {
+      "epoch": 0.48999591670069415,
+      "grad_norm": 0.14076441526412964,
+      "learning_rate": 4.806368580039566e-05,
+      "loss": 0.0832,
+      "step": 900
+    },
+    {
+      "epoch": 0.4927181162379202,
+      "grad_norm": 0.15563461184501648,
+      "learning_rate": 4.801563013112844e-05,
+      "loss": 0.0873,
+      "step": 905
+    },
+    {
+      "epoch": 0.4954403157751463,
+      "grad_norm": 0.13865019381046295,
+      "learning_rate": 4.7967010050978635e-05,
+      "loss": 0.1083,
+      "step": 910
+    },
+    {
+      "epoch": 0.4981625153123724,
+      "grad_norm": 0.1487104445695877,
+      "learning_rate": 4.791782675225348e-05,
+      "loss": 0.0853,
+      "step": 915
+    },
+    {
+      "epoch": 0.5008847148495985,
+      "grad_norm": 0.14385932683944702,
+      "learning_rate": 4.7868081441071975e-05,
+      "loss": 0.0882,
+      "step": 920
+    },
+    {
+      "epoch": 0.5036069143868246,
+      "grad_norm": 0.144570991396904,
+      "learning_rate": 4.781777533733534e-05,
+      "loss": 0.0859,
+      "step": 925
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 0.1305367350578308,
+      "learning_rate": 4.776690967469708e-05,
+      "loss": 0.0788,
+      "step": 930
+    },
+    {
+      "epoch": 0.5090513134612767,
+      "grad_norm": 0.14406158030033112,
+      "learning_rate": 4.771548570053268e-05,
+      "loss": 0.075,
+      "step": 935
+    },
+    {
+      "epoch": 0.5117735129985028,
+      "grad_norm": 0.14757996797561646,
+      "learning_rate": 4.766350467590911e-05,
+      "loss": 0.0991,
+      "step": 940
+    },
+    {
+      "epoch": 0.5144957125357289,
+      "grad_norm": 0.13629554212093353,
+      "learning_rate": 4.7610967875553846e-05,
+      "loss": 0.082,
+      "step": 945
+    },
+    {
+      "epoch": 0.517217912072955,
+      "grad_norm": 0.1259639412164688,
+      "learning_rate": 4.755787658782361e-05,
+      "loss": 0.0848,
+      "step": 950
+    },
+    {
+      "epoch": 0.519940111610181,
+      "grad_norm": 0.14720089733600616,
+      "learning_rate": 4.750423211467278e-05,
+      "loss": 0.083,
+      "step": 955
+    },
+    {
+      "epoch": 0.5226623111474071,
+      "grad_norm": 0.13509656488895416,
+      "learning_rate": 4.745003577162148e-05,
+      "loss": 0.0856,
+      "step": 960
+    },
+    {
+      "epoch": 0.5253845106846332,
+      "grad_norm": 0.15491896867752075,
+      "learning_rate": 4.7395288887723296e-05,
+      "loss": 0.0953,
+      "step": 965
+    },
+    {
+      "epoch": 0.5281067102218593,
+      "grad_norm": 0.14023731648921967,
+      "learning_rate": 4.73399928055327e-05,
+      "loss": 0.0887,
+      "step": 970
+    },
+    {
+      "epoch": 0.5308289097590854,
+      "grad_norm": 0.13751854002475739,
+      "learning_rate": 4.728414888107211e-05,
+      "loss": 0.0908,
+      "step": 975
+    },
+    {
+      "epoch": 0.5335511092963114,
+      "grad_norm": 0.16226938366889954,
+      "learning_rate": 4.722775848379866e-05,
+      "loss": 0.0898,
+      "step": 980
+    },
+    {
+      "epoch": 0.5362733088335375,
+      "grad_norm": 0.12918660044670105,
+      "learning_rate": 4.717082299657058e-05,
+      "loss": 0.0985,
+      "step": 985
+    },
+    {
+      "epoch": 0.5389955083707636,
+      "grad_norm": 0.14950741827487946,
+      "learning_rate": 4.711334381561333e-05,
+      "loss": 0.0894,
+      "step": 990
+    },
+    {
+      "epoch": 0.5417177079079897,
+      "grad_norm": 0.158179372549057,
+      "learning_rate": 4.7055322350485344e-05,
+      "loss": 0.0823,
+      "step": 995
+    },
+    {
+      "epoch": 0.5444399074452158,
+      "grad_norm": 0.14156126976013184,
+      "learning_rate": 4.699676002404342e-05,
+      "loss": 0.0851,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5444399074452158,
+      "eval_loss": 0.08648888021707535,
+      "eval_runtime": 254.9955,
+      "eval_samples_per_second": 2.353,
+      "eval_steps_per_second": 0.588,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5471621069824418,
+      "grad_norm": 0.14035391807556152,
+      "learning_rate": 4.6937658272407905e-05,
+      "loss": 0.0827,
+      "step": 1005
+    },
+    {
+      "epoch": 0.5498843065196679,
+      "grad_norm": 0.1346469521522522,
+      "learning_rate": 4.6878018544927415e-05,
+      "loss": 0.0986,
+      "step": 1010
+    },
+    {
+      "epoch": 0.552606506056894,
+      "grad_norm": 0.16363918781280518,
+      "learning_rate": 4.681784230414332e-05,
+      "loss": 0.0883,
+      "step": 1015
+    },
+    {
+      "epoch": 0.5553287055941201,
+      "grad_norm": 0.14541316032409668,
+      "learning_rate": 4.6757131025753886e-05,
+      "loss": 0.0881,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5580509051313461,
+      "grad_norm": 0.1540316641330719,
+      "learning_rate": 4.6695886198578034e-05,
+      "loss": 0.0825,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5607731046685722,
+      "grad_norm": 0.15897440910339355,
+      "learning_rate": 4.6634109324518914e-05,
+      "loss": 0.0978,
+      "step": 1030
+    },
+    {
+      "epoch": 0.5634953042057983,
+      "grad_norm": 0.13335077464580536,
+      "learning_rate": 4.657180191852701e-05,
+      "loss": 0.1565,
+      "step": 1035
+    },
+    {
+      "epoch": 0.5662175037430244,
+      "grad_norm": 0.1566026508808136,
+      "learning_rate": 4.6508965508562995e-05,
+      "loss": 0.0927,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5689397032802505,
+      "grad_norm": 0.1556750237941742,
+      "learning_rate": 4.644560163556031e-05,
+      "loss": 0.0839,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5716619028174765,
+      "grad_norm": 0.15351781249046326,
+      "learning_rate": 4.638171185338729e-05,
+      "loss": 0.0812,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5743841023547026,
+      "grad_norm": 0.16494281589984894,
+      "learning_rate": 4.6317297728809147e-05,
+      "loss": 0.0889,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5771063018919287,
+      "grad_norm": 0.15591537952423096,
+      "learning_rate": 4.62523608414495e-05,
+      "loss": 0.0858,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5798285014291548,
+      "grad_norm": 0.1396636664867401,
+      "learning_rate": 4.618690278375164e-05,
+      "loss": 0.0856,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5825507009663808,
+      "grad_norm": 0.1384498029947281,
+      "learning_rate": 4.61209251609395e-05,
+      "loss": 0.082,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5852729005036069,
+      "grad_norm": 0.13901008665561676,
+      "learning_rate": 4.605442959097826e-05,
+      "loss": 0.0821,
+      "step": 1075
+    },
+    {
+      "epoch": 0.587995100040833,
+      "grad_norm": 0.1609126627445221,
+      "learning_rate": 4.5987417704534697e-05,
+      "loss": 0.1273,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5907172995780591,
+      "grad_norm": 0.1475275307893753,
+      "learning_rate": 4.591989114493718e-05,
+      "loss": 0.0807,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5934394991152852,
+      "grad_norm": 0.14419402182102203,
+      "learning_rate": 4.5851851568135376e-05,
+      "loss": 0.0737,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5961616986525112,
+      "grad_norm": 0.13583903014659882,
+      "learning_rate": 4.5783300642659644e-05,
+      "loss": 0.0904,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5988838981897373,
+      "grad_norm": 0.1405334770679474,
+      "learning_rate": 4.571424004958012e-05,
+      "loss": 0.0863,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6016060977269634,
+      "grad_norm": 0.1326403170824051,
+      "learning_rate": 4.564467148246548e-05,
+      "loss": 0.0839,
+      "step": 1105
+    },
+    {
+      "epoch": 0.6043282972641895,
+      "grad_norm": 0.14316369593143463,
+      "learning_rate": 4.557459664734141e-05,
+      "loss": 0.0862,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6070504968014155,
+      "grad_norm": 0.14497336745262146,
+      "learning_rate": 4.550401726264879e-05,
+      "loss": 0.0733,
+      "step": 1115
+    },
+    {
+      "epoch": 0.6097726963386416,
+      "grad_norm": 0.15112735331058502,
+      "learning_rate": 4.5432935059201544e-05,
+      "loss": 0.0712,
+      "step": 1120
+    },
+    {
+      "epoch": 0.6124948958758677,
+      "grad_norm": 0.1457735002040863,
+      "learning_rate": 4.536135178014415e-05,
+      "loss": 0.0808,
+      "step": 1125
+    },
+    {
+      "epoch": 0.6152170954130938,
+      "grad_norm": 0.14395588636398315,
+      "learning_rate": 4.528926918090898e-05,
+      "loss": 0.0863,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6179392949503199,
+      "grad_norm": 0.13603851199150085,
+      "learning_rate": 4.521668902917317e-05,
+      "loss": 0.0848,
+      "step": 1135
+    },
+    {
+      "epoch": 0.6206614944875459,
+      "grad_norm": 0.16027645766735077,
+      "learning_rate": 4.514361310481533e-05,
+      "loss": 0.0929,
+      "step": 1140
+    },
+    {
+      "epoch": 0.623383694024772,
+      "grad_norm": 0.13826999068260193,
+      "learning_rate": 4.507004319987185e-05,
+      "loss": 0.0914,
+      "step": 1145
+    },
+    {
+      "epoch": 0.6261058935619981,
+      "grad_norm": 0.14140823483467102,
+      "learning_rate": 4.499598111849299e-05,
+      "loss": 0.0885,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6288280930992242,
+      "grad_norm": 0.13926826417446136,
+      "learning_rate": 4.492142867689861e-05,
+      "loss": 0.083,
+      "step": 1155
+    },
+    {
+      "epoch": 0.6315502926364502,
+      "grad_norm": 0.148858904838562,
+      "learning_rate": 4.484638770333367e-05,
+      "loss": 0.0866,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6342724921736763,
+      "grad_norm": 0.12866345047950745,
+      "learning_rate": 4.4770860038023335e-05,
+      "loss": 0.0991,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6369946917109024,
+      "grad_norm": 0.13434378802776337,
+      "learning_rate": 4.4694847533127903e-05,
+      "loss": 0.079,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6397168912481285,
+      "grad_norm": 0.13786840438842773,
+      "learning_rate": 4.461835205269736e-05,
+      "loss": 0.0842,
+      "step": 1175
+    },
+    {
+      "epoch": 0.6424390907853545,
+      "grad_norm": 0.12938237190246582,
+      "learning_rate": 4.454137547262566e-05,
+      "loss": 0.0771,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 0.13730603456497192,
+      "learning_rate": 4.446391968060475e-05,
+      "loss": 0.0831,
+      "step": 1185
+    },
+    {
+      "epoch": 0.6478834898598067,
+      "grad_norm": 0.142304465174675,
+      "learning_rate": 4.4385986576078254e-05,
+      "loss": 0.0918,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6506056893970328,
+      "grad_norm": 0.1353188157081604,
+      "learning_rate": 4.43075780701949e-05,
+      "loss": 0.0897,
+      "step": 1195
+    },
+    {
+      "epoch": 0.6533278889342589,
+      "grad_norm": 0.1252220869064331,
+      "learning_rate": 4.422869608576167e-05,
+      "loss": 0.1353,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6560500884714849,
+      "grad_norm": 0.14703655242919922,
+      "learning_rate": 4.4149342557196605e-05,
+      "loss": 0.2083,
+      "step": 1205
+    },
+    {
+      "epoch": 0.658772288008711,
+      "grad_norm": 0.13858124613761902,
+      "learning_rate": 4.406951943048141e-05,
+      "loss": 0.0824,
+      "step": 1210
+    },
+    {
+      "epoch": 0.6614944875459371,
+      "grad_norm": 0.1428638994693756,
+      "learning_rate": 4.3989228663113714e-05,
+      "loss": 0.0774,
+      "step": 1215
+    },
+    {
+      "epoch": 0.6642166870831632,
+      "grad_norm": 0.13073915243148804,
+      "learning_rate": 4.3908472224059064e-05,
+      "loss": 0.0788,
+      "step": 1220
+    },
+    {
+      "epoch": 0.6669388866203892,
+      "grad_norm": 0.14101268351078033,
+      "learning_rate": 4.3827252093702656e-05,
+      "loss": 0.0882,
+      "step": 1225
+    },
+    {
+      "epoch": 0.6696610861576153,
+      "grad_norm": 0.13420827686786652,
+      "learning_rate": 4.374557026380075e-05,
+      "loss": 0.0783,
+      "step": 1230
+    },
+    {
+      "epoch": 0.6723832856948414,
+      "grad_norm": 0.12414020299911499,
+      "learning_rate": 4.366342873743185e-05,
+      "loss": 0.0874,
+      "step": 1235
+    },
+    {
+      "epoch": 0.6751054852320675,
+      "grad_norm": 0.12837223708629608,
+      "learning_rate": 4.358082952894753e-05,
+      "loss": 0.0863,
+      "step": 1240
+    },
+    {
+      "epoch": 0.6778276847692936,
+      "grad_norm": 0.14144425094127655,
+      "learning_rate": 4.349777466392313e-05,
+      "loss": 0.0784,
+      "step": 1245
+    },
+    {
+      "epoch": 0.6805498843065196,
+      "grad_norm": 0.13347384333610535,
+      "learning_rate": 4.341426617910798e-05,
+      "loss": 0.0823,
+      "step": 1250
+    },
+    {
+      "epoch": 0.6832720838437457,
+      "grad_norm": 0.1294572502374649,
+      "learning_rate": 4.3330306122375516e-05,
+      "loss": 0.0787,
+      "step": 1255
+    },
+    {
+      "epoch": 0.6859942833809718,
+      "grad_norm": 0.13768264651298523,
+      "learning_rate": 4.324589655267306e-05,
+      "loss": 0.0812,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6887164829181979,
+      "grad_norm": 0.14431719481945038,
+      "learning_rate": 4.3161039539971295e-05,
+      "loss": 0.0917,
+      "step": 1265
+    },
+    {
+      "epoch": 0.691438682455424,
+      "grad_norm": 0.13428707420825958,
+      "learning_rate": 4.307573716521353e-05,
+      "loss": 0.0866,
+      "step": 1270
+    },
+    {
+      "epoch": 0.69416088199265,
+      "grad_norm": 0.14333197474479675,
+      "learning_rate": 4.298999152026465e-05,
+      "loss": 0.0822,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6968830815298761,
+      "grad_norm": 0.13545657694339752,
+      "learning_rate": 4.2903804707859835e-05,
+      "loss": 0.1482,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6996052810671022,
+      "grad_norm": 0.13254009187221527,
+      "learning_rate": 4.281717884155298e-05,
+      "loss": 0.0792,
+      "step": 1285
+    },
+    {
+      "epoch": 0.7023274806043283,
+      "grad_norm": 0.12849317491054535,
+      "learning_rate": 4.2730116045664905e-05,
+      "loss": 0.0909,
+      "step": 1290
+    },
+    {
+      "epoch": 0.7050496801415543,
+      "grad_norm": 0.15678727626800537,
+      "learning_rate": 4.264261845523116e-05,
+      "loss": 0.0902,
+      "step": 1295
+    },
+    {
+      "epoch": 0.7077718796787804,
+      "grad_norm": 0.1492096483707428,
+      "learning_rate": 4.255468821594981e-05,
+      "loss": 0.0796,
+      "step": 1300
+    },
+    {
+      "epoch": 0.7104940792160065,
+      "grad_norm": 0.12939919531345367,
+      "learning_rate": 4.2466327484128685e-05,
+      "loss": 0.0929,
+      "step": 1305
+    },
+    {
+      "epoch": 0.7132162787532326,
+      "grad_norm": 0.13585573434829712,
+      "learning_rate": 4.2377538426632595e-05,
+      "loss": 0.0997,
+      "step": 1310
+    },
+    {
+      "epoch": 0.7159384782904586,
+      "grad_norm": 0.15193824470043182,
+      "learning_rate": 4.228832322083013e-05,
+      "loss": 0.0806,
+      "step": 1315
+    },
+    {
+      "epoch": 0.7186606778276847,
+      "grad_norm": 0.1575016975402832,
+      "learning_rate": 4.2198684054540285e-05,
+      "loss": 0.0877,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7213828773649108,
+      "grad_norm": 0.12813854217529297,
+      "learning_rate": 4.210862312597884e-05,
+      "loss": 0.0802,
+      "step": 1325
+    },
+    {
+      "epoch": 0.7241050769021369,
+      "grad_norm": 0.13729339838027954,
+      "learning_rate": 4.201814264370441e-05,
+      "loss": 0.0816,
+      "step": 1330
+    },
+    {
+      "epoch": 0.726827276439363,
+      "grad_norm": 0.12651780247688293,
+      "learning_rate": 4.192724482656428e-05,
+      "loss": 0.0812,
+      "step": 1335
+    },
+    {
+      "epoch": 0.729549475976589,
+      "grad_norm": 0.13558164238929749,
+      "learning_rate": 4.1835931903640046e-05,
+      "loss": 0.0926,
+      "step": 1340
+    },
+    {
+      "epoch": 0.7322716755138151,
+      "grad_norm": 0.1335664838552475,
+      "learning_rate": 4.17442061141929e-05,
+      "loss": 0.078,
+      "step": 1345
+    },
+    {
+      "epoch": 0.7349938750510412,
+      "grad_norm": 0.12852634489536285,
+      "learning_rate": 4.165206970760874e-05,
+      "loss": 0.0805,
+      "step": 1350
+    },
+    {
+      "epoch": 0.7377160745882673,
+      "grad_norm": 0.14529520273208618,
+      "learning_rate": 4.1559524943342985e-05,
+      "loss": 0.0837,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7404382741254933,
+      "grad_norm": 0.14452920854091644,
+      "learning_rate": 4.1466574090865225e-05,
+      "loss": 0.0857,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7431604736627194,
+      "grad_norm": 0.13917720317840576,
+      "learning_rate": 4.1373219429603473e-05,
+      "loss": 0.0814,
+      "step": 1365
+    },
+    {
+      "epoch": 0.7458826731999455,
+      "grad_norm": 0.11816851049661636,
+      "learning_rate": 4.127946324888836e-05,
+      "loss": 0.0743,
+      "step": 1370
+    },
+    {
+      "epoch": 0.7486048727371717,
+      "grad_norm": 0.14422592520713806,
+      "learning_rate": 4.118530784789694e-05,
+      "loss": 0.0768,
+      "step": 1375
+    },
+    {
+      "epoch": 0.7513270722743978,
+      "grad_norm": 0.14289607107639313,
+      "learning_rate": 4.109075553559633e-05,
+      "loss": 0.0806,
+      "step": 1380
+    },
+    {
+      "epoch": 0.7540492718116238,
+      "grad_norm": 0.14432035386562347,
+      "learning_rate": 4.099580863068706e-05,
+      "loss": 0.0915,
+      "step": 1385
+    },
+    {
+      "epoch": 0.7567714713488499,
+      "grad_norm": 0.16059063374996185,
+      "learning_rate": 4.0900469461546235e-05,
+      "loss": 0.0799,
+      "step": 1390
+    },
+    {
+      "epoch": 0.759493670886076,
+      "grad_norm": 0.14347511529922485,
+      "learning_rate": 4.0804740366170454e-05,
+      "loss": 0.0899,
+      "step": 1395
+    },
+    {
+      "epoch": 0.7622158704233021,
+      "grad_norm": 0.12717047333717346,
+      "learning_rate": 4.070862369211843e-05,
+      "loss": 0.0827,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7649380699605282,
+      "grad_norm": 0.12884029746055603,
+      "learning_rate": 4.061212179645345e-05,
+      "loss": 0.0856,
+      "step": 1405
+    },
+    {
+      "epoch": 0.7676602694977542,
+      "grad_norm": 0.14441439509391785,
+      "learning_rate": 4.051523704568557e-05,
+      "loss": 0.1416,
+      "step": 1410
+    },
+    {
+      "epoch": 0.7703824690349803,
+      "grad_norm": 0.12841658294200897,
+      "learning_rate": 4.0417971815713584e-05,
+      "loss": 0.0827,
+      "step": 1415
+    },
+    {
+      "epoch": 0.7731046685722064,
+      "grad_norm": 0.14411351084709167,
+      "learning_rate": 4.032032849176673e-05,
+      "loss": 0.0808,
+      "step": 1420
+    },
+    {
+      "epoch": 0.7758268681094325,
+      "grad_norm": 0.13570953905582428,
+      "learning_rate": 4.022230946834624e-05,
+      "loss": 0.08,
+      "step": 1425
+    },
+    {
+      "epoch": 0.7785490676466585,
+      "grad_norm": 0.13861487805843353,
+      "learning_rate": 4.012391714916661e-05,
+      "loss": 0.0867,
+      "step": 1430
+    },
+    {
+      "epoch": 0.7812712671838846,
+      "grad_norm": 0.14143118262290955,
+      "learning_rate": 4.0025153947096624e-05,
+      "loss": 0.0865,
+      "step": 1435
+    },
+    {
+      "epoch": 0.7839934667211107,
+      "grad_norm": 0.1405879408121109,
+      "learning_rate": 3.992602228410023e-05,
+      "loss": 0.0983,
+      "step": 1440
+    },
+    {
+      "epoch": 0.7867156662583368,
+      "grad_norm": 0.14435367286205292,
+      "learning_rate": 3.982652459117707e-05,
+      "loss": 0.0829,
+      "step": 1445
+    },
+    {
+      "epoch": 0.7894378657955629,
+      "grad_norm": 0.14033889770507812,
+      "learning_rate": 3.972666330830299e-05,
+      "loss": 0.0903,
+      "step": 1450
+    },
+    {
+      "epoch": 0.7921600653327889,
+      "grad_norm": 0.13285259902477264,
+      "learning_rate": 3.9626440884370065e-05,
+      "loss": 0.0748,
+      "step": 1455
+    },
+    {
+      "epoch": 0.794882264870015,
+      "grad_norm": 0.1237478256225586,
+      "learning_rate": 3.952585977712664e-05,
+      "loss": 0.083,
+      "step": 1460
+    },
+    {
+      "epoch": 0.7976044644072411,
+      "grad_norm": 0.14227712154388428,
+      "learning_rate": 3.942492245311703e-05,
+      "loss": 0.0926,
+      "step": 1465
+    },
+    {
+      "epoch": 0.8003266639444672,
+      "grad_norm": 0.13259918987751007,
+      "learning_rate": 3.9323631387621015e-05,
+      "loss": 0.0841,
+      "step": 1470
+    },
+    {
+      "epoch": 0.8030488634816932,
+      "grad_norm": 0.12796998023986816,
+      "learning_rate": 3.9221989064593175e-05,
+      "loss": 0.076,
+      "step": 1475
+    },
+    {
+      "epoch": 0.8057710630189193,
+      "grad_norm": 0.13424307107925415,
+      "learning_rate": 3.9119997976601954e-05,
+      "loss": 0.097,
+      "step": 1480
+    },
+    {
+      "epoch": 0.8084932625561454,
+      "grad_norm": 0.12660135328769684,
+      "learning_rate": 3.9017660624768515e-05,
+      "loss": 0.0834,
+      "step": 1485
+    },
+    {
+      "epoch": 0.8112154620933715,
+      "grad_norm": 0.12413927167654037,
+      "learning_rate": 3.891497951870545e-05,
+      "loss": 0.0765,
+      "step": 1490
+    },
+    {
+      "epoch": 0.8139376616305976,
+      "grad_norm": 0.1326495110988617,
+      "learning_rate": 3.8811957176455214e-05,
+      "loss": 0.0832,
+      "step": 1495
+    },
+    {
+      "epoch": 0.8166598611678236,
+      "grad_norm": 0.133337140083313,
+      "learning_rate": 3.870859612442837e-05,
+      "loss": 0.0819,
+      "step": 1500
+    },
+    {
+      "epoch": 0.8166598611678236,
+      "eval_loss": 0.08255165070295334,
+      "eval_runtime": 238.131,
+      "eval_samples_per_second": 2.52,
+      "eval_steps_per_second": 0.63,
+      "step": 1500
+    },
+    {
+      "epoch": 0.8193820607050497,
+      "grad_norm": 0.14871715009212494,
+      "learning_rate": 3.8604898897341644e-05,
+      "loss": 0.1878,
+      "step": 1505
+    },
+    {
+      "epoch": 0.8221042602422758,
+      "grad_norm": 0.1360073983669281,
+      "learning_rate": 3.850086803815576e-05,
+      "loss": 0.0879,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8248264597795019,
+      "grad_norm": 0.12314160168170929,
+      "learning_rate": 3.8396506098013076e-05,
+      "loss": 0.0767,
+      "step": 1515
+    },
+    {
+      "epoch": 0.8275486593167279,
+      "grad_norm": 0.14723584055900574,
+      "learning_rate": 3.829181563617503e-05,
+      "loss": 0.0972,
+      "step": 1520
+    },
+    {
+      "epoch": 0.830270858853954,
+      "grad_norm": 0.14206384122371674,
+      "learning_rate": 3.81867992199594e-05,
+      "loss": 0.0839,
+      "step": 1525
+    },
+    {
+      "epoch": 0.8329930583911801,
+      "grad_norm": 0.11804749816656113,
+      "learning_rate": 3.808145942467729e-05,
+      "loss": 0.0923,
+      "step": 1530
+    },
+    {
+      "epoch": 0.8357152579284062,
+      "grad_norm": 0.12313738465309143,
+      "learning_rate": 3.797579883357002e-05,
+      "loss": 0.0872,
+      "step": 1535
+    },
+    {
+      "epoch": 0.8384374574656323,
+      "grad_norm": 0.13978321850299835,
+      "learning_rate": 3.7869820037745776e-05,
+      "loss": 0.1243,
+      "step": 1540
+    },
+    {
+      "epoch": 0.8411596570028583,
+      "grad_norm": 0.12310109287500381,
+      "learning_rate": 3.776352563611604e-05,
+      "loss": 0.0839,
+      "step": 1545
+    },
+    {
+      "epoch": 0.8438818565400844,
+      "grad_norm": 0.13480614125728607,
+      "learning_rate": 3.765691823533191e-05,
+      "loss": 0.0873,
+      "step": 1550
+    },
+    {
+      "epoch": 0.8466040560773105,
+      "grad_norm": 0.13653361797332764,
+      "learning_rate": 3.75500004497201e-05,
+      "loss": 0.0771,
+      "step": 1555
+    },
+    {
+      "epoch": 0.8493262556145366,
+      "grad_norm": 0.1283886879682541,
+      "learning_rate": 3.74427749012189e-05,
+      "loss": 0.0816,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8520484551517626,
+      "grad_norm": 0.12764351069927216,
+      "learning_rate": 3.733524421931385e-05,
+      "loss": 0.0859,
+      "step": 1565
+    },
+    {
+      "epoch": 0.8547706546889887,
+      "grad_norm": 0.1281007081270218,
+      "learning_rate": 3.722741104097323e-05,
+      "loss": 0.0806,
+      "step": 1570
+    },
+    {
+      "epoch": 0.8574928542262148,
+      "grad_norm": 0.1311092972755432,
+      "learning_rate": 3.711927801058347e-05,
+      "loss": 0.0792,
+      "step": 1575
+    },
+    {
+      "epoch": 0.8602150537634409,
+      "grad_norm": 0.11376714706420898,
+      "learning_rate": 3.7010847779884204e-05,
+      "loss": 0.0763,
+      "step": 1580
+    },
+    {
+      "epoch": 0.862937253300667,
+      "grad_norm": 0.14167912304401398,
+      "learning_rate": 3.690212300790333e-05,
+      "loss": 0.0871,
+      "step": 1585
+    },
+    {
+      "epoch": 0.865659452837893,
+      "grad_norm": 0.13373591005802155,
+      "learning_rate": 3.679310636089174e-05,
+      "loss": 0.1189,
+      "step": 1590
+    },
+    {
+      "epoch": 0.8683816523751191,
+      "grad_norm": 0.14792831242084503,
+      "learning_rate": 3.668380051225794e-05,
+      "loss": 0.0811,
+      "step": 1595
+    },
+    {
+      "epoch": 0.8711038519123452,
+      "grad_norm": 0.14200051128864288,
+      "learning_rate": 3.657420814250258e-05,
+      "loss": 0.086,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8738260514495713,
+      "grad_norm": 0.1501314491033554,
+      "learning_rate": 3.646433193915257e-05,
+      "loss": 0.0885,
+      "step": 1605
+    },
+    {
+      "epoch": 0.8765482509867973,
+      "grad_norm": 0.10102449357509613,
+      "learning_rate": 3.635417459669532e-05,
+      "loss": 0.0738,
+      "step": 1610
+    },
+    {
+      "epoch": 0.8792704505240234,
+      "grad_norm": 0.12374872714281082,
+      "learning_rate": 3.624373881651254e-05,
+      "loss": 0.0883,
+      "step": 1615
+    },
+    {
+      "epoch": 0.8819926500612495,
+      "grad_norm": 0.1358879953622818,
+      "learning_rate": 3.6133027306814085e-05,
+      "loss": 0.0869,
+      "step": 1620
+    },
+    {
+      "epoch": 0.8847148495984756,
+      "grad_norm": 0.12947793304920197,
+      "learning_rate": 3.6022042782571494e-05,
+      "loss": 0.0868,
+      "step": 1625
+    },
+    {
+      "epoch": 0.8874370491357016,
+      "grad_norm": 0.14391177892684937,
+      "learning_rate": 3.591078796545144e-05,
+      "loss": 0.0829,
+      "step": 1630
+    },
+    {
+      "epoch": 0.8901592486729277,
+      "grad_norm": 0.13038809597492218,
+      "learning_rate": 3.579926558374897e-05,
+      "loss": 0.0888,
+      "step": 1635
+    },
+    {
+      "epoch": 0.8928814482101538,
+      "grad_norm": 0.15558017790317535,
+      "learning_rate": 3.5687478372320576e-05,
+      "loss": 0.0853,
+      "step": 1640
+    },
+    {
+      "epoch": 0.8956036477473799,
+      "grad_norm": 0.1262005716562271,
+      "learning_rate": 3.557542907251718e-05,
+      "loss": 0.0763,
+      "step": 1645
+    },
+    {
+      "epoch": 0.898325847284606,
+      "grad_norm": 0.1222100704908371,
+      "learning_rate": 3.546312043211687e-05,
+      "loss": 0.0835,
+      "step": 1650
+    },
+    {
+      "epoch": 0.901048046821832,
+      "grad_norm": 0.14368890225887299,
+      "learning_rate": 3.535055520525753e-05,
+      "loss": 0.0795,
+      "step": 1655
+    },
+    {
+      "epoch": 0.9037702463590581,
+      "grad_norm": 0.13927970826625824,
+      "learning_rate": 3.52377361523693e-05,
+      "loss": 0.0838,
+      "step": 1660
+    },
+    {
+      "epoch": 0.9064924458962842,
+      "grad_norm": 0.14974816143512726,
+      "learning_rate": 3.512466604010688e-05,
+      "loss": 0.0834,
+      "step": 1665
+    },
+    {
+      "epoch": 0.9092146454335103,
+      "grad_norm": 0.14425551891326904,
+      "learning_rate": 3.501134764128167e-05,
+      "loss": 0.0933,
+      "step": 1670
+    },
+    {
+      "epoch": 0.9119368449707363,
+      "grad_norm": 0.13006910681724548,
+      "learning_rate": 3.4897783734793794e-05,
+      "loss": 0.0829,
+      "step": 1675
+    },
+    {
+      "epoch": 0.9146590445079624,
+      "grad_norm": 0.10913355648517609,
+      "learning_rate": 3.478397710556397e-05,
+      "loss": 0.0775,
+      "step": 1680
+    },
+    {
+      "epoch": 0.9173812440451885,
+      "grad_norm": 0.13194523751735687,
+      "learning_rate": 3.466993054446515e-05,
+      "loss": 0.0762,
+      "step": 1685
+    },
+    {
+      "epoch": 0.9201034435824146,
+      "grad_norm": 0.1317387819290161,
+      "learning_rate": 3.4555646848254136e-05,
+      "loss": 0.094,
+      "step": 1690
+    },
+    {
+      "epoch": 0.9228256431196407,
+      "grad_norm": 0.1169929951429367,
+      "learning_rate": 3.4441128819503e-05,
+      "loss": 0.0839,
+      "step": 1695
+    },
+    {
+      "epoch": 0.9255478426568667,
+      "grad_norm": 0.13555140793323517,
+      "learning_rate": 3.4326379266530314e-05,
+      "loss": 0.0831,
+      "step": 1700
+    },
+    {
+      "epoch": 0.9282700421940928,
+      "grad_norm": 0.1173190027475357,
+      "learning_rate": 3.421140100333231e-05,
+      "loss": 0.0842,
+      "step": 1705
+    },
+    {
+      "epoch": 0.9309922417313189,
+      "grad_norm": 0.1393524408340454,
+      "learning_rate": 3.409619684951386e-05,
+      "loss": 0.1127,
+      "step": 1710
+    },
+    {
+      "epoch": 0.933714441268545,
+      "grad_norm": 0.12765948474407196,
+      "learning_rate": 3.3980769630219354e-05,
+      "loss": 0.0781,
+      "step": 1715
+    },
+    {
+      "epoch": 0.936436640805771,
+      "grad_norm": 0.11775423586368561,
+      "learning_rate": 3.386512217606339e-05,
+      "loss": 0.071,
+      "step": 1720
+    },
+    {
+      "epoch": 0.9391588403429971,
+      "grad_norm": 0.1447712928056717,
+      "learning_rate": 3.3749257323061376e-05,
+      "loss": 0.0852,
+      "step": 1725
+    },
+    {
+      "epoch": 0.9418810398802232,
+      "grad_norm": 0.14077867567539215,
+      "learning_rate": 3.3633177912559984e-05,
+      "loss": 0.0793,
+      "step": 1730
+    },
+    {
+      "epoch": 0.9446032394174493,
+      "grad_norm": 0.13743150234222412,
+      "learning_rate": 3.3516886791167444e-05,
+      "loss": 0.0804,
+      "step": 1735
+    },
+    {
+      "epoch": 0.9473254389546754,
+      "grad_norm": 0.13769537210464478,
+      "learning_rate": 3.34003868106838e-05,
+      "loss": 0.0865,
+      "step": 1740
+    },
+    {
+      "epoch": 0.9500476384919014,
+      "grad_norm": 0.1354909986257553,
+      "learning_rate": 3.328368082803088e-05,
+      "loss": 0.0793,
+      "step": 1745
+    },
+    {
+      "epoch": 0.9527698380291275,
+      "grad_norm": 0.12832538783550262,
+      "learning_rate": 3.316677170518235e-05,
+      "loss": 0.0769,
+      "step": 1750
+    },
+    {
+      "epoch": 0.9554920375663536,
+      "grad_norm": 0.1272445023059845,
+      "learning_rate": 3.304966230909342e-05,
+      "loss": 0.0931,
+      "step": 1755
+    },
+    {
+      "epoch": 0.9582142371035797,
+      "grad_norm": 0.14885401725769043,
+      "learning_rate": 3.293235551163063e-05,
+      "loss": 0.0835,
+      "step": 1760
+    },
+    {
+      "epoch": 0.9609364366408057,
+      "grad_norm": 0.13052543997764587,
+      "learning_rate": 3.2814854189501346e-05,
+      "loss": 0.0866,
+      "step": 1765
+    },
+    {
+      "epoch": 0.9636586361780318,
+      "grad_norm": 0.12114793807268143,
+      "learning_rate": 3.269716122418326e-05,
+      "loss": 0.0851,
+      "step": 1770
+    },
+    {
+      "epoch": 0.9663808357152579,
+      "grad_norm": 0.13619346916675568,
+      "learning_rate": 3.2579279501853744e-05,
+      "loss": 0.0834,
+      "step": 1775
+    },
+    {
+      "epoch": 0.969103035252484,
+      "grad_norm": 0.13156457245349884,
+      "learning_rate": 3.246121191331902e-05,
+      "loss": 0.0868,
+      "step": 1780
+    },
+    {
+      "epoch": 0.9718252347897101,
+      "grad_norm": 0.1379031538963318,
+      "learning_rate": 3.234296135394329e-05,
+      "loss": 0.081,
+      "step": 1785
+    },
+    {
+      "epoch": 0.9745474343269361,
+      "grad_norm": 0.1381215900182724,
+      "learning_rate": 3.222453072357777e-05,
+      "loss": 0.0758,
+      "step": 1790
+    },
+    {
+      "epoch": 0.9772696338641622,
+      "grad_norm": 0.11675938963890076,
+      "learning_rate": 3.2105922926489504e-05,
+      "loss": 0.1002,
+      "step": 1795
+    },
+    {
+      "epoch": 0.9799918334013883,
+      "grad_norm": 0.12285912781953812,
+      "learning_rate": 3.1987140871290236e-05,
+      "loss": 0.0736,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9827140329386144,
+      "grad_norm": 0.13735820353031158,
+      "learning_rate": 3.1868187470864984e-05,
+      "loss": 0.0779,
+      "step": 1805
+    },
+    {
+      "epoch": 0.9854362324758404,
+      "grad_norm": 0.13905146718025208,
+      "learning_rate": 3.1749065642300674e-05,
+      "loss": 0.0833,
+      "step": 1810
+    },
+    {
+      "epoch": 0.9881584320130665,
+      "grad_norm": 0.13085448741912842,
+      "learning_rate": 3.1629778306814585e-05,
+      "loss": 0.0859,
+      "step": 1815
+    },
+    {
+      "epoch": 0.9908806315502926,
+      "grad_norm": 0.13941799104213715,
+      "learning_rate": 3.151032838968271e-05,
+      "loss": 0.0845,
+      "step": 1820
+    },
+    {
+      "epoch": 0.9936028310875187,
+      "grad_norm": 0.12749053537845612,
+      "learning_rate": 3.139071882016802e-05,
+      "loss": 0.0935,
+      "step": 1825
+    },
+    {
+      "epoch": 0.9963250306247448,
+      "grad_norm": 0.12027429789304733,
+      "learning_rate": 3.127095253144864e-05,
+      "loss": 0.077,
+      "step": 1830
+    },
+    {
+      "epoch": 0.9990472301619708,
+      "grad_norm": 0.13250844180583954,
+      "learning_rate": 3.11510324605459e-05,
+      "loss": 0.0795,
+      "step": 1835
+    },
+    {
+      "epoch": 1.001769429699197,
+      "grad_norm": 0.12925724685192108,
+      "learning_rate": 3.103096154825233e-05,
+      "loss": 0.1021,
+      "step": 1840
+    },
+    {
+      "epoch": 1.004491629236423,
+      "grad_norm": 0.1351374089717865,
+      "learning_rate": 3.0910742739059527e-05,
+      "loss": 0.0833,
+      "step": 1845
+    },
+    {
+      "epoch": 1.0072138287736492,
+      "grad_norm": 0.13750001788139343,
+      "learning_rate": 3.0790378981085956e-05,
+      "loss": 0.076,
+      "step": 1850
+    },
+    {
+      "epoch": 1.0099360283108751,
+      "grad_norm": 0.1203581839799881,
+      "learning_rate": 3.0669873226004655e-05,
+      "loss": 0.0769,
+      "step": 1855
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 0.12651003897190094,
+      "learning_rate": 3.054922842897084e-05,
+      "loss": 0.0729,
+      "step": 1860
+    },
+    {
+      "epoch": 1.0153804273853273,
+      "grad_norm": 0.14429447054862976,
+      "learning_rate": 3.0428447548549467e-05,
+      "loss": 0.0772,
+      "step": 1865
+    },
+    {
+      "epoch": 1.0181026269225535,
+      "grad_norm": 0.13154245913028717,
+      "learning_rate": 3.030753354664262e-05,
+      "loss": 0.1681,
+      "step": 1870
+    },
+    {
+      "epoch": 1.0208248264597795,
+      "grad_norm": 0.11439166218042374,
+      "learning_rate": 3.018648938841695e-05,
+      "loss": 0.0753,
+      "step": 1875
+    },
+    {
+      "epoch": 1.0235470259970056,
+      "grad_norm": 0.12332076579332352,
+      "learning_rate": 3.00653180422309e-05,
+      "loss": 0.0746,
+      "step": 1880
+    },
+    {
+      "epoch": 1.0262692255342316,
+      "grad_norm": 0.12956227362155914,
+      "learning_rate": 2.994402247956194e-05,
+      "loss": 0.0733,
+      "step": 1885
+    },
+    {
+      "epoch": 1.0289914250714578,
+      "grad_norm": 0.13542629778385162,
+      "learning_rate": 2.9822605674933696e-05,
+      "loss": 0.076,
+      "step": 1890
+    },
+    {
+      "epoch": 1.0317136246086838,
+      "grad_norm": 0.14228489995002747,
+      "learning_rate": 2.9701070605843e-05,
+      "loss": 0.0756,
+      "step": 1895
+    },
+    {
+      "epoch": 1.03443582414591,
+      "grad_norm": 0.13206136226654053,
+      "learning_rate": 2.957942025268689e-05,
+      "loss": 0.0733,
+      "step": 1900
+    },
+    {
+      "epoch": 1.037158023683136,
+      "grad_norm": 0.14194877445697784,
+      "learning_rate": 2.945765759868949e-05,
+      "loss": 0.0837,
+      "step": 1905
+    },
+    {
+      "epoch": 1.039880223220362,
+      "grad_norm": 0.1430080085992813,
+      "learning_rate": 2.933578562982888e-05,
+      "loss": 0.0814,
+      "step": 1910
+    },
+    {
+      "epoch": 1.042602422757588,
+      "grad_norm": 0.12564821541309357,
+      "learning_rate": 2.9213807334763854e-05,
+      "loss": 0.0748,
+      "step": 1915
+    },
+    {
+      "epoch": 1.0453246222948143,
+      "grad_norm": 0.12261082231998444,
+      "learning_rate": 2.9091725704760638e-05,
+      "loss": 0.0668,
+      "step": 1920
+    },
+    {
+      "epoch": 1.0480468218320402,
+      "grad_norm": 0.14175981283187866,
+      "learning_rate": 2.8969543733619554e-05,
+      "loss": 0.0742,
+      "step": 1925
+    },
+    {
+      "epoch": 1.0507690213692664,
+      "grad_norm": 0.1397016942501068,
+      "learning_rate": 2.884726441760155e-05,
+      "loss": 0.0716,
+      "step": 1930
+    },
+    {
+      "epoch": 1.0534912209064924,
+      "grad_norm": 0.1348811835050583,
+      "learning_rate": 2.8724890755354783e-05,
+      "loss": 0.0882,
+      "step": 1935
+    },
+    {
+      "epoch": 1.0562134204437186,
+      "grad_norm": 0.11297186464071274,
+      "learning_rate": 2.8602425747841057e-05,
+      "loss": 0.0715,
+      "step": 1940
+    },
+    {
+      "epoch": 1.0589356199809445,
+      "grad_norm": 0.1416822224855423,
+      "learning_rate": 2.8479872398262198e-05,
+      "loss": 0.0805,
+      "step": 1945
+    },
+    {
+      "epoch": 1.0616578195181707,
+      "grad_norm": 0.147059828042984,
+      "learning_rate": 2.8357233711986487e-05,
+      "loss": 0.0764,
+      "step": 1950
+    },
+    {
+      "epoch": 1.0643800190553967,
+      "grad_norm": 0.137193962931633,
+      "learning_rate": 2.8234512696474875e-05,
+      "loss": 0.0733,
+      "step": 1955
+    },
+    {
+      "epoch": 1.0671022185926229,
+      "grad_norm": 0.12843886017799377,
+      "learning_rate": 2.8111712361207283e-05,
+      "loss": 0.0751,
+      "step": 1960
+    },
+    {
+      "epoch": 1.0698244181298489,
+      "grad_norm": 0.1351711004972458,
+      "learning_rate": 2.7988835717608784e-05,
+      "loss": 0.0776,
+      "step": 1965
+    },
+    {
+      "epoch": 1.072546617667075,
+      "grad_norm": 0.1511557698249817,
+      "learning_rate": 2.7865885778975743e-05,
+      "loss": 0.0764,
+      "step": 1970
+    },
+    {
+      "epoch": 1.075268817204301,
+      "grad_norm": 0.12794414162635803,
+      "learning_rate": 2.774286556040196e-05,
+      "loss": 0.0792,
+      "step": 1975
+    },
+    {
+      "epoch": 1.0779910167415272,
+      "grad_norm": 0.13065066933631897,
+      "learning_rate": 2.7619778078704685e-05,
+      "loss": 0.0706,
+      "step": 1980
+    },
+    {
+      "epoch": 1.0807132162787532,
+      "grad_norm": 0.13252925872802734,
+      "learning_rate": 2.7496626352350664e-05,
+      "loss": 0.0826,
+      "step": 1985
+    },
+    {
+      "epoch": 1.0834354158159794,
+      "grad_norm": 0.1248321682214737,
+      "learning_rate": 2.7373413401382104e-05,
+      "loss": 0.076,
+      "step": 1990
+    },
+    {
+      "epoch": 1.0861576153532053,
+      "grad_norm": 0.14091134071350098,
+      "learning_rate": 2.7250142247342637e-05,
+      "loss": 0.1298,
+      "step": 1995
+    },
+    {
+      "epoch": 1.0888798148904315,
+      "grad_norm": 0.13955925405025482,
+      "learning_rate": 2.7126815913203178e-05,
+      "loss": 0.0753,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0888798148904315,
+      "eval_loss": 0.08035612851381302,
+      "eval_runtime": 228.5599,
+      "eval_samples_per_second": 2.625,
+      "eval_steps_per_second": 0.656,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0916020144276575,
+      "grad_norm": 0.1300753802061081,
+      "learning_rate": 2.7003437423287857e-05,
+      "loss": 0.1036,
+      "step": 2005
+    },
+    {
+      "epoch": 1.0943242139648837,
+      "grad_norm": 0.12375082820653915,
+      "learning_rate": 2.688000980319979e-05,
+      "loss": 0.0744,
+      "step": 2010
+    },
+    {
+      "epoch": 1.0970464135021096,
+      "grad_norm": 0.14749085903167725,
+      "learning_rate": 2.6756536079746907e-05,
+      "loss": 0.0738,
+      "step": 2015
+    },
+    {
+      "epoch": 1.0997686130393358,
+      "grad_norm": 0.13898104429244995,
+      "learning_rate": 2.663301928086774e-05,
+      "loss": 0.0701,
+      "step": 2020
+    },
+    {
+      "epoch": 1.1024908125765618,
+      "grad_norm": 0.12244424223899841,
+      "learning_rate": 2.6509462435557152e-05,
+      "loss": 0.0707,
+      "step": 2025
+    },
+    {
+      "epoch": 1.105213012113788,
+      "grad_norm": 0.13717930018901825,
+      "learning_rate": 2.6385868573792072e-05,
+      "loss": 0.07,
+      "step": 2030
+    },
+    {
+      "epoch": 1.107935211651014,
+      "grad_norm": 0.16028448939323425,
+      "learning_rate": 2.6262240726457167e-05,
+      "loss": 0.0815,
+      "step": 2035
+    },
+    {
+      "epoch": 1.1106574111882401,
+      "grad_norm": 0.13453496992588043,
+      "learning_rate": 2.6138581925270533e-05,
+      "loss": 0.0802,
+      "step": 2040
+    },
+    {
+      "epoch": 1.113379610725466,
+      "grad_norm": 0.14346693456172943,
+      "learning_rate": 2.6014895202709354e-05,
+      "loss": 0.0777,
+      "step": 2045
+    },
+    {
+      "epoch": 1.1161018102626923,
+      "grad_norm": 0.12791860103607178,
+      "learning_rate": 2.5891183591935515e-05,
+      "loss": 0.0798,
+      "step": 2050
+    },
+    {
+      "epoch": 1.1188240097999183,
+      "grad_norm": 0.1344967633485794,
+      "learning_rate": 2.5767450126721255e-05,
+      "loss": 0.0679,
+      "step": 2055
+    },
+    {
+      "epoch": 1.1215462093371444,
+      "grad_norm": 0.1335950344800949,
+      "learning_rate": 2.564369784137472e-05,
+      "loss": 0.0742,
+      "step": 2060
+    },
+    {
+      "epoch": 1.1242684088743704,
+      "grad_norm": 0.12762115895748138,
+      "learning_rate": 2.5519929770665597e-05,
+      "loss": 0.0885,
+      "step": 2065
+    },
+    {
+      "epoch": 1.1269906084115966,
+      "grad_norm": 0.1343313604593277,
+      "learning_rate": 2.539614894975067e-05,
+      "loss": 0.0672,
+      "step": 2070
+    },
+    {
+      "epoch": 1.1297128079488226,
+      "grad_norm": 0.13929401338100433,
+      "learning_rate": 2.5272358414099413e-05,
+      "loss": 0.0765,
+      "step": 2075
+    },
+    {
+      "epoch": 1.1324350074860488,
+      "grad_norm": 0.1336519867181778,
+      "learning_rate": 2.5148561199419516e-05,
+      "loss": 0.0728,
+      "step": 2080
+    },
+    {
+      "epoch": 1.1351572070232747,
+      "grad_norm": 0.13433049619197845,
+      "learning_rate": 2.5024760341582453e-05,
+      "loss": 0.0785,
+      "step": 2085
+    },
+    {
+      "epoch": 1.137879406560501,
+      "grad_norm": 0.12692618370056152,
+      "learning_rate": 2.490095887654906e-05,
+      "loss": 0.0971,
+      "step": 2090
+    },
+    {
+      "epoch": 1.1406016060977269,
+      "grad_norm": 0.1449315845966339,
+      "learning_rate": 2.4777159840295046e-05,
+      "loss": 0.0669,
+      "step": 2095
+    },
+    {
+      "epoch": 1.143323805634953,
+      "grad_norm": 0.14388057589530945,
+      "learning_rate": 2.4653366268736565e-05,
+      "loss": 0.0776,
+      "step": 2100
+    },
+    {
+      "epoch": 1.146046005172179,
+      "grad_norm": 0.12486547976732254,
+      "learning_rate": 2.452958119765577e-05,
+      "loss": 0.0782,
+      "step": 2105
+    },
+    {
+      "epoch": 1.1487682047094052,
+      "grad_norm": 0.135404571890831,
+      "learning_rate": 2.4405807662626338e-05,
+      "loss": 0.0792,
+      "step": 2110
+    },
+    {
+      "epoch": 1.1514904042466312,
+      "grad_norm": 0.14270424842834473,
+      "learning_rate": 2.4282048698939062e-05,
+      "loss": 0.0729,
+      "step": 2115
+    },
+    {
+      "epoch": 1.1542126037838574,
+      "grad_norm": 0.13348889350891113,
+      "learning_rate": 2.41583073415274e-05,
+      "loss": 0.0842,
+      "step": 2120
+    },
+    {
+      "epoch": 1.1569348033210836,
+      "grad_norm": 0.15408015251159668,
+      "learning_rate": 2.403458662489304e-05,
+      "loss": 0.0903,
+      "step": 2125
+    },
+    {
+      "epoch": 1.1596570028583095,
+      "grad_norm": 0.12234120815992355,
+      "learning_rate": 2.3910889583031533e-05,
+      "loss": 0.0793,
+      "step": 2130
+    },
+    {
+      "epoch": 1.1623792023955355,
+      "grad_norm": 0.15008710324764252,
+      "learning_rate": 2.3787219249357803e-05,
+      "loss": 0.0759,
+      "step": 2135
+    },
+    {
+      "epoch": 1.1651014019327617,
+      "grad_norm": 0.133488729596138,
+      "learning_rate": 2.3663578656631858e-05,
+      "loss": 0.0783,
+      "step": 2140
+    },
+    {
+      "epoch": 1.1678236014699879,
+      "grad_norm": 0.12147729843854904,
+      "learning_rate": 2.3539970836884347e-05,
+      "loss": 0.0816,
+      "step": 2145
+    },
+    {
+      "epoch": 1.1705458010072138,
+      "grad_norm": 0.14398206770420074,
+      "learning_rate": 2.3416398821342238e-05,
+      "loss": 0.0739,
+      "step": 2150
+    },
+    {
+      "epoch": 1.1732680005444398,
+      "grad_norm": 0.14384213089942932,
+      "learning_rate": 2.329286564035446e-05,
+      "loss": 0.0764,
+      "step": 2155
+    },
+    {
+      "epoch": 1.175990200081666,
+      "grad_norm": 0.1491377055644989,
+      "learning_rate": 2.3169374323317624e-05,
+      "loss": 0.0716,
+      "step": 2160
+    },
+    {
+      "epoch": 1.1787123996188922,
+      "grad_norm": 0.13119584321975708,
+      "learning_rate": 2.3045927898601702e-05,
+      "loss": 0.0737,
+      "step": 2165
+    },
+    {
+      "epoch": 1.1814345991561181,
+      "grad_norm": 0.13395731151103973,
+      "learning_rate": 2.292252939347577e-05,
+      "loss": 0.0674,
+      "step": 2170
+    },
+    {
+      "epoch": 1.1841567986933441,
+      "grad_norm": 0.1500498354434967,
+      "learning_rate": 2.27991818340338e-05,
+      "loss": 0.0872,
+      "step": 2175
+    },
+    {
+      "epoch": 1.1868789982305703,
+      "grad_norm": 0.15766474604606628,
+      "learning_rate": 2.2675888245120382e-05,
+      "loss": 0.1069,
+      "step": 2180
+    },
+    {
+      "epoch": 1.1896011977677965,
+      "grad_norm": 0.1266118288040161,
+      "learning_rate": 2.255265165025663e-05,
+      "loss": 0.0753,
+      "step": 2185
+    },
+    {
+      "epoch": 1.1923233973050225,
+      "grad_norm": 0.12611445784568787,
+      "learning_rate": 2.2429475071565987e-05,
+      "loss": 0.0704,
+      "step": 2190
+    },
+    {
+      "epoch": 1.1950455968422484,
+      "grad_norm": 0.1382066011428833,
+      "learning_rate": 2.2306361529700125e-05,
+      "loss": 0.081,
+      "step": 2195
+    },
+    {
+      "epoch": 1.1977677963794746,
+      "grad_norm": 0.13451717793941498,
+      "learning_rate": 2.218331404376484e-05,
+      "loss": 0.0813,
+      "step": 2200
+    },
+    {
+      "epoch": 1.2004899959167008,
+      "grad_norm": 0.1681748777627945,
+      "learning_rate": 2.2060335631246075e-05,
+      "loss": 0.0793,
+      "step": 2205
+    },
+    {
+      "epoch": 1.2032121954539268,
+      "grad_norm": 0.14408931136131287,
+      "learning_rate": 2.1937429307935887e-05,
+      "loss": 0.0763,
+      "step": 2210
+    },
+    {
+      "epoch": 1.2059343949911527,
+      "grad_norm": 0.14158952236175537,
+      "learning_rate": 2.1814598087858476e-05,
+      "loss": 0.0882,
+      "step": 2215
+    },
+    {
+      "epoch": 1.208656594528379,
+      "grad_norm": 0.12508870661258698,
+      "learning_rate": 2.16918449831963e-05,
+      "loss": 0.0762,
+      "step": 2220
+    },
+    {
+      "epoch": 1.2113787940656051,
+      "grad_norm": 0.13746081292629242,
+      "learning_rate": 2.15691730042162e-05,
+      "loss": 0.0802,
+      "step": 2225
+    },
+    {
+      "epoch": 1.214100993602831,
+      "grad_norm": 0.12565705180168152,
+      "learning_rate": 2.144658515919557e-05,
+      "loss": 0.071,
+      "step": 2230
+    },
+    {
+      "epoch": 1.216823193140057,
+      "grad_norm": 0.13257178664207458,
+      "learning_rate": 2.1324084454348592e-05,
+      "loss": 0.069,
+      "step": 2235
+    },
+    {
+      "epoch": 1.2195453926772832,
+      "grad_norm": 0.15064120292663574,
+      "learning_rate": 2.1201673893752534e-05,
+      "loss": 0.0793,
+      "step": 2240
+    },
+    {
+      "epoch": 1.2222675922145094,
+      "grad_norm": 0.1452379673719406,
+      "learning_rate": 2.107935647927404e-05,
+      "loss": 0.0653,
+      "step": 2245
+    },
+    {
+      "epoch": 1.2249897917517354,
+      "grad_norm": 0.16073830425739288,
+      "learning_rate": 2.095713521049554e-05,
+      "loss": 0.0726,
+      "step": 2250
+    },
+    {
+      "epoch": 1.2277119912889614,
+      "grad_norm": 0.15698719024658203,
+      "learning_rate": 2.0835013084641704e-05,
+      "loss": 0.0781,
+      "step": 2255
+    },
+    {
+      "epoch": 1.2304341908261875,
+      "grad_norm": 0.13535834848880768,
+      "learning_rate": 2.07129930965059e-05,
+      "loss": 0.0754,
+      "step": 2260
+    },
+    {
+      "epoch": 1.2331563903634137,
+      "grad_norm": 0.13440349698066711,
+      "learning_rate": 2.0591078238376803e-05,
+      "loss": 0.1275,
+      "step": 2265
+    },
+    {
+      "epoch": 1.2358785899006397,
+      "grad_norm": 0.15323837101459503,
+      "learning_rate": 2.0469271499964993e-05,
+      "loss": 0.0738,
+      "step": 2270
+    },
+    {
+      "epoch": 1.238600789437866,
+      "grad_norm": 0.14659421145915985,
+      "learning_rate": 2.034757586832961e-05,
+      "loss": 0.0719,
+      "step": 2275
+    },
+    {
+      "epoch": 1.2413229889750919,
+      "grad_norm": 0.14592771232128143,
+      "learning_rate": 2.022599432780515e-05,
+      "loss": 0.0793,
+      "step": 2280
+    },
+    {
+      "epoch": 1.244045188512318,
+      "grad_norm": 0.12547121942043304,
+      "learning_rate": 2.0104529859928254e-05,
+      "loss": 0.0754,
+      "step": 2285
+    },
+    {
+      "epoch": 1.246767388049544,
+      "grad_norm": 0.13453277945518494,
+      "learning_rate": 1.9983185443364615e-05,
+      "loss": 0.0679,
+      "step": 2290
+    },
+    {
+      "epoch": 1.2494895875867702,
+      "grad_norm": 0.13268929719924927,
+      "learning_rate": 1.9861964053835885e-05,
+      "loss": 0.0778,
+      "step": 2295
+    },
+    {
+      "epoch": 1.2522117871239962,
+      "grad_norm": 0.14037257432937622,
+      "learning_rate": 1.974086866404675e-05,
+      "loss": 0.0874,
+      "step": 2300
+    },
+    {
+      "epoch": 1.2549339866612224,
+      "grad_norm": 0.1446313112974167,
+      "learning_rate": 1.961990224361201e-05,
+      "loss": 0.0708,
+      "step": 2305
+    },
+    {
+      "epoch": 1.2576561861984483,
+      "grad_norm": 0.15174619853496552,
+      "learning_rate": 1.949906775898375e-05,
+      "loss": 0.0841,
+      "step": 2310
+    },
+    {
+      "epoch": 1.2603783857356743,
+      "grad_norm": 0.14297954738140106,
+      "learning_rate": 1.9378368173378618e-05,
+      "loss": 0.0802,
+      "step": 2315
+    },
+    {
+      "epoch": 1.2631005852729005,
+      "grad_norm": 0.1344052255153656,
+      "learning_rate": 1.9257806446705116e-05,
+      "loss": 0.0675,
+      "step": 2320
+    },
+    {
+      "epoch": 1.2658227848101267,
+      "grad_norm": 0.14959311485290527,
+      "learning_rate": 1.913738553549106e-05,
+      "loss": 0.0848,
+      "step": 2325
+    },
+    {
+      "epoch": 1.2685449843473526,
+      "grad_norm": 0.12404653429985046,
+      "learning_rate": 1.9017108392811065e-05,
+      "loss": 0.0687,
+      "step": 2330
+    },
+    {
+      "epoch": 1.2712671838845788,
+      "grad_norm": 0.13011127710342407,
+      "learning_rate": 1.8896977968214078e-05,
+      "loss": 0.113,
+      "step": 2335
+    },
+    {
+      "epoch": 1.2739893834218048,
+      "grad_norm": 0.14289557933807373,
+      "learning_rate": 1.877699720765114e-05,
+      "loss": 0.0771,
+      "step": 2340
+    },
+    {
+      "epoch": 1.276711582959031,
+      "grad_norm": 0.14011026918888092,
+      "learning_rate": 1.8657169053403052e-05,
+      "loss": 0.0771,
+      "step": 2345
+    },
+    {
+      "epoch": 1.279433782496257,
+      "grad_norm": 0.1479315608739853,
+      "learning_rate": 1.8537496444008283e-05,
+      "loss": 0.07,
+      "step": 2350
+    },
+    {
+      "epoch": 1.2821559820334831,
+      "grad_norm": 0.1444329172372818,
+      "learning_rate": 1.841798231419087e-05,
+      "loss": 0.0815,
+      "step": 2355
+    },
+    {
+      "epoch": 1.284878181570709,
+      "grad_norm": 0.11986621469259262,
+      "learning_rate": 1.8298629594788467e-05,
+      "loss": 0.0689,
+      "step": 2360
+    },
+    {
+      "epoch": 1.2876003811079353,
+      "grad_norm": 0.13101747632026672,
+      "learning_rate": 1.817944121268048e-05,
+      "loss": 0.0728,
+      "step": 2365
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 0.1301647424697876,
+      "learning_rate": 1.8060420090716265e-05,
+      "loss": 0.0789,
+      "step": 2370
+    },
+    {
+      "epoch": 1.2930447801823874,
+      "grad_norm": 0.14230471849441528,
+      "learning_rate": 1.794156914764349e-05,
+      "loss": 0.0702,
+      "step": 2375
+    },
+    {
+      "epoch": 1.2957669797196134,
+      "grad_norm": 0.15737561881542206,
+      "learning_rate": 1.7822891298036515e-05,
+      "loss": 0.0781,
+      "step": 2380
+    },
+    {
+      "epoch": 1.2984891792568396,
+      "grad_norm": 0.16721278429031372,
+      "learning_rate": 1.7704389452224944e-05,
+      "loss": 0.0805,
+      "step": 2385
+    },
+    {
+      "epoch": 1.3012113787940656,
+      "grad_norm": 0.1321476846933365,
+      "learning_rate": 1.7586066516222276e-05,
+      "loss": 0.0768,
+      "step": 2390
+    },
+    {
+      "epoch": 1.3039335783312918,
+      "grad_norm": 0.15134398639202118,
+      "learning_rate": 1.7467925391654585e-05,
+      "loss": 0.0825,
+      "step": 2395
+    },
+    {
+      "epoch": 1.3066557778685177,
+      "grad_norm": 0.14100222289562225,
+      "learning_rate": 1.73499689756894e-05,
+      "loss": 0.0823,
+      "step": 2400
+    },
+    {
+      "epoch": 1.309377977405744,
+      "grad_norm": 0.1558840423822403,
+      "learning_rate": 1.7232200160964657e-05,
+      "loss": 0.0832,
+      "step": 2405
+    },
+    {
+      "epoch": 1.3121001769429699,
+      "grad_norm": 0.14366985857486725,
+      "learning_rate": 1.7114621835517773e-05,
+      "loss": 0.0791,
+      "step": 2410
+    },
+    {
+      "epoch": 1.314822376480196,
+      "grad_norm": 0.12937745451927185,
+      "learning_rate": 1.699723688271477e-05,
+      "loss": 0.0703,
+      "step": 2415
+    },
+    {
+      "epoch": 1.317544576017422,
+      "grad_norm": 0.12883788347244263,
+      "learning_rate": 1.6880048181179652e-05,
+      "loss": 0.0797,
+      "step": 2420
+    },
+    {
+      "epoch": 1.3202667755546482,
+      "grad_norm": 0.13123267889022827,
+      "learning_rate": 1.6763058604723723e-05,
+      "loss": 0.0727,
+      "step": 2425
+    },
+    {
+      "epoch": 1.3229889750918742,
+      "grad_norm": 0.10518030822277069,
+      "learning_rate": 1.6646271022275185e-05,
+      "loss": 0.066,
+      "step": 2430
+    },
+    {
+      "epoch": 1.3257111746291004,
+      "grad_norm": 0.13036532700061798,
+      "learning_rate": 1.6529688297808726e-05,
+      "loss": 0.078,
+      "step": 2435
+    },
+    {
+      "epoch": 1.3284333741663263,
+      "grad_norm": 0.14793965220451355,
+      "learning_rate": 1.6413313290275355e-05,
+      "loss": 0.0779,
+      "step": 2440
+    },
+    {
+      "epoch": 1.3311555737035525,
+      "grad_norm": 0.15071183443069458,
+      "learning_rate": 1.629714885353221e-05,
+      "loss": 0.0835,
+      "step": 2445
+    },
+    {
+      "epoch": 1.3338777732407785,
+      "grad_norm": 0.13086527585983276,
+      "learning_rate": 1.618119783627263e-05,
+      "loss": 0.0759,
+      "step": 2450
+    },
+    {
+      "epoch": 1.3365999727780047,
+      "grad_norm": 0.13066627085208893,
+      "learning_rate": 1.6065463081956292e-05,
+      "loss": 0.0745,
+      "step": 2455
+    },
+    {
+      "epoch": 1.3393221723152307,
+      "grad_norm": 0.15003521740436554,
+      "learning_rate": 1.5949947428739448e-05,
+      "loss": 0.084,
+      "step": 2460
+    },
+    {
+      "epoch": 1.3420443718524568,
+      "grad_norm": 0.12341570854187012,
+      "learning_rate": 1.5834653709405368e-05,
+      "loss": 0.0707,
+      "step": 2465
+    },
+    {
+      "epoch": 1.3447665713896828,
+      "grad_norm": 0.1332896649837494,
+      "learning_rate": 1.571958475129484e-05,
+      "loss": 0.0701,
+      "step": 2470
+    },
+    {
+      "epoch": 1.347488770926909,
+      "grad_norm": 0.12761184573173523,
+      "learning_rate": 1.5604743376236847e-05,
+      "loss": 0.0708,
+      "step": 2475
+    },
+    {
+      "epoch": 1.350210970464135,
+      "grad_norm": 0.12873777747154236,
+      "learning_rate": 1.549013240047937e-05,
+      "loss": 0.0731,
+      "step": 2480
+    },
+    {
+      "epoch": 1.3529331700013612,
+      "grad_norm": 0.1278139352798462,
+      "learning_rate": 1.537575463462031e-05,
+      "loss": 0.0712,
+      "step": 2485
+    },
+    {
+      "epoch": 1.3556553695385871,
+      "grad_norm": 0.16153199970722198,
+      "learning_rate": 1.526161288353861e-05,
+      "loss": 0.0825,
+      "step": 2490
+    },
+    {
+      "epoch": 1.3583775690758133,
+      "grad_norm": 0.13839781284332275,
+      "learning_rate": 1.5147709946325395e-05,
+      "loss": 0.0769,
+      "step": 2495
+    },
+    {
+      "epoch": 1.3610997686130393,
+      "grad_norm": 0.1404309719800949,
+      "learning_rate": 1.5034048616215402e-05,
+      "loss": 0.0737,
+      "step": 2500
+    },
+    {
+      "epoch": 1.3610997686130393,
+      "eval_loss": 0.07860012352466583,
+      "eval_runtime": 248.5723,
+      "eval_samples_per_second": 2.414,
+      "eval_steps_per_second": 0.603,
+      "step": 2500
+    },
+    {
+      "epoch": 1.3638219681502655,
+      "grad_norm": 0.14883238077163696,
+      "learning_rate": 1.4920631680518432e-05,
+      "loss": 0.0695,
+      "step": 2505
+    },
+    {
+      "epoch": 1.3665441676874914,
+      "grad_norm": 0.14661747217178345,
+      "learning_rate": 1.4807461920551028e-05,
+      "loss": 0.0752,
+      "step": 2510
+    },
+    {
+      "epoch": 1.3692663672247176,
+      "grad_norm": 0.13054049015045166,
+      "learning_rate": 1.469454211156826e-05,
+      "loss": 0.0698,
+      "step": 2515
+    },
+    {
+      "epoch": 1.3719885667619436,
+      "grad_norm": 0.13900673389434814,
+      "learning_rate": 1.4581875022695653e-05,
+      "loss": 0.0897,
+      "step": 2520
+    },
+    {
+      "epoch": 1.3747107662991698,
+      "grad_norm": 0.1521766632795334,
+      "learning_rate": 1.4469463416861307e-05,
+      "loss": 0.0751,
+      "step": 2525
+    },
+    {
+      "epoch": 1.3774329658363957,
+      "grad_norm": 0.14469105005264282,
+      "learning_rate": 1.4357310050728115e-05,
+      "loss": 0.0807,
+      "step": 2530
+    },
+    {
+      "epoch": 1.380155165373622,
+      "grad_norm": 0.132577583193779,
+      "learning_rate": 1.4245417674626183e-05,
+      "loss": 0.0693,
+      "step": 2535
+    },
+    {
+      "epoch": 1.382877364910848,
+      "grad_norm": 0.12606105208396912,
+      "learning_rate": 1.4133789032485367e-05,
+      "loss": 0.0739,
+      "step": 2540
+    },
+    {
+      "epoch": 1.385599564448074,
+      "grad_norm": 0.1424499899148941,
+      "learning_rate": 1.4022426861767998e-05,
+      "loss": 0.0806,
+      "step": 2545
+    },
+    {
+      "epoch": 1.3883217639853,
+      "grad_norm": 0.15580779314041138,
+      "learning_rate": 1.391133389340174e-05,
+      "loss": 0.0737,
+      "step": 2550
+    },
+    {
+      "epoch": 1.3910439635225262,
+      "grad_norm": 0.14672575891017914,
+      "learning_rate": 1.3800512851712635e-05,
+      "loss": 0.0746,
+      "step": 2555
+    },
+    {
+      "epoch": 1.3937661630597522,
+      "grad_norm": 0.15802086889743805,
+      "learning_rate": 1.3689966454358256e-05,
+      "loss": 0.0768,
+      "step": 2560
+    },
+    {
+      "epoch": 1.3964883625969784,
+      "grad_norm": 0.1357397735118866,
+      "learning_rate": 1.3579697412261117e-05,
+      "loss": 0.0724,
+      "step": 2565
+    },
+    {
+      "epoch": 1.3992105621342044,
+      "grad_norm": 0.14769281446933746,
+      "learning_rate": 1.3469708429542157e-05,
+      "loss": 0.0738,
+      "step": 2570
+    },
+    {
+      "epoch": 1.4019327616714305,
+      "grad_norm": 0.1663895845413208,
+      "learning_rate": 1.3360002203454442e-05,
+      "loss": 0.0739,
+      "step": 2575
+    },
+    {
+      "epoch": 1.4046549612086565,
+      "grad_norm": 0.13682472705841064,
+      "learning_rate": 1.325058142431701e-05,
+      "loss": 0.0732,
+      "step": 2580
+    },
+    {
+      "epoch": 1.4073771607458827,
+      "grad_norm": 0.1464831680059433,
+      "learning_rate": 1.3141448775448875e-05,
+      "loss": 0.0781,
+      "step": 2585
+    },
+    {
+      "epoch": 1.4100993602831087,
+      "grad_norm": 0.14444297552108765,
+      "learning_rate": 1.3032606933103305e-05,
+      "loss": 0.0742,
+      "step": 2590
+    },
+    {
+      "epoch": 1.4128215598203349,
+      "grad_norm": 0.14417661726474762,
+      "learning_rate": 1.2924058566402098e-05,
+      "loss": 0.0761,
+      "step": 2595
+    },
+    {
+      "epoch": 1.4155437593575608,
+      "grad_norm": 0.12622784078121185,
+      "learning_rate": 1.2815806337270186e-05,
+      "loss": 0.0693,
+      "step": 2600
+    },
+    {
+      "epoch": 1.418265958894787,
+      "grad_norm": 0.1415053904056549,
+      "learning_rate": 1.270785290037031e-05,
+      "loss": 0.0771,
+      "step": 2605
+    },
+    {
+      "epoch": 1.4209881584320132,
+      "grad_norm": 0.14631399512290955,
+      "learning_rate": 1.260020090303797e-05,
+      "loss": 0.1018,
+      "step": 2610
+    },
+    {
+      "epoch": 1.4237103579692392,
+      "grad_norm": 0.1438084840774536,
+      "learning_rate": 1.2492852985216483e-05,
+      "loss": 0.0708,
+      "step": 2615
+    },
+    {
+      "epoch": 1.4264325575064651,
+      "grad_norm": 0.12973198294639587,
+      "learning_rate": 1.2385811779392236e-05,
+      "loss": 0.0795,
+      "step": 2620
+    },
+    {
+      "epoch": 1.4291547570436913,
+      "grad_norm": 0.15162228047847748,
+      "learning_rate": 1.2279079910530147e-05,
+      "loss": 0.0811,
+      "step": 2625
+    },
+    {
+      "epoch": 1.4318769565809175,
+      "grad_norm": 0.14159157872200012,
+      "learning_rate": 1.2172659996009254e-05,
+      "loss": 0.0762,
+      "step": 2630
+    },
+    {
+      "epoch": 1.4345991561181435,
+      "grad_norm": 0.13606242835521698,
+      "learning_rate": 1.2066554645558578e-05,
+      "loss": 0.0739,
+      "step": 2635
+    },
+    {
+      "epoch": 1.4373213556553694,
+      "grad_norm": 0.15046873688697815,
+      "learning_rate": 1.1960766461193124e-05,
+      "loss": 0.0663,
+      "step": 2640
+    },
+    {
+      "epoch": 1.4400435551925956,
+      "grad_norm": 0.12745556235313416,
+      "learning_rate": 1.1855298037150022e-05,
+      "loss": 0.073,
+      "step": 2645
+    },
+    {
+      "epoch": 1.4427657547298218,
+      "grad_norm": 0.156600683927536,
+      "learning_rate": 1.1750151959824961e-05,
+      "loss": 0.0746,
+      "step": 2650
+    },
+    {
+      "epoch": 1.4454879542670478,
+      "grad_norm": 0.13598720729351044,
+      "learning_rate": 1.1645330807708713e-05,
+      "loss": 0.0687,
+      "step": 2655
+    },
+    {
+      "epoch": 1.4482101538042738,
+      "grad_norm": 0.13264238834381104,
+      "learning_rate": 1.1540837151323951e-05,
+      "loss": 0.0814,
+      "step": 2660
+    },
+    {
+      "epoch": 1.4509323533415,
+      "grad_norm": 0.13126635551452637,
+      "learning_rate": 1.143667355316219e-05,
+      "loss": 0.0768,
+      "step": 2665
+    },
+    {
+      "epoch": 1.4536545528787261,
+      "grad_norm": 0.14652326703071594,
+      "learning_rate": 1.1332842567620941e-05,
+      "loss": 0.0731,
+      "step": 2670
+    },
+    {
+      "epoch": 1.456376752415952,
+      "grad_norm": 0.14913715422153473,
+      "learning_rate": 1.1229346740941088e-05,
+      "loss": 0.0893,
+      "step": 2675
+    },
+    {
+      "epoch": 1.459098951953178,
+      "grad_norm": 0.15281103551387787,
+      "learning_rate": 1.1126188611144406e-05,
+      "loss": 0.0981,
+      "step": 2680
+    },
+    {
+      "epoch": 1.4618211514904043,
+      "grad_norm": 0.14536678791046143,
+      "learning_rate": 1.102337070797137e-05,
+      "loss": 0.0775,
+      "step": 2685
+    },
+    {
+      "epoch": 1.4645433510276304,
+      "grad_norm": 0.14888562262058258,
+      "learning_rate": 1.0920895552819118e-05,
+      "loss": 0.0828,
+      "step": 2690
+    },
+    {
+      "epoch": 1.4672655505648564,
+      "grad_norm": 0.12814417481422424,
+      "learning_rate": 1.0818765658679576e-05,
+      "loss": 0.0766,
+      "step": 2695
+    },
+    {
+      "epoch": 1.4699877501020824,
+      "grad_norm": 0.12727950513362885,
+      "learning_rate": 1.0716983530077843e-05,
+      "loss": 0.0776,
+      "step": 2700
+    },
+    {
+      "epoch": 1.4727099496393086,
+      "grad_norm": 0.12727922201156616,
+      "learning_rate": 1.0615551663010806e-05,
+      "loss": 0.0689,
+      "step": 2705
+    },
+    {
+      "epoch": 1.4754321491765348,
+      "grad_norm": 0.14102789759635925,
+      "learning_rate": 1.051447254488591e-05,
+      "loss": 0.0634,
+      "step": 2710
+    },
+    {
+      "epoch": 1.4781543487137607,
+      "grad_norm": 0.14436770975589752,
+      "learning_rate": 1.0413748654460149e-05,
+      "loss": 0.0796,
+      "step": 2715
+    },
+    {
+      "epoch": 1.4808765482509867,
+      "grad_norm": 0.1291683316230774,
+      "learning_rate": 1.0313382461779306e-05,
+      "loss": 0.0755,
+      "step": 2720
+    },
+    {
+      "epoch": 1.4835987477882129,
+      "grad_norm": 0.12600617110729218,
+      "learning_rate": 1.0213376428117333e-05,
+      "loss": 0.0658,
+      "step": 2725
+    },
+    {
+      "epoch": 1.486320947325439,
+      "grad_norm": 0.14813588559627533,
+      "learning_rate": 1.0113733005916057e-05,
+      "loss": 0.075,
+      "step": 2730
+    },
+    {
+      "epoch": 1.489043146862665,
+      "grad_norm": 0.15321391820907593,
+      "learning_rate": 1.0014454638724982e-05,
+      "loss": 0.0774,
+      "step": 2735
+    },
+    {
+      "epoch": 1.491765346399891,
+      "grad_norm": 0.13844044506549835,
+      "learning_rate": 9.915543761141432e-06,
+      "loss": 0.074,
+      "step": 2740
+    },
+    {
+      "epoch": 1.4944875459371172,
+      "grad_norm": 0.14296895265579224,
+      "learning_rate": 9.81700279875075e-06,
+      "loss": 0.0796,
+      "step": 2745
+    },
+    {
+      "epoch": 1.4972097454743434,
+      "grad_norm": 0.14961597323417664,
+      "learning_rate": 9.718834168066904e-06,
+      "loss": 0.0713,
+      "step": 2750
+    },
+    {
+      "epoch": 1.4999319450115693,
+      "grad_norm": 0.1663280427455902,
+      "learning_rate": 9.62104027647319e-06,
+      "loss": 0.077,
+      "step": 2755
+    },
+    {
+      "epoch": 1.5026541445487953,
+      "grad_norm": 0.14651760458946228,
+      "learning_rate": 9.523623522163197e-06,
+      "loss": 0.0719,
+      "step": 2760
+    },
+    {
+      "epoch": 1.5053763440860215,
+      "grad_norm": 0.14970749616622925,
+      "learning_rate": 9.426586294082013e-06,
+      "loss": 0.0776,
+      "step": 2765
+    },
+    {
+      "epoch": 1.5080985436232477,
+      "grad_norm": 0.15019264817237854,
+      "learning_rate": 9.329930971867595e-06,
+      "loss": 0.0771,
+      "step": 2770
+    },
+    {
+      "epoch": 1.5108207431604737,
+      "grad_norm": 0.14440083503723145,
+      "learning_rate": 9.233659925792477e-06,
+      "loss": 0.072,
+      "step": 2775
+    },
+    {
+      "epoch": 1.5135429426976996,
+      "grad_norm": 0.138129323720932,
+      "learning_rate": 9.137775516705604e-06,
+      "loss": 0.0758,
+      "step": 2780
+    },
+    {
+      "epoch": 1.5162651422349258,
+      "grad_norm": 0.14640666544437408,
+      "learning_rate": 9.042280095974434e-06,
+      "loss": 0.0844,
+      "step": 2785
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 0.14141102135181427,
+      "learning_rate": 8.947176005427324e-06,
+      "loss": 0.0798,
+      "step": 2790
+    },
+    {
+      "epoch": 1.521709541309378,
+      "grad_norm": 0.14003820717334747,
+      "learning_rate": 8.852465577296015e-06,
+      "loss": 0.0761,
+      "step": 2795
+    },
+    {
+      "epoch": 1.524431740846604,
+      "grad_norm": 0.13643573224544525,
+      "learning_rate": 8.75815113415852e-06,
+      "loss": 0.0753,
+      "step": 2800
+    },
+    {
+      "epoch": 1.5271539403838301,
+      "grad_norm": 0.13099689781665802,
+      "learning_rate": 8.66423498888213e-06,
+      "loss": 0.0764,
+      "step": 2805
+    },
+    {
+      "epoch": 1.5298761399210563,
+      "grad_norm": 0.16020213067531586,
+      "learning_rate": 8.570719444566702e-06,
+      "loss": 0.0733,
+      "step": 2810
+    },
+    {
+      "epoch": 1.5325983394582823,
+      "grad_norm": 0.1546473205089569,
+      "learning_rate": 8.477606794488183e-06,
+      "loss": 0.0839,
+      "step": 2815
+    },
+    {
+      "epoch": 1.5353205389955082,
+      "grad_norm": 0.14650870859622955,
+      "learning_rate": 8.384899322042356e-06,
+      "loss": 0.0837,
+      "step": 2820
+    },
+    {
+      "epoch": 1.5380427385327344,
+      "grad_norm": 0.11506952345371246,
+      "learning_rate": 8.29259930068887e-06,
+      "loss": 0.0775,
+      "step": 2825
+    },
+    {
+      "epoch": 1.5407649380699606,
+      "grad_norm": 0.14615637063980103,
+      "learning_rate": 8.200708993895476e-06,
+      "loss": 0.0744,
+      "step": 2830
+    },
+    {
+      "epoch": 1.5434871376071866,
+      "grad_norm": 0.13769914209842682,
+      "learning_rate": 8.109230655082517e-06,
+      "loss": 0.0795,
+      "step": 2835
+    },
+    {
+      "epoch": 1.5462093371444126,
+      "grad_norm": 0.13748113811016083,
+      "learning_rate": 8.018166527567672e-06,
+      "loss": 0.0675,
+      "step": 2840
+    },
+    {
+      "epoch": 1.5489315366816387,
+      "grad_norm": 0.15049228072166443,
+      "learning_rate": 7.927518844510941e-06,
+      "loss": 0.0703,
+      "step": 2845
+    },
+    {
+      "epoch": 1.551653736218865,
+      "grad_norm": 0.13132914900779724,
+      "learning_rate": 7.837289828859884e-06,
+      "loss": 0.0742,
+      "step": 2850
+    },
+    {
+      "epoch": 1.554375935756091,
+      "grad_norm": 0.14202824234962463,
+      "learning_rate": 7.7474816932951e-06,
+      "loss": 0.0699,
+      "step": 2855
+    },
+    {
+      "epoch": 1.5570981352933169,
+      "grad_norm": 0.16140979528427124,
+      "learning_rate": 7.658096640175985e-06,
+      "loss": 0.0777,
+      "step": 2860
+    },
+    {
+      "epoch": 1.559820334830543,
+      "grad_norm": 0.1300312578678131,
+      "learning_rate": 7.569136861486686e-06,
+      "loss": 0.0648,
+      "step": 2865
+    },
+    {
+      "epoch": 1.5625425343677692,
+      "grad_norm": 0.14492963254451752,
+      "learning_rate": 7.4806045387823935e-06,
+      "loss": 0.0713,
+      "step": 2870
+    },
+    {
+      "epoch": 1.5652647339049952,
+      "grad_norm": 0.13518249988555908,
+      "learning_rate": 7.392501843135816e-06,
+      "loss": 0.0649,
+      "step": 2875
+    },
+    {
+      "epoch": 1.5679869334422212,
+      "grad_norm": 0.14563652873039246,
+      "learning_rate": 7.304830935083934e-06,
+      "loss": 0.0724,
+      "step": 2880
+    },
+    {
+      "epoch": 1.5707091329794474,
+      "grad_norm": 0.1495407372713089,
+      "learning_rate": 7.2175939645750454e-06,
+      "loss": 0.0695,
+      "step": 2885
+    },
+    {
+      "epoch": 1.5734313325166736,
+      "grad_norm": 0.15107296407222748,
+      "learning_rate": 7.130793070916006e-06,
+      "loss": 0.0762,
+      "step": 2890
+    },
+    {
+      "epoch": 1.5761535320538995,
+      "grad_norm": 0.13882042467594147,
+      "learning_rate": 7.0444303827197994e-06,
+      "loss": 0.0698,
+      "step": 2895
+    },
+    {
+      "epoch": 1.5788757315911255,
+      "grad_norm": 0.15497738122940063,
+      "learning_rate": 6.9585080178533155e-06,
+      "loss": 0.0729,
+      "step": 2900
+    },
+    {
+      "epoch": 1.5815979311283517,
+      "grad_norm": 0.14168910682201385,
+      "learning_rate": 6.873028083385435e-06,
+      "loss": 0.0681,
+      "step": 2905
+    },
+    {
+      "epoch": 1.5843201306655779,
+      "grad_norm": 0.1269499659538269,
+      "learning_rate": 6.787992675535318e-06,
+      "loss": 0.0803,
+      "step": 2910
+    },
+    {
+      "epoch": 1.5870423302028038,
+      "grad_norm": 0.14480414986610413,
+      "learning_rate": 6.703403879621048e-06,
+      "loss": 0.0719,
+      "step": 2915
+    },
+    {
+      "epoch": 1.5897645297400298,
+      "grad_norm": 0.14816336333751678,
+      "learning_rate": 6.619263770008471e-06,
+      "loss": 0.0764,
+      "step": 2920
+    },
+    {
+      "epoch": 1.592486729277256,
+      "grad_norm": 0.13959960639476776,
+      "learning_rate": 6.535574410060321e-06,
+      "loss": 0.0777,
+      "step": 2925
+    },
+    {
+      "epoch": 1.5952089288144822,
+      "grad_norm": 0.13289447128772736,
+      "learning_rate": 6.452337852085621e-06,
+      "loss": 0.077,
+      "step": 2930
+    },
+    {
+      "epoch": 1.5979311283517081,
+      "grad_norm": 0.14249089360237122,
+      "learning_rate": 6.369556137289373e-06,
+      "loss": 0.0678,
+      "step": 2935
+    },
+    {
+      "epoch": 1.600653327888934,
+      "grad_norm": 0.14189301431179047,
+      "learning_rate": 6.28723129572247e-06,
+      "loss": 0.077,
+      "step": 2940
+    },
+    {
+      "epoch": 1.6033755274261603,
+      "grad_norm": 0.13121193647384644,
+      "learning_rate": 6.2053653462319475e-06,
+      "loss": 0.0651,
+      "step": 2945
+    },
+    {
+      "epoch": 1.6060977269633865,
+      "grad_norm": 0.13624070584774017,
+      "learning_rate": 6.123960296411449e-06,
+      "loss": 0.0726,
+      "step": 2950
+    },
+    {
+      "epoch": 1.6088199265006125,
+      "grad_norm": 0.1546383649110794,
+      "learning_rate": 6.043018142552009e-06,
+      "loss": 0.0747,
+      "step": 2955
+    },
+    {
+      "epoch": 1.6115421260378384,
+      "grad_norm": 0.14421793818473816,
+      "learning_rate": 5.962540869593081e-06,
+      "loss": 0.0818,
+      "step": 2960
+    },
+    {
+      "epoch": 1.6142643255750646,
+      "grad_norm": 0.13135267794132233,
+      "learning_rate": 5.882530451073886e-06,
+      "loss": 0.0724,
+      "step": 2965
+    },
+    {
+      "epoch": 1.6169865251122908,
+      "grad_norm": 0.14589843153953552,
+      "learning_rate": 5.8029888490850005e-06,
+      "loss": 0.0769,
+      "step": 2970
+    },
+    {
+      "epoch": 1.6197087246495168,
+      "grad_norm": 0.1423359513282776,
+      "learning_rate": 5.723918014220236e-06,
+      "loss": 0.0662,
+      "step": 2975
+    },
+    {
+      "epoch": 1.6224309241867427,
+      "grad_norm": 0.12941160798072815,
+      "learning_rate": 5.645319885528824e-06,
+      "loss": 0.0938,
+      "step": 2980
+    },
+    {
+      "epoch": 1.625153123723969,
+      "grad_norm": 0.16445757448673248,
+      "learning_rate": 5.5671963904678185e-06,
+      "loss": 0.0752,
+      "step": 2985
+    },
+    {
+      "epoch": 1.627875323261195,
+      "grad_norm": 0.15879502892494202,
+      "learning_rate": 5.489549444854908e-06,
+      "loss": 0.0875,
+      "step": 2990
+    },
+    {
+      "epoch": 1.630597522798421,
+      "grad_norm": 0.1418047547340393,
+      "learning_rate": 5.412380952821358e-06,
+      "loss": 0.0778,
+      "step": 2995
+    },
+    {
+      "epoch": 1.633319722335647,
+      "grad_norm": 0.1362149566411972,
+      "learning_rate": 5.33569280676536e-06,
+      "loss": 0.0692,
+      "step": 3000
+    },
+    {
+      "epoch": 1.633319722335647,
+      "eval_loss": 0.07731131464242935,
+      "eval_runtime": 262.646,
+      "eval_samples_per_second": 2.284,
+      "eval_steps_per_second": 0.571,
+      "step": 3000
+    },
+    {
+      "epoch": 1.6360419218728732,
+      "grad_norm": 0.14646309614181519,
+      "learning_rate": 5.259486887305592e-06,
+      "loss": 0.068,
+      "step": 3005
+    },
+    {
+      "epoch": 1.6387641214100994,
+      "grad_norm": 0.13968642055988312,
+      "learning_rate": 5.183765063235138e-06,
+      "loss": 0.0727,
+      "step": 3010
+    },
+    {
+      "epoch": 1.6414863209473254,
+      "grad_norm": 0.1375684291124344,
+      "learning_rate": 5.108529191475636e-06,
+      "loss": 0.0794,
+      "step": 3015
+    },
+    {
+      "epoch": 1.6442085204845516,
+      "grad_norm": 0.1326666921377182,
+      "learning_rate": 5.033781117031738e-06,
+      "loss": 0.072,
+      "step": 3020
+    },
+    {
+      "epoch": 1.6469307200217775,
+      "grad_norm": 0.14453214406967163,
+      "learning_rate": 4.959522672945891e-06,
+      "loss": 0.0763,
+      "step": 3025
+    },
+    {
+      "epoch": 1.6496529195590037,
+      "grad_norm": 0.15132352709770203,
+      "learning_rate": 4.885755680253334e-06,
+      "loss": 0.1175,
+      "step": 3030
+    },
+    {
+      "epoch": 1.65237511909623,
+      "grad_norm": 0.15359079837799072,
+      "learning_rate": 4.812481947937497e-06,
+      "loss": 0.0837,
+      "step": 3035
+    },
+    {
+      "epoch": 1.6550973186334559,
+      "grad_norm": 0.13241231441497803,
+      "learning_rate": 4.739703272885626e-06,
+      "loss": 0.0738,
+      "step": 3040
+    },
+    {
+      "epoch": 1.6578195181706818,
+      "grad_norm": 0.14873500168323517,
+      "learning_rate": 4.667421439844691e-06,
+      "loss": 0.0708,
+      "step": 3045
+    },
+    {
+      "epoch": 1.660541717707908,
+      "grad_norm": 0.13607364892959595,
+      "learning_rate": 4.5956382213776295e-06,
+      "loss": 0.1151,
+      "step": 3050
+    },
+    {
+      "epoch": 1.6632639172451342,
+      "grad_norm": 0.15463878214359283,
+      "learning_rate": 4.524355377819897e-06,
+      "loss": 0.0831,
+      "step": 3055
+    },
+    {
+      "epoch": 1.6659861167823602,
+      "grad_norm": 0.1272389441728592,
+      "learning_rate": 4.45357465723629e-06,
+      "loss": 0.0714,
+      "step": 3060
+    },
+    {
+      "epoch": 1.6687083163195862,
+      "grad_norm": 0.14218097925186157,
+      "learning_rate": 4.383297795378061e-06,
+      "loss": 0.0675,
+      "step": 3065
+    },
+    {
+      "epoch": 1.6714305158568123,
+      "grad_norm": 0.1373002678155899,
+      "learning_rate": 4.3135265156403756e-06,
+      "loss": 0.0791,
+      "step": 3070
+    },
+    {
+      "epoch": 1.6741527153940385,
+      "grad_norm": 0.16303496062755585,
+      "learning_rate": 4.244262529020041e-06,
+      "loss": 0.0757,
+      "step": 3075
+    },
+    {
+      "epoch": 1.6768749149312645,
+      "grad_norm": 0.1444956660270691,
+      "learning_rate": 4.175507534073533e-06,
+      "loss": 0.0732,
+      "step": 3080
+    },
+    {
+      "epoch": 1.6795971144684905,
+      "grad_norm": 0.1505775898694992,
+      "learning_rate": 4.107263216875362e-06,
+      "loss": 0.0751,
+      "step": 3085
+    },
+    {
+      "epoch": 1.6823193140057167,
+      "grad_norm": 0.14044207334518433,
+      "learning_rate": 4.039531250976735e-06,
+      "loss": 0.069,
+      "step": 3090
+    },
+    {
+      "epoch": 1.6850415135429428,
+      "grad_norm": 0.14999620616436005,
+      "learning_rate": 3.972313297364494e-06,
+      "loss": 0.0708,
+      "step": 3095
+    },
+    {
+      "epoch": 1.6877637130801688,
+      "grad_norm": 0.1265767216682434,
+      "learning_rate": 3.90561100442036e-06,
+      "loss": 0.0655,
+      "step": 3100
+    },
+    {
+      "epoch": 1.6904859126173948,
+      "grad_norm": 0.16221390664577484,
+      "learning_rate": 3.8394260078805755e-06,
+      "loss": 0.0736,
+      "step": 3105
+    },
+    {
+      "epoch": 1.693208112154621,
+      "grad_norm": 0.15090176463127136,
+      "learning_rate": 3.7737599307957365e-06,
+      "loss": 0.0669,
+      "step": 3110
+    },
+    {
+      "epoch": 1.6959303116918472,
+      "grad_norm": 0.14406718313694,
+      "learning_rate": 3.708614383491016e-06,
+      "loss": 0.0737,
+      "step": 3115
+    },
+    {
+      "epoch": 1.6986525112290731,
+      "grad_norm": 0.12115464359521866,
+      "learning_rate": 3.643990963526672e-06,
+      "loss": 0.0627,
+      "step": 3120
+    },
+    {
+      "epoch": 1.701374710766299,
+      "grad_norm": 0.16066431999206543,
+      "learning_rate": 3.579891255658846e-06,
+      "loss": 0.0757,
+      "step": 3125
+    },
+    {
+      "epoch": 1.7040969103035253,
+      "grad_norm": 0.13997657597064972,
+      "learning_rate": 3.5163168318007365e-06,
+      "loss": 0.1398,
+      "step": 3130
+    },
+    {
+      "epoch": 1.7068191098407515,
+      "grad_norm": 0.13934873044490814,
+      "learning_rate": 3.4532692509840313e-06,
+      "loss": 0.0723,
+      "step": 3135
+    },
+    {
+      "epoch": 1.7095413093779774,
+      "grad_norm": 0.13022823631763458,
+      "learning_rate": 3.390750059320688e-06,
+      "loss": 0.0789,
+      "step": 3140
+    },
+    {
+      "epoch": 1.7122635089152034,
+      "grad_norm": 0.13600829243659973,
+      "learning_rate": 3.328760789964988e-06,
+      "loss": 0.0728,
+      "step": 3145
+    },
+    {
+      "epoch": 1.7149857084524296,
+      "grad_norm": 0.1362736076116562,
+      "learning_rate": 3.2673029630759745e-06,
+      "loss": 0.0694,
+      "step": 3150
+    },
+    {
+      "epoch": 1.7177079079896558,
+      "grad_norm": 0.14168758690357208,
+      "learning_rate": 3.2063780857801596e-06,
+      "loss": 0.0726,
+      "step": 3155
+    },
+    {
+      "epoch": 1.7204301075268817,
+      "grad_norm": 0.14045333862304688,
+      "learning_rate": 3.145987652134563e-06,
+      "loss": 0.0666,
+      "step": 3160
+    },
+    {
+      "epoch": 1.7231523070641077,
+      "grad_norm": 0.15181277692317963,
+      "learning_rate": 3.0861331430900807e-06,
+      "loss": 0.0788,
+      "step": 3165
+    },
+    {
+      "epoch": 1.725874506601334,
+      "grad_norm": 0.14058373868465424,
+      "learning_rate": 3.0268160264551484e-06,
+      "loss": 0.0701,
+      "step": 3170
+    },
+    {
+      "epoch": 1.72859670613856,
+      "grad_norm": 0.1474706381559372,
+      "learning_rate": 2.968037756859776e-06,
+      "loss": 0.0745,
+      "step": 3175
+    },
+    {
+      "epoch": 1.731318905675786,
+      "grad_norm": 0.13882222771644592,
+      "learning_rate": 2.9097997757198516e-06,
+      "loss": 0.0717,
+      "step": 3180
+    },
+    {
+      "epoch": 1.734041105213012,
+      "grad_norm": 0.14754945039749146,
+      "learning_rate": 2.8521035112018063e-06,
+      "loss": 0.0644,
+      "step": 3185
+    },
+    {
+      "epoch": 1.7367633047502382,
+      "grad_norm": 0.1717384308576584,
+      "learning_rate": 2.7949503781875912e-06,
+      "loss": 0.0768,
+      "step": 3190
+    },
+    {
+      "epoch": 1.7394855042874644,
+      "grad_norm": 0.1543627679347992,
+      "learning_rate": 2.7383417782399693e-06,
+      "loss": 0.0709,
+      "step": 3195
+    },
+    {
+      "epoch": 1.7422077038246904,
+      "grad_norm": 0.1619829684495926,
+      "learning_rate": 2.6822790995681577e-06,
+      "loss": 0.0814,
+      "step": 3200
+    },
+    {
+      "epoch": 1.7449299033619163,
+      "grad_norm": 0.13856656849384308,
+      "learning_rate": 2.62676371699378e-06,
+      "loss": 0.0701,
+      "step": 3205
+    },
+    {
+      "epoch": 1.7476521028991425,
+      "grad_norm": 0.14767934381961823,
+      "learning_rate": 2.5717969919171553e-06,
+      "loss": 0.076,
+      "step": 3210
+    },
+    {
+      "epoch": 1.7503743024363687,
+      "grad_norm": 0.1731363832950592,
+      "learning_rate": 2.5173802722838867e-06,
+      "loss": 0.0711,
+      "step": 3215
+    },
+    {
+      "epoch": 1.7530965019735947,
+      "grad_norm": 0.14000795781612396,
+      "learning_rate": 2.4635148925518577e-06,
+      "loss": 0.0675,
+      "step": 3220
+    },
+    {
+      "epoch": 1.7558187015108206,
+      "grad_norm": 0.14177638292312622,
+      "learning_rate": 2.4102021736584575e-06,
+      "loss": 0.0659,
+      "step": 3225
+    },
+    {
+      "epoch": 1.7585409010480468,
+      "grad_norm": 0.1450289785861969,
+      "learning_rate": 2.3574434229882145e-06,
+      "loss": 0.0836,
+      "step": 3230
+    },
+    {
+      "epoch": 1.761263100585273,
+      "grad_norm": 0.15300573408603668,
+      "learning_rate": 2.3052399343407277e-06,
+      "loss": 0.079,
+      "step": 3235
+    },
+    {
+      "epoch": 1.763985300122499,
+      "grad_norm": 0.12552043795585632,
+      "learning_rate": 2.253592987898942e-06,
+      "loss": 0.075,
+      "step": 3240
+    },
+    {
+      "epoch": 1.766707499659725,
+      "grad_norm": 0.13973963260650635,
+      "learning_rate": 2.2025038501977486e-06,
+      "loss": 0.075,
+      "step": 3245
+    },
+    {
+      "epoch": 1.7694296991969511,
+      "grad_norm": 0.14422014355659485,
+      "learning_rate": 2.1519737740929274e-06,
+      "loss": 0.0847,
+      "step": 3250
+    },
+    {
+      "epoch": 1.7721518987341773,
+      "grad_norm": 0.15774239599704742,
+      "learning_rate": 2.1020039987304285e-06,
+      "loss": 0.0852,
+      "step": 3255
+    },
+    {
+      "epoch": 1.7748740982714033,
+      "grad_norm": 0.14551898837089539,
+      "learning_rate": 2.05259574951599e-06,
+      "loss": 0.0726,
+      "step": 3260
+    },
+    {
+      "epoch": 1.7775962978086293,
+      "grad_norm": 0.1303868442773819,
+      "learning_rate": 2.003750238085053e-06,
+      "loss": 0.076,
+      "step": 3265
+    },
+    {
+      "epoch": 1.7803184973458555,
+      "grad_norm": 0.16554243862628937,
+      "learning_rate": 1.9554686622730993e-06,
+      "loss": 0.0992,
+      "step": 3270
+    },
+    {
+      "epoch": 1.7830406968830816,
+      "grad_norm": 0.15280041098594666,
+      "learning_rate": 1.907752206086247e-06,
+      "loss": 0.0819,
+      "step": 3275
+    },
+    {
+      "epoch": 1.7857628964203076,
+      "grad_norm": 0.14237718284130096,
+      "learning_rate": 1.8606020396722129e-06,
+      "loss": 0.0726,
+      "step": 3280
+    },
+    {
+      "epoch": 1.7884850959575336,
+      "grad_norm": 0.15816614031791687,
+      "learning_rate": 1.814019319291635e-06,
+      "loss": 0.0754,
+      "step": 3285
+    },
+    {
+      "epoch": 1.7912072954947598,
+      "grad_norm": 0.13710108399391174,
+      "learning_rate": 1.7680051872896869e-06,
+      "loss": 0.0658,
+      "step": 3290
+    },
+    {
+      "epoch": 1.793929495031986,
+      "grad_norm": 0.15852783620357513,
+      "learning_rate": 1.7225607720681131e-06,
+      "loss": 0.0732,
+      "step": 3295
+    },
+    {
+      "epoch": 1.796651694569212,
+      "grad_norm": 0.15130798518657684,
+      "learning_rate": 1.6776871880575084e-06,
+      "loss": 0.0833,
+      "step": 3300
+    },
+    {
+      "epoch": 1.7993738941064379,
+      "grad_norm": 0.15700189769268036,
+      "learning_rate": 1.6333855356900185e-06,
+      "loss": 0.1652,
+      "step": 3305
+    },
+    {
+      "epoch": 1.802096093643664,
+      "grad_norm": 0.14298690855503082,
+      "learning_rate": 1.5896569013723329e-06,
+      "loss": 0.0797,
+      "step": 3310
+    },
+    {
+      "epoch": 1.8048182931808903,
+      "grad_norm": 0.14160804450511932,
+      "learning_rate": 1.5465023574590676e-06,
+      "loss": 0.0757,
+      "step": 3315
+    },
+    {
+      "epoch": 1.8075404927181162,
+      "grad_norm": 0.15241742134094238,
+      "learning_rate": 1.5039229622264479e-06,
+      "loss": 0.0781,
+      "step": 3320
+    },
+    {
+      "epoch": 1.8102626922553422,
+      "grad_norm": 0.15816597640514374,
+      "learning_rate": 1.4619197598463641e-06,
+      "loss": 0.0764,
+      "step": 3325
+    },
+    {
+      "epoch": 1.8129848917925684,
+      "grad_norm": 0.15082836151123047,
+      "learning_rate": 1.420493780360771e-06,
+      "loss": 0.0708,
+      "step": 3330
+    },
+    {
+      "epoch": 1.8157070913297946,
+      "grad_norm": 0.13604667782783508,
+      "learning_rate": 1.3796460396564098e-06,
+      "loss": 0.0711,
+      "step": 3335
+    },
+    {
+      "epoch": 1.8184292908670205,
+      "grad_norm": 0.14523722231388092,
+      "learning_rate": 1.3393775394399123e-06,
+      "loss": 0.0797,
+      "step": 3340
+    },
+    {
+      "epoch": 1.8211514904042465,
+      "grad_norm": 0.14779233932495117,
+      "learning_rate": 1.2996892672132416e-06,
+      "loss": 0.0845,
+      "step": 3345
+    },
+    {
+      "epoch": 1.8238736899414727,
+      "grad_norm": 0.13566653430461884,
+      "learning_rate": 1.2605821962494574e-06,
+      "loss": 0.1118,
+      "step": 3350
+    },
+    {
+      "epoch": 1.8265958894786989,
+      "grad_norm": 0.15038765966892242,
+      "learning_rate": 1.2220572855688507e-06,
+      "loss": 0.0782,
+      "step": 3355
+    },
+    {
+      "epoch": 1.8293180890159249,
+      "grad_norm": 0.14022061228752136,
+      "learning_rate": 1.1841154799154374e-06,
+      "loss": 0.0737,
+      "step": 3360
+    },
+    {
+      "epoch": 1.8320402885531508,
+      "grad_norm": 0.14298531413078308,
+      "learning_rate": 1.1467577097337867e-06,
+      "loss": 0.0939,
+      "step": 3365
+    },
+    {
+      "epoch": 1.834762488090377,
+      "grad_norm": 0.15060581266880035,
+      "learning_rate": 1.1099848911462014e-06,
+      "loss": 0.0752,
+      "step": 3370
+    },
+    {
+      "epoch": 1.8374846876276032,
+      "grad_norm": 0.1449151188135147,
+      "learning_rate": 1.0737979259302478e-06,
+      "loss": 0.0688,
+      "step": 3375
+    },
+    {
+      "epoch": 1.8402068871648292,
+      "grad_norm": 0.13491316139698029,
+      "learning_rate": 1.0381977014966543e-06,
+      "loss": 0.0996,
+      "step": 3380
+    },
+    {
+      "epoch": 1.8429290867020551,
+      "grad_norm": 0.128363698720932,
+      "learning_rate": 1.0031850908675283e-06,
+      "loss": 0.1217,
+      "step": 3385
+    },
+    {
+      "epoch": 1.8456512862392813,
+      "grad_norm": 0.14816632866859436,
+      "learning_rate": 9.68760952654968e-07,
+      "loss": 0.0653,
+      "step": 3390
+    },
+    {
+      "epoch": 1.8483734857765075,
+      "grad_norm": 0.16936026513576508,
+      "learning_rate": 9.349261310400037e-07,
+      "loss": 0.0735,
+      "step": 3395
+    },
+    {
+      "epoch": 1.8510956853137335,
+      "grad_norm": 0.14492131769657135,
+      "learning_rate": 9.016814557518849e-07,
+      "loss": 0.0683,
+      "step": 3400
+    },
+    {
+      "epoch": 1.8538178848509594,
+      "grad_norm": 0.14610859751701355,
+      "learning_rate": 8.690277420477372e-07,
+      "loss": 0.079,
+      "step": 3405
+    },
+    {
+      "epoch": 1.8565400843881856,
+      "grad_norm": 0.1451197862625122,
+      "learning_rate": 8.369657906925732e-07,
+      "loss": 0.0753,
+      "step": 3410
+    },
+    {
+      "epoch": 1.8592622839254118,
+      "grad_norm": 0.14033135771751404,
+      "learning_rate": 8.054963879396554e-07,
+      "loss": 0.0646,
+      "step": 3415
+    },
+    {
+      "epoch": 1.8619844834626378,
+      "grad_norm": 0.13456685841083527,
+      "learning_rate": 7.746203055112145e-07,
+      "loss": 0.1086,
+      "step": 3420
+    },
+    {
+      "epoch": 1.8647066829998638,
+      "grad_norm": 0.15288980305194855,
+      "learning_rate": 7.443383005795224e-07,
+      "loss": 0.0946,
+      "step": 3425
+    },
+    {
+      "epoch": 1.86742888253709,
+      "grad_norm": 0.14935944974422455,
+      "learning_rate": 7.146511157483216e-07,
+      "loss": 0.0829,
+      "step": 3430
+    },
+    {
+      "epoch": 1.8701510820743161,
+      "grad_norm": 0.14166168868541718,
+      "learning_rate": 6.85559479034617e-07,
+      "loss": 0.0679,
+      "step": 3435
+    },
+    {
+      "epoch": 1.872873281611542,
+      "grad_norm": 0.14961621165275574,
+      "learning_rate": 6.570641038508296e-07,
+      "loss": 0.0762,
+      "step": 3440
+    },
+    {
+      "epoch": 1.875595481148768,
+      "grad_norm": 0.1400919258594513,
+      "learning_rate": 6.291656889873016e-07,
+      "loss": 0.0803,
+      "step": 3445
+    },
+    {
+      "epoch": 1.8783176806859943,
+      "grad_norm": 0.13169647753238678,
+      "learning_rate": 6.018649185951325e-07,
+      "loss": 0.0707,
+      "step": 3450
+    },
+    {
+      "epoch": 1.8810398802232204,
+      "grad_norm": 0.13241390883922577,
+      "learning_rate": 5.751624621694429e-07,
+      "loss": 0.0636,
+      "step": 3455
+    },
+    {
+      "epoch": 1.8837620797604464,
+      "grad_norm": 0.16256216168403625,
+      "learning_rate": 5.490589745329261e-07,
+      "loss": 0.0774,
+      "step": 3460
+    },
+    {
+      "epoch": 1.8864842792976724,
+      "grad_norm": 0.15392127633094788,
+      "learning_rate": 5.235550958198083e-07,
+      "loss": 0.073,
+      "step": 3465
+    },
+    {
+      "epoch": 1.8892064788348986,
+      "grad_norm": 0.14630888402462006,
+      "learning_rate": 4.986514514601415e-07,
+      "loss": 0.0742,
+      "step": 3470
+    },
+    {
+      "epoch": 1.8919286783721248,
+      "grad_norm": 0.16385363042354584,
+      "learning_rate": 4.74348652164458e-07,
+      "loss": 0.083,
+      "step": 3475
+    },
+    {
+      "epoch": 1.8946508779093507,
+      "grad_norm": 0.16106607019901276,
+      "learning_rate": 4.5064729390881246e-07,
+      "loss": 0.0755,
+      "step": 3480
+    },
+    {
+      "epoch": 1.8973730774465767,
+      "grad_norm": 0.13656839728355408,
+      "learning_rate": 4.275479579201602e-07,
+      "loss": 0.0703,
+      "step": 3485
+    },
+    {
+      "epoch": 1.9000952769838029,
+      "grad_norm": 0.12616074085235596,
+      "learning_rate": 4.050512106620913e-07,
+      "loss": 0.0692,
+      "step": 3490
+    },
+    {
+      "epoch": 1.902817476521029,
+      "grad_norm": 0.15891359746456146,
+      "learning_rate": 3.8315760382096057e-07,
+      "loss": 0.0794,
+      "step": 3495
+    },
+    {
+      "epoch": 1.905539676058255,
+      "grad_norm": 0.1297856718301773,
+      "learning_rate": 3.6186767429234323e-07,
+      "loss": 0.0671,
+      "step": 3500
+    },
+    {
+      "epoch": 1.905539676058255,
+      "eval_loss": 0.07686587423086166,
+      "eval_runtime": 218.8338,
+      "eval_samples_per_second": 2.742,
+      "eval_steps_per_second": 0.685,
+      "step": 3500
+    },
+    {
+      "epoch": 1.908261875595481,
+      "grad_norm": 0.16892676055431366,
+      "learning_rate": 3.4118194416786764e-07,
+      "loss": 0.076,
+      "step": 3505
+    },
+    {
+      "epoch": 1.9109840751327072,
+      "grad_norm": 0.14325669407844543,
+      "learning_rate": 3.211009207224225e-07,
+      "loss": 0.0723,
+      "step": 3510
+    },
+    {
+      "epoch": 1.9137062746699334,
+      "grad_norm": 0.12773250043392181,
+      "learning_rate": 3.016250964017142e-07,
+      "loss": 0.0727,
+      "step": 3515
+    },
+    {
+      "epoch": 1.9164284742071593,
+      "grad_norm": 0.15402384102344513,
+      "learning_rate": 2.827549488101849e-07,
+      "loss": 0.0699,
+      "step": 3520
+    },
+    {
+      "epoch": 1.9191506737443853,
+      "grad_norm": 0.13994073867797852,
+      "learning_rate": 2.644909406992996e-07,
+      "loss": 0.0674,
+      "step": 3525
+    },
+    {
+      "epoch": 1.9218728732816115,
+      "grad_norm": 0.15044575929641724,
+      "learning_rate": 2.468335199562105e-07,
+      "loss": 0.0779,
+      "step": 3530
+    },
+    {
+      "epoch": 1.9245950728188377,
+      "grad_norm": 0.13634447753429413,
+      "learning_rate": 2.29783119592758e-07,
+      "loss": 0.0697,
+      "step": 3535
+    },
+    {
+      "epoch": 1.9273172723560636,
+      "grad_norm": 0.1296970546245575,
+      "learning_rate": 2.1334015773486203e-07,
+      "loss": 0.0682,
+      "step": 3540
+    },
+    {
+      "epoch": 1.9300394718932896,
+      "grad_norm": 0.14532914757728577,
+      "learning_rate": 1.975050376122667e-07,
+      "loss": 0.0787,
+      "step": 3545
+    },
+    {
+      "epoch": 1.9327616714305158,
+      "grad_norm": 0.15663999319076538,
+      "learning_rate": 1.8227814754865068e-07,
+      "loss": 0.0735,
+      "step": 3550
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 0.13645371794700623,
+      "learning_rate": 1.6765986095209908e-07,
+      "loss": 0.0684,
+      "step": 3555
+    },
+    {
+      "epoch": 1.938206070504968,
+      "grad_norm": 0.14688655734062195,
+      "learning_rate": 1.536505363059576e-07,
+      "loss": 0.0738,
+      "step": 3560
+    },
+    {
+      "epoch": 1.9409282700421941,
+      "grad_norm": 0.14657242596149445,
+      "learning_rate": 1.4025051716003157e-07,
+      "loss": 0.083,
+      "step": 3565
+    },
+    {
+      "epoch": 1.9436504695794201,
+      "grad_norm": 0.1277332305908203,
+      "learning_rate": 1.2746013212217022e-07,
+      "loss": 0.0758,
+      "step": 3570
+    },
+    {
+      "epoch": 1.9463726691166463,
+      "grad_norm": 0.1419767141342163,
+      "learning_rate": 1.1527969485019275e-07,
+      "loss": 0.0755,
+      "step": 3575
+    },
+    {
+      "epoch": 1.9490948686538725,
+      "grad_norm": 0.15132427215576172,
+      "learning_rate": 1.0370950404421931e-07,
+      "loss": 0.0651,
+      "step": 3580
+    },
+    {
+      "epoch": 1.9518170681910985,
+      "grad_norm": 0.1558247208595276,
+      "learning_rate": 9.274984343932702e-08,
+      "loss": 0.0785,
+      "step": 3585
+    },
+    {
+      "epoch": 1.9545392677283244,
+      "grad_norm": 0.1476377248764038,
+      "learning_rate": 8.240098179859712e-08,
+      "loss": 0.0826,
+      "step": 3590
+    },
+    {
+      "epoch": 1.9572614672655506,
+      "grad_norm": 0.14471116662025452,
+      "learning_rate": 7.26631729065258e-08,
+      "loss": 0.0816,
+      "step": 3595
+    },
+    {
+      "epoch": 1.9599836668027768,
+      "grad_norm": 0.14453163743019104,
+      "learning_rate": 6.353665556280697e-08,
+      "loss": 0.0699,
+      "step": 3600
+    },
+    {
+      "epoch": 1.9627058663400028,
+      "grad_norm": 0.13174843788146973,
+      "learning_rate": 5.5021653576459164e-08,
+      "loss": 0.062,
+      "step": 3605
+    },
+    {
+      "epoch": 1.9654280658772287,
+      "grad_norm": 0.1312907189130783,
+      "learning_rate": 4.7118375760357716e-08,
+      "loss": 0.0804,
+      "step": 3610
+    },
+    {
+      "epoch": 1.968150265414455,
+      "grad_norm": 0.15283526480197906,
+      "learning_rate": 3.982701592609439e-08,
+      "loss": 0.1359,
+      "step": 3615
+    },
+    {
+      "epoch": 1.9708724649516811,
+      "grad_norm": 0.14205031096935272,
+      "learning_rate": 3.314775287923677e-08,
+      "loss": 0.071,
+      "step": 3620
+    },
+    {
+      "epoch": 1.973594664488907,
+      "grad_norm": 0.16160933673381805,
+      "learning_rate": 2.708075041494562e-08,
+      "loss": 0.0784,
+      "step": 3625
+    },
+    {
+      "epoch": 1.976316864026133,
+      "grad_norm": 0.12819606065750122,
+      "learning_rate": 2.1626157313950345e-08,
+      "loss": 0.097,
+      "step": 3630
+    },
+    {
+      "epoch": 1.9790390635633592,
+      "grad_norm": 0.1536770612001419,
+      "learning_rate": 1.6784107338899124e-08,
+      "loss": 0.0775,
+      "step": 3635
+    },
+    {
+      "epoch": 1.9817612631005854,
+      "grad_norm": 0.12998686730861664,
+      "learning_rate": 1.2554719231083755e-08,
+      "loss": 0.0696,
+      "step": 3640
+    },
+    {
+      "epoch": 1.9844834626378114,
+      "grad_norm": 0.11487606167793274,
+      "learning_rate": 8.93809670753365e-09,
+      "loss": 0.0674,
+      "step": 3645
+    },
+    {
+      "epoch": 1.9872056621750374,
+      "grad_norm": 0.13673199713230133,
+      "learning_rate": 5.934328458459537e-09,
+      "loss": 0.0712,
+      "step": 3650
+    },
+    {
+      "epoch": 1.9899278617122635,
+      "grad_norm": 0.1361059844493866,
+      "learning_rate": 3.543488145082985e-09,
+      "loss": 0.0614,
+      "step": 3655
+    },
+    {
+      "epoch": 1.9926500612494897,
+      "grad_norm": 0.15082420408725739,
+      "learning_rate": 1.7656343978378342e-09,
+      "loss": 0.0762,
+      "step": 3660
+    },
+    {
+      "epoch": 1.9953722607867157,
+      "grad_norm": 0.1459612250328064,
+      "learning_rate": 6.008108149185843e-10,
+      "loss": 0.0841,
+      "step": 3665
+    },
+    {
+      "epoch": 1.9980944603239417,
+      "grad_norm": 0.15308088064193726,
+      "learning_rate": 4.904596122290439e-11,
+      "loss": 0.0805,
+      "step": 3670
+    },
+    {
+      "epoch": 1.9991833401388321,
+      "step": 3672,
+      "total_flos": 2.724210105233965e+18,
+      "train_loss": 0.10618524583693684,
+      "train_runtime": 81161.5668,
+      "train_samples_per_second": 0.724,
+      "train_steps_per_second": 0.045
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3672,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.724210105233965e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}