diff --git "a/checkpoint-1130/trainer_state.json" "b/checkpoint-1130/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1130/trainer_state.json"
@@ -0,0 +1,7995 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9991154356479433,
+  "eval_steps": 142,
+  "global_step": 1130,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.680586814880371,
+      "learning_rate": 1e-05,
+      "loss": 3.3182,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 3.3362529277801514,
+      "eval_runtime": 14.4477,
+      "eval_samples_per_second": 33.016,
+      "eval_steps_per_second": 8.306,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.609802722930908,
+      "learning_rate": 2e-05,
+      "loss": 3.2788,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.793943405151367,
+      "learning_rate": 3e-05,
+      "loss": 3.3432,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.687256336212158,
+      "learning_rate": 4e-05,
+      "loss": 3.2521,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.676945209503174,
+      "learning_rate": 5e-05,
+      "loss": 3.1085,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.490086078643799,
+      "learning_rate": 6e-05,
+      "loss": 2.8093,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.042544364929199,
+      "learning_rate": 7e-05,
+      "loss": 2.3501,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4973549842834473,
+      "learning_rate": 8e-05,
+      "loss": 1.6118,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9255049228668213,
+      "learning_rate": 9e-05,
+      "loss": 0.9938,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9458708763122559,
+      "learning_rate": 0.0001,
+      "loss": 0.4821,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3317277431488037,
+      "learning_rate": 9.999991309598974e-05,
+      "loss": 0.3336,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7918155193328857,
+      "learning_rate": 9.999965238426104e-05,
+      "loss": 0.1707,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.779201090335846,
+      "learning_rate": 9.999921786572015e-05,
+      "loss": 0.1089,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8333582878112793,
+      "learning_rate": 9.999860954187756e-05,
+      "loss": 0.1829,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7508969902992249,
+      "learning_rate": 9.999782741484788e-05,
+      "loss": 0.1284,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6024438142776489,
+      "learning_rate": 9.999687148734995e-05,
+      "loss": 0.1321,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2141278237104416,
+      "learning_rate": 9.999574176270667e-05,
+      "loss": 0.1294,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8326414227485657,
+      "learning_rate": 9.999443824484519e-05,
+      "loss": 0.1414,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.25254812836647034,
+      "learning_rate": 9.999296093829672e-05,
+      "loss": 0.1389,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.24321849644184113,
+      "learning_rate": 9.999130984819662e-05,
+      "loss": 0.1354,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.27592459321022034,
+      "learning_rate": 9.998948498028435e-05,
+      "loss": 0.1032,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5619893670082092,
+      "learning_rate": 9.998748634090344e-05,
+      "loss": 0.1264,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7668437361717224,
+      "learning_rate": 9.998531393700148e-05,
+      "loss": 0.1223,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.38381776213645935,
+      "learning_rate": 9.99829677761301e-05,
+      "loss": 0.0988,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8528074026107788,
+      "learning_rate": 9.998044786644491e-05,
+      "loss": 0.1421,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.61940336227417,
+      "learning_rate": 9.997775421670556e-05,
+      "loss": 0.2738,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5642948150634766,
+      "learning_rate": 9.997488683627559e-05,
+      "loss": 0.1113,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.31713128089904785,
+      "learning_rate": 9.997184573512245e-05,
+      "loss": 0.0593,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.540770411491394,
+      "learning_rate": 9.996863092381752e-05,
+      "loss": 0.2008,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5343081951141357,
+      "learning_rate": 9.9965242413536e-05,
+      "loss": 0.1141,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3453178107738495,
+      "learning_rate": 9.99616802160569e-05,
+      "loss": 0.1137,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.24307988584041595,
+      "learning_rate": 9.995794434376297e-05,
+      "loss": 0.0971,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1228247806429863,
+      "learning_rate": 9.995403480964072e-05,
+      "loss": 0.1246,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.28919702768325806,
+      "learning_rate": 9.994995162728029e-05,
+      "loss": 0.1485,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.16266010701656342,
+      "learning_rate": 9.994569481087552e-05,
+      "loss": 0.1196,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.13942277431488037,
+      "learning_rate": 9.994126437522375e-05,
+      "loss": 0.1266,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.22137008607387543,
+      "learning_rate": 9.99366603357259e-05,
+      "loss": 0.1226,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4418635070323944,
+      "learning_rate": 9.993188270838635e-05,
+      "loss": 0.1577,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1568412482738495,
+      "learning_rate": 9.992693150981292e-05,
+      "loss": 0.1205,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.33117881417274475,
+      "learning_rate": 9.992180675721672e-05,
+      "loss": 0.1179,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.18135391175746918,
+      "learning_rate": 9.991650846841226e-05,
+      "loss": 0.098,
+      "step": 41
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.23688584566116333,
+      "learning_rate": 9.99110366618172e-05,
+      "loss": 0.0839,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.40056225657463074,
+      "learning_rate": 9.990539135645245e-05,
+      "loss": 0.1648,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.16901437938213348,
+      "learning_rate": 9.9899572571942e-05,
+      "loss": 0.052,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6529514193534851,
+      "learning_rate": 9.989358032851284e-05,
+      "loss": 0.1448,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.25050362944602966,
+      "learning_rate": 9.9887414646995e-05,
+      "loss": 0.083,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.38230955600738525,
+      "learning_rate": 9.988107554882138e-05,
+      "loss": 0.0912,
+      "step": 47
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.24738825857639313,
+      "learning_rate": 9.987456305602769e-05,
+      "loss": 0.1337,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.22692906856536865,
+      "learning_rate": 9.986787719125241e-05,
+      "loss": 0.0924,
+      "step": 49
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6348592638969421,
+      "learning_rate": 9.986101797773667e-05,
+      "loss": 0.1401,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.27844250202178955,
+      "learning_rate": 9.985398543932421e-05,
+      "loss": 0.1028,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4353552460670471,
+      "learning_rate": 9.984677960046123e-05,
+      "loss": 0.1245,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3049747943878174,
+      "learning_rate": 9.98394004861964e-05,
+      "loss": 0.0936,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.34531188011169434,
+      "learning_rate": 9.983184812218072e-05,
+      "loss": 0.0775,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.252056360244751,
+      "learning_rate": 9.98241225346674e-05,
+      "loss": 0.1082,
+      "step": 55
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.14545601606369019,
+      "learning_rate": 9.981622375051183e-05,
+      "loss": 0.0857,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.313376784324646,
+      "learning_rate": 9.980815179717145e-05,
+      "loss": 0.0798,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3164367079734802,
+      "learning_rate": 9.979990670270564e-05,
+      "loss": 0.1103,
+      "step": 58
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.30412447452545166,
+      "learning_rate": 9.979148849577572e-05,
+      "loss": 0.0889,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2582318186759949,
+      "learning_rate": 9.978289720564471e-05,
+      "loss": 0.0844,
+      "step": 60
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2912735939025879,
+      "learning_rate": 9.977413286217728e-05,
+      "loss": 0.077,
+      "step": 61
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3888057768344879,
+      "learning_rate": 9.976519549583974e-05,
+      "loss": 0.1386,
+      "step": 62
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42121654748916626,
+      "learning_rate": 9.975608513769976e-05,
+      "loss": 0.0823,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4052259624004364,
+      "learning_rate": 9.974680181942645e-05,
+      "loss": 0.0846,
+      "step": 64
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2273682802915573,
+      "learning_rate": 9.973734557329009e-05,
+      "loss": 0.0589,
+      "step": 65
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5051669478416443,
+      "learning_rate": 9.972771643216212e-05,
+      "loss": 0.1111,
+      "step": 66
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2672370970249176,
+      "learning_rate": 9.971791442951497e-05,
+      "loss": 0.0819,
+      "step": 67
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.28557881712913513,
+      "learning_rate": 9.970793959942198e-05,
+      "loss": 0.0912,
+      "step": 68
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.34148702025413513,
+      "learning_rate": 9.969779197655726e-05,
+      "loss": 0.1036,
+      "step": 69
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3550778925418854,
+      "learning_rate": 9.968747159619556e-05,
+      "loss": 0.0833,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3434258699417114,
+      "learning_rate": 9.967697849421221e-05,
+      "loss": 0.1186,
+      "step": 71
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.23545867204666138,
+      "learning_rate": 9.966631270708287e-05,
+      "loss": 0.1185,
+      "step": 72
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.28094542026519775,
+      "learning_rate": 9.965547427188357e-05,
+      "loss": 0.052,
+      "step": 73
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13517600297927856,
+      "learning_rate": 9.964446322629043e-05,
+      "loss": 0.0695,
+      "step": 74
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.16696467995643616,
+      "learning_rate": 9.963327960857962e-05,
+      "loss": 0.1003,
+      "step": 75
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.18569788336753845,
+      "learning_rate": 9.962192345762717e-05,
+      "loss": 0.0495,
+      "step": 76
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.19817449152469635,
+      "learning_rate": 9.961039481290888e-05,
+      "loss": 0.067,
+      "step": 77
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.14672966301441193,
+      "learning_rate": 9.959869371450021e-05,
+      "loss": 0.0737,
+      "step": 78
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.302121639251709,
+      "learning_rate": 9.958682020307601e-05,
+      "loss": 0.0779,
+      "step": 79
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.35970303416252136,
+      "learning_rate": 9.957477431991054e-05,
+      "loss": 0.134,
+      "step": 80
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.308292955160141,
+      "learning_rate": 9.956255610687719e-05,
+      "loss": 0.1006,
+      "step": 81
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.27124735713005066,
+      "learning_rate": 9.955016560644847e-05,
+      "loss": 0.0572,
+      "step": 82
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1820615977048874,
+      "learning_rate": 9.953760286169571e-05,
+      "loss": 0.0595,
+      "step": 83
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.36385104060173035,
+      "learning_rate": 9.952486791628905e-05,
+      "loss": 0.0874,
+      "step": 84
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.848340630531311,
+      "learning_rate": 9.95119608144972e-05,
+      "loss": 0.1178,
+      "step": 85
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7947489023208618,
+      "learning_rate": 9.94988816011873e-05,
+      "loss": 0.1115,
+      "step": 86
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.33932074904441833,
+      "learning_rate": 9.94856303218248e-05,
+      "loss": 0.0546,
+      "step": 87
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.26873940229415894,
+      "learning_rate": 9.947220702247329e-05,
+      "loss": 0.0873,
+      "step": 88
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3373044431209564,
+      "learning_rate": 9.945861174979429e-05,
+      "loss": 0.1051,
+      "step": 89
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.24391719698905945,
+      "learning_rate": 9.944484455104717e-05,
+      "loss": 0.0986,
+      "step": 90
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4301680326461792,
+      "learning_rate": 9.943090547408888e-05,
+      "loss": 0.1524,
+      "step": 91
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4246341288089752,
+      "learning_rate": 9.941679456737394e-05,
+      "loss": 0.1619,
+      "step": 92
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2680893838405609,
+      "learning_rate": 9.940251187995411e-05,
+      "loss": 0.1187,
+      "step": 93
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.18920297920703888,
+      "learning_rate": 9.938805746147827e-05,
+      "loss": 0.105,
+      "step": 94
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.22168701887130737,
+      "learning_rate": 9.937343136219233e-05,
+      "loss": 0.0856,
+      "step": 95
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2235199213027954,
+      "learning_rate": 9.935863363293896e-05,
+      "loss": 0.1026,
+      "step": 96
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.16578496992588043,
+      "learning_rate": 9.93436643251574e-05,
+      "loss": 0.0777,
+      "step": 97
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.15994016826152802,
+      "learning_rate": 9.932852349088342e-05,
+      "loss": 0.0957,
+      "step": 98
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.18692170083522797,
+      "learning_rate": 9.931321118274897e-05,
+      "loss": 0.0913,
+      "step": 99
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.15477485954761505,
+      "learning_rate": 9.929772745398206e-05,
+      "loss": 0.0911,
+      "step": 100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.28473320603370667,
+      "learning_rate": 9.928207235840664e-05,
+      "loss": 0.1283,
+      "step": 101
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38557159900665283,
+      "learning_rate": 9.926624595044234e-05,
+      "loss": 0.1125,
+      "step": 102
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13523289561271667,
+      "learning_rate": 9.925024828510427e-05,
+      "loss": 0.0555,
+      "step": 103
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.15305563807487488,
+      "learning_rate": 9.923407941800291e-05,
+      "loss": 0.1003,
+      "step": 104
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.21130621433258057,
+      "learning_rate": 9.921773940534382e-05,
+      "loss": 0.0945,
+      "step": 105
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2301904559135437,
+      "learning_rate": 9.920122830392748e-05,
+      "loss": 0.1019,
+      "step": 106
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.16425654292106628,
+      "learning_rate": 9.918454617114918e-05,
+      "loss": 0.0781,
+      "step": 107
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2672991454601288,
+      "learning_rate": 9.916769306499866e-05,
+      "loss": 0.085,
+      "step": 108
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3746218681335449,
+      "learning_rate": 9.915066904406e-05,
+      "loss": 0.1698,
+      "step": 109
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1691233068704605,
+      "learning_rate": 9.913347416751148e-05,
+      "loss": 0.046,
+      "step": 110
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3089153468608856,
+      "learning_rate": 9.91161084951252e-05,
+      "loss": 0.131,
+      "step": 111
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1581045240163803,
+      "learning_rate": 9.909857208726705e-05,
+      "loss": 0.0654,
+      "step": 112
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2545772194862366,
+      "learning_rate": 9.908086500489637e-05,
+      "loss": 0.1021,
+      "step": 113
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2257249355316162,
+      "learning_rate": 9.906298730956586e-05,
+      "loss": 0.0636,
+      "step": 114
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.17862719297409058,
+      "learning_rate": 9.904493906342123e-05,
+      "loss": 0.0942,
+      "step": 115
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.23423053324222565,
+      "learning_rate": 9.902672032920106e-05,
+      "loss": 0.0676,
+      "step": 116
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.26653358340263367,
+      "learning_rate": 9.900833117023664e-05,
+      "loss": 0.0918,
+      "step": 117
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4517073631286621,
+      "learning_rate": 9.89897716504516e-05,
+      "loss": 0.1102,
+      "step": 118
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.20187437534332275,
+      "learning_rate": 9.897104183436183e-05,
+      "loss": 0.0713,
+      "step": 119
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5759711861610413,
+      "learning_rate": 9.895214178707516e-05,
+      "loss": 0.0837,
+      "step": 120
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.43704915046691895,
+      "learning_rate": 9.89330715742912e-05,
+      "loss": 0.0868,
+      "step": 121
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.30784374475479126,
+      "learning_rate": 9.891383126230104e-05,
+      "loss": 0.1171,
+      "step": 122
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.23538921773433685,
+      "learning_rate": 9.889442091798712e-05,
+      "loss": 0.055,
+      "step": 123
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.27727362513542175,
+      "learning_rate": 9.887484060882291e-05,
+      "loss": 0.041,
+      "step": 124
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.21666617691516876,
+      "learning_rate": 9.885509040287268e-05,
+      "loss": 0.0624,
+      "step": 125
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2829636335372925,
+      "learning_rate": 9.883517036879132e-05,
+      "loss": 0.0946,
+      "step": 126
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.34035512804985046,
+      "learning_rate": 9.88150805758241e-05,
+      "loss": 0.0635,
+      "step": 127
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.44064444303512573,
+      "learning_rate": 9.879482109380634e-05,
+      "loss": 0.0931,
+      "step": 128
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5164741277694702,
+      "learning_rate": 9.877439199316323e-05,
+      "loss": 0.0891,
+      "step": 129
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5549228191375732,
+      "learning_rate": 9.875379334490962e-05,
+      "loss": 0.1144,
+      "step": 130
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41133901476860046,
+      "learning_rate": 9.873302522064972e-05,
+      "loss": 0.1022,
+      "step": 131
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5204330682754517,
+      "learning_rate": 9.871208769257685e-05,
+      "loss": 0.0867,
+      "step": 132
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3383274972438812,
+      "learning_rate": 9.869098083347323e-05,
+      "loss": 0.0558,
+      "step": 133
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7084139585494995,
+      "learning_rate": 9.866970471670967e-05,
+      "loss": 0.1208,
+      "step": 134
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2974587380886078,
+      "learning_rate": 9.864825941624537e-05,
+      "loss": 0.1199,
+      "step": 135
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1811504065990448,
+      "learning_rate": 9.862664500662764e-05,
+      "loss": 0.1025,
+      "step": 136
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.23748300969600677,
+      "learning_rate": 9.860486156299164e-05,
+      "loss": 0.0864,
+      "step": 137
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.21784676611423492,
+      "learning_rate": 9.85829091610601e-05,
+      "loss": 0.095,
+      "step": 138
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.26979225873947144,
+      "learning_rate": 9.856078787714309e-05,
+      "loss": 0.0864,
+      "step": 139
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1479984074831009,
+      "learning_rate": 9.853849778813777e-05,
+      "loss": 0.0904,
+      "step": 140
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.17924343049526215,
+      "learning_rate": 9.851603897152803e-05,
+      "loss": 0.0752,
+      "step": 141
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.16448016464710236,
+      "learning_rate": 9.849341150538434e-05,
+      "loss": 0.0781,
+      "step": 142
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.0848281979560852,
+      "eval_runtime": 14.6961,
+      "eval_samples_per_second": 32.458,
+      "eval_steps_per_second": 8.165,
+      "step": 142
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.14405055344104767,
+      "learning_rate": 9.847061546836339e-05,
+      "loss": 0.1007,
+      "step": 143
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.20907168090343475,
+      "learning_rate": 9.844765093970787e-05,
+      "loss": 0.1126,
+      "step": 144
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.1777975857257843,
+      "learning_rate": 9.842451799924616e-05,
+      "loss": 0.0928,
+      "step": 145
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.1817995309829712,
+      "learning_rate": 9.840121672739208e-05,
+      "loss": 0.046,
+      "step": 146
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2099136859178543,
+      "learning_rate": 9.837774720514457e-05,
+      "loss": 0.1032,
+      "step": 147
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.19467169046401978,
+      "learning_rate": 9.835410951408748e-05,
+      "loss": 0.0913,
+      "step": 148
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.24700500071048737,
+      "learning_rate": 9.833030373638919e-05,
+      "loss": 0.1101,
+      "step": 149
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22854459285736084,
+      "learning_rate": 9.830632995480242e-05,
+      "loss": 0.0729,
+      "step": 150
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.206742063164711,
+      "learning_rate": 9.828218825266388e-05,
+      "loss": 0.0861,
+      "step": 151
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40378740429878235,
+      "learning_rate": 9.8257878713894e-05,
+      "loss": 0.0948,
+      "step": 152
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5055291652679443,
+      "learning_rate": 9.823340142299662e-05,
+      "loss": 0.193,
+      "step": 153
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3036790192127228,
+      "learning_rate": 9.820875646505874e-05,
+      "loss": 0.0859,
+      "step": 154
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1878231018781662,
+      "learning_rate": 9.818394392575019e-05,
+      "loss": 0.0702,
+      "step": 155
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.17990007996559143,
+      "learning_rate": 9.815896389132333e-05,
+      "loss": 0.0967,
+      "step": 156
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.22680750489234924,
+      "learning_rate": 9.813381644861277e-05,
+      "loss": 0.0959,
+      "step": 157
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.299663782119751,
+      "learning_rate": 9.810850168503506e-05,
+      "loss": 0.0801,
+      "step": 158
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3132835924625397,
+      "learning_rate": 9.808301968858837e-05,
+      "loss": 0.1151,
+      "step": 159
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.16891297698020935,
+      "learning_rate": 9.805737054785222e-05,
+      "loss": 0.0799,
+      "step": 160
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.23542983829975128,
+      "learning_rate": 9.803155435198712e-05,
+      "loss": 0.0645,
+      "step": 161
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1784803569316864,
+      "learning_rate": 9.800557119073433e-05,
+      "loss": 0.0475,
+      "step": 162
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.37001606822013855,
+      "learning_rate": 9.797942115441545e-05,
+      "loss": 0.1331,
+      "step": 163
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.18926851451396942,
+      "learning_rate": 9.795310433393226e-05,
+      "loss": 0.0744,
+      "step": 164
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.15572589635849,
+      "learning_rate": 9.792662082076618e-05,
+      "loss": 0.0551,
+      "step": 165
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2562514841556549,
+      "learning_rate": 9.789997070697821e-05,
+      "loss": 0.106,
+      "step": 166
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4756919741630554,
+      "learning_rate": 9.787315408520838e-05,
+      "loss": 0.1229,
+      "step": 167
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2322833091020584,
+      "learning_rate": 9.78461710486756e-05,
+      "loss": 0.1212,
+      "step": 168
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.28180065751075745,
+      "learning_rate": 9.78190216911772e-05,
+      "loss": 0.0855,
+      "step": 169
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2058788239955902,
+      "learning_rate": 9.779170610708872e-05,
+      "loss": 0.0445,
+      "step": 170
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.16885802149772644,
+      "learning_rate": 9.776422439136352e-05,
+      "loss": 0.0619,
+      "step": 171
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2562006413936615,
+      "learning_rate": 9.773657663953243e-05,
+      "loss": 0.1111,
+      "step": 172
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2394249439239502,
+      "learning_rate": 9.770876294770349e-05,
+      "loss": 0.0562,
+      "step": 173
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.214800164103508,
+      "learning_rate": 9.768078341256155e-05,
+      "loss": 0.0428,
+      "step": 174
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.30056923627853394,
+      "learning_rate": 9.765263813136796e-05,
+      "loss": 0.1173,
+      "step": 175
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3174525499343872,
+      "learning_rate": 9.762432720196024e-05,
+      "loss": 0.0871,
+      "step": 176
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.31560906767845154,
+      "learning_rate": 9.75958507227517e-05,
+      "loss": 0.1133,
+      "step": 177
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.21752357482910156,
+      "learning_rate": 9.756720879273117e-05,
+      "loss": 0.0421,
+      "step": 178
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3215218484401703,
+      "learning_rate": 9.753840151146259e-05,
+      "loss": 0.0596,
+      "step": 179
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3161137104034424,
+      "learning_rate": 9.750942897908468e-05,
+      "loss": 0.122,
+      "step": 180
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4206744134426117,
+      "learning_rate": 9.748029129631062e-05,
+      "loss": 0.0966,
+      "step": 181
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.28242579102516174,
+      "learning_rate": 9.745098856442768e-05,
+      "loss": 0.0853,
+      "step": 182
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.24647079408168793,
+      "learning_rate": 9.742152088529684e-05,
+      "loss": 0.1077,
+      "step": 183
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.29940977692604065,
+      "learning_rate": 9.739188836135247e-05,
+      "loss": 0.0837,
+      "step": 184
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.21811984479427338,
+      "learning_rate": 9.7362091095602e-05,
+      "loss": 0.1,
+      "step": 185
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.24434742331504822,
+      "learning_rate": 9.733212919162549e-05,
+      "loss": 0.0839,
+      "step": 186
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1656690537929535,
+      "learning_rate": 9.730200275357535e-05,
+      "loss": 0.0894,
+      "step": 187
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.16984042525291443,
+      "learning_rate": 9.727171188617587e-05,
+      "loss": 0.0732,
+      "step": 188
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.19889003038406372,
+      "learning_rate": 9.7241256694723e-05,
+      "loss": 0.0832,
+      "step": 189
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1639273464679718,
+      "learning_rate": 9.721063728508383e-05,
+      "loss": 0.0912,
+      "step": 190
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.26211172342300415,
+      "learning_rate": 9.717985376369639e-05,
+      "loss": 0.0986,
+      "step": 191
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12403538823127747,
+      "learning_rate": 9.714890623756912e-05,
+      "loss": 0.0844,
+      "step": 192
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1882586032152176,
+      "learning_rate": 9.711779481428057e-05,
+      "loss": 0.1163,
+      "step": 193
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1780715435743332,
+      "learning_rate": 9.708651960197904e-05,
+      "loss": 0.1038,
+      "step": 194
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1291002333164215,
+      "learning_rate": 9.705508070938218e-05,
+      "loss": 0.0746,
+      "step": 195
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2384466975927353,
+      "learning_rate": 9.702347824577666e-05,
+      "loss": 0.0909,
+      "step": 196
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.25463247299194336,
+      "learning_rate": 9.699171232101768e-05,
+      "loss": 0.0977,
+      "step": 197
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.19303986430168152,
+      "learning_rate": 9.69597830455287e-05,
+      "loss": 0.1137,
+      "step": 198
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.21899022161960602,
+      "learning_rate": 9.692769053030099e-05,
+      "loss": 0.0671,
+      "step": 199
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.16923530399799347,
+      "learning_rate": 9.689543488689332e-05,
+      "loss": 0.0776,
+      "step": 200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.260955810546875,
+      "learning_rate": 9.686301622743144e-05,
+      "loss": 0.092,
+      "step": 201
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1771455854177475,
+      "learning_rate": 9.683043466460782e-05,
+      "loss": 0.06,
+      "step": 202
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.18851810693740845,
+      "learning_rate": 9.67976903116812e-05,
+      "loss": 0.0965,
+      "step": 203
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.589522659778595,
+      "learning_rate": 9.676478328247622e-05,
+      "loss": 0.1673,
+      "step": 204
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21747058629989624,
+      "learning_rate": 9.673171369138296e-05,
+      "loss": 0.0997,
+      "step": 205
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4168107807636261,
+      "learning_rate": 9.669848165335666e-05,
+      "loss": 0.0795,
+      "step": 206
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6520416140556335,
+      "learning_rate": 9.666508728391719e-05,
+      "loss": 0.1177,
+      "step": 207
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3752453029155731,
+      "learning_rate": 9.663153069914875e-05,
+      "loss": 0.0871,
+      "step": 208
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.26546868681907654,
+      "learning_rate": 9.65978120156994e-05,
+      "loss": 0.0647,
+      "step": 209
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20044176280498505,
+      "learning_rate": 9.656393135078068e-05,
+      "loss": 0.1072,
+      "step": 210
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.25033503770828247,
+      "learning_rate": 9.652988882216724e-05,
+      "loss": 0.1326,
+      "step": 211
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.24569682776927948,
+      "learning_rate": 9.649568454819637e-05,
+      "loss": 0.0931,
+      "step": 212
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5623157620429993,
+      "learning_rate": 9.64613186477676e-05,
+      "loss": 0.2157,
+      "step": 213
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.15052182972431183,
+      "learning_rate": 9.642679124034233e-05,
+      "loss": 0.1236,
+      "step": 214
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.14209671318531036,
+      "learning_rate": 9.639210244594334e-05,
+      "loss": 0.0971,
+      "step": 215
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1627768725156784,
+      "learning_rate": 9.635725238515445e-05,
+      "loss": 0.1161,
+      "step": 216
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.10190293937921524,
+      "learning_rate": 9.63222411791201e-05,
+      "loss": 0.0999,
+      "step": 217
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.11575043201446533,
+      "learning_rate": 9.62870689495448e-05,
+      "loss": 0.0986,
+      "step": 218
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.09842410683631897,
+      "learning_rate": 9.62517358186929e-05,
+      "loss": 0.1176,
+      "step": 219
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.14816004037857056,
+      "learning_rate": 9.621624190938803e-05,
+      "loss": 0.0833,
+      "step": 220
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.11311839520931244,
+      "learning_rate": 9.618058734501269e-05,
+      "loss": 0.0815,
+      "step": 221
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.17481163144111633,
+      "learning_rate": 9.614477224950789e-05,
+      "loss": 0.0678,
+      "step": 222
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2277013659477234,
+      "learning_rate": 9.610879674737264e-05,
+      "loss": 0.0941,
+      "step": 223
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.14689870178699493,
+      "learning_rate": 9.607266096366352e-05,
+      "loss": 0.0991,
+      "step": 224
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.24558769166469574,
+      "learning_rate": 9.603636502399436e-05,
+      "loss": 0.0878,
+      "step": 225
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.1541660875082016,
+      "learning_rate": 9.599990905453567e-05,
+      "loss": 0.0784,
+      "step": 226
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12188339233398438,
+      "learning_rate": 9.59632931820142e-05,
+      "loss": 0.0464,
+      "step": 227
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.32710394263267517,
+      "learning_rate": 9.592651753371265e-05,
+      "loss": 0.0541,
+      "step": 228
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3118465840816498,
+      "learning_rate": 9.588958223746903e-05,
+      "loss": 0.0845,
+      "step": 229
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26805219054222107,
+      "learning_rate": 9.585248742167639e-05,
+      "loss": 0.0485,
+      "step": 230
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7972936630249023,
+      "learning_rate": 9.581523321528223e-05,
+      "loss": 0.1013,
+      "step": 231
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6285438537597656,
+      "learning_rate": 9.577781974778817e-05,
+      "loss": 0.0767,
+      "step": 232
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6384493708610535,
+      "learning_rate": 9.57402471492494e-05,
+      "loss": 0.1855,
+      "step": 233
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5759001970291138,
+      "learning_rate": 9.570251555027432e-05,
+      "loss": 0.1585,
+      "step": 234
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.42002353072166443,
+      "learning_rate": 9.566462508202402e-05,
+      "loss": 0.1479,
+      "step": 235
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.18405884504318237,
+      "learning_rate": 9.562657587621184e-05,
+      "loss": 0.09,
+      "step": 236
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20893922448158264,
+      "learning_rate": 9.558836806510291e-05,
+      "loss": 0.0685,
+      "step": 237
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.31388092041015625,
+      "learning_rate": 9.555000178151374e-05,
+      "loss": 0.0983,
+      "step": 238
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20344533026218414,
+      "learning_rate": 9.551147715881166e-05,
+      "loss": 0.0944,
+      "step": 239
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1582648903131485,
+      "learning_rate": 9.547279433091446e-05,
+      "loss": 0.0662,
+      "step": 240
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.16737405955791473,
+      "learning_rate": 9.543395343228983e-05,
+      "loss": 0.1565,
+      "step": 241
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.21974924206733704,
+      "learning_rate": 9.539495459795499e-05,
+      "loss": 0.1243,
+      "step": 242
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1147058829665184,
+      "learning_rate": 9.535579796347612e-05,
+      "loss": 0.0727,
+      "step": 243
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13460345566272736,
+      "learning_rate": 9.531648366496799e-05,
+      "loss": 0.0691,
+      "step": 244
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1404263824224472,
+      "learning_rate": 9.527701183909336e-05,
+      "loss": 0.0975,
+      "step": 245
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.17380090057849884,
+      "learning_rate": 9.523738262306269e-05,
+      "loss": 0.0873,
+      "step": 246
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13862797617912292,
+      "learning_rate": 9.519759615463346e-05,
+      "loss": 0.0738,
+      "step": 247
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.17551685869693756,
+      "learning_rate": 9.51576525721098e-05,
+      "loss": 0.0676,
+      "step": 248
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.20715269446372986,
+      "learning_rate": 9.511755201434205e-05,
+      "loss": 0.0737,
+      "step": 249
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.14763356745243073,
+      "learning_rate": 9.507729462072614e-05,
+      "loss": 0.07,
+      "step": 250
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.317452073097229,
+      "learning_rate": 9.503688053120327e-05,
+      "loss": 0.1252,
+      "step": 251
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.21908459067344666,
+      "learning_rate": 9.499630988625925e-05,
+      "loss": 0.0877,
+      "step": 252
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3233601450920105,
+      "learning_rate": 9.49555828269242e-05,
+      "loss": 0.0891,
+      "step": 253
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4098372161388397,
+      "learning_rate": 9.491469949477187e-05,
+      "loss": 0.0805,
+      "step": 254
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40573808550834656,
+      "learning_rate": 9.487366003191931e-05,
+      "loss": 0.1284,
+      "step": 255
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3391616940498352,
+      "learning_rate": 9.483246458102625e-05,
+      "loss": 0.0901,
+      "step": 256
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1822938323020935,
+      "learning_rate": 9.479111328529473e-05,
+      "loss": 0.0398,
+      "step": 257
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4700302183628082,
+      "learning_rate": 9.474960628846843e-05,
+      "loss": 0.1509,
+      "step": 258
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.20210890471935272,
+      "learning_rate": 9.470794373483236e-05,
+      "loss": 0.0765,
+      "step": 259
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.28329914808273315,
+      "learning_rate": 9.466612576921223e-05,
+      "loss": 0.0666,
+      "step": 260
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.41083166003227234,
+      "learning_rate": 9.462415253697401e-05,
+      "loss": 0.1248,
+      "step": 261
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.17644570767879486,
+      "learning_rate": 9.458202418402338e-05,
+      "loss": 0.0532,
+      "step": 262
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2667219936847687,
+      "learning_rate": 9.453974085680526e-05,
+      "loss": 0.0937,
+      "step": 263
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.20900332927703857,
+      "learning_rate": 9.449730270230326e-05,
+      "loss": 0.0853,
+      "step": 264
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.26425743103027344,
+      "learning_rate": 9.445470986803922e-05,
+      "loss": 0.12,
+      "step": 265
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1956167221069336,
+      "learning_rate": 9.441196250207267e-05,
+      "loss": 0.0965,
+      "step": 266
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.21896903216838837,
+      "learning_rate": 9.436906075300032e-05,
+      "loss": 0.0867,
+      "step": 267
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2082919031381607,
+      "learning_rate": 9.432600476995551e-05,
+      "loss": 0.0847,
+      "step": 268
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1674569696187973,
+      "learning_rate": 9.428279470260776e-05,
+      "loss": 0.0846,
+      "step": 269
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.23109744489192963,
+      "learning_rate": 9.423943070116218e-05,
+      "loss": 0.136,
+      "step": 270
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.21344415843486786,
+      "learning_rate": 9.4195912916359e-05,
+      "loss": 0.1091,
+      "step": 271
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.16391590237617493,
+      "learning_rate": 9.415224149947306e-05,
+      "loss": 0.0901,
+      "step": 272
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2023243010044098,
+      "learning_rate": 9.410841660231315e-05,
+      "loss": 0.0635,
+      "step": 273
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1723608821630478,
+      "learning_rate": 9.406443837722168e-05,
+      "loss": 0.1001,
+      "step": 274
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1470147669315338,
+      "learning_rate": 9.402030697707398e-05,
+      "loss": 0.0721,
+      "step": 275
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.15082985162734985,
+      "learning_rate": 9.397602255527791e-05,
+      "loss": 0.0698,
+      "step": 276
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.16322006285190582,
+      "learning_rate": 9.393158526577323e-05,
+      "loss": 0.0809,
+      "step": 277
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.10098633915185928,
+      "learning_rate": 9.388699526303105e-05,
+      "loss": 0.0386,
+      "step": 278
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.19049708545207977,
+      "learning_rate": 9.38422527020534e-05,
+      "loss": 0.0559,
+      "step": 279
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.22742775082588196,
+      "learning_rate": 9.37973577383726e-05,
+      "loss": 0.0802,
+      "step": 280
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2055177539587021,
+      "learning_rate": 9.375231052805072e-05,
+      "loss": 0.1048,
+      "step": 281
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1366245150566101,
+      "learning_rate": 9.370711122767913e-05,
+      "loss": 0.0204,
+      "step": 282
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3235447406768799,
+      "learning_rate": 9.36617599943778e-05,
+      "loss": 0.0974,
+      "step": 283
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.09579204767942429,
+      "learning_rate": 9.361625698579493e-05,
+      "loss": 0.0151,
+      "step": 284
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.07987037301063538,
+      "eval_runtime": 14.6437,
+      "eval_samples_per_second": 32.574,
+      "eval_steps_per_second": 8.195,
+      "step": 284
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.22850771248340607,
+      "learning_rate": 9.357060236010625e-05,
+      "loss": 0.0458,
+      "step": 285
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4980478882789612,
+      "learning_rate": 9.352479627601457e-05,
+      "loss": 0.1306,
+      "step": 286
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18770304322242737,
+      "learning_rate": 9.347883889274923e-05,
+      "loss": 0.0218,
+      "step": 287
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6386083364486694,
+      "learning_rate": 9.34327303700654e-05,
+      "loss": 0.0912,
+      "step": 288
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4997164309024811,
+      "learning_rate": 9.338647086824372e-05,
+      "loss": 0.1083,
+      "step": 289
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.31682559847831726,
+      "learning_rate": 9.334006054808966e-05,
+      "loss": 0.0947,
+      "step": 290
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.28325051069259644,
+      "learning_rate": 9.329349957093292e-05,
+      "loss": 0.0794,
+      "step": 291
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5778185725212097,
+      "learning_rate": 9.324678809862695e-05,
+      "loss": 0.1223,
+      "step": 292
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2953624725341797,
+      "learning_rate": 9.319992629354828e-05,
+      "loss": 0.0747,
+      "step": 293
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.28283196687698364,
+      "learning_rate": 9.31529143185961e-05,
+      "loss": 0.1099,
+      "step": 294
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.36138102412223816,
+      "learning_rate": 9.310575233719154e-05,
+      "loss": 0.1303,
+      "step": 295
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.16202205419540405,
+      "learning_rate": 9.305844051327725e-05,
+      "loss": 0.0805,
+      "step": 296
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.11523901671171188,
+      "learning_rate": 9.30109790113167e-05,
+      "loss": 0.0775,
+      "step": 297
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.10756238549947739,
+      "learning_rate": 9.296336799629369e-05,
+      "loss": 0.0795,
+      "step": 298
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.18142195045948029,
+      "learning_rate": 9.291560763371173e-05,
+      "loss": 0.0833,
+      "step": 299
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14596430957317352,
+      "learning_rate": 9.28676980895935e-05,
+      "loss": 0.0904,
+      "step": 300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.10054739564657211,
+      "learning_rate": 9.28196395304803e-05,
+      "loss": 0.0898,
+      "step": 301
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.24579764902591705,
+      "learning_rate": 9.277143212343134e-05,
+      "loss": 0.145,
+      "step": 302
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13506978750228882,
+      "learning_rate": 9.272307603602334e-05,
+      "loss": 0.0847,
+      "step": 303
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.17480792105197906,
+      "learning_rate": 9.267457143634979e-05,
+      "loss": 0.125,
+      "step": 304
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31458401679992676,
+      "learning_rate": 9.262591849302048e-05,
+      "loss": 0.1047,
+      "step": 305
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.17494355142116547,
+      "learning_rate": 9.257711737516082e-05,
+      "loss": 0.0576,
+      "step": 306
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2996468245983124,
+      "learning_rate": 9.252816825241134e-05,
+      "loss": 0.1012,
+      "step": 307
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.19122976064682007,
+      "learning_rate": 9.247907129492707e-05,
+      "loss": 0.0878,
+      "step": 308
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.16079925000667572,
+      "learning_rate": 9.242982667337685e-05,
+      "loss": 0.0778,
+      "step": 309
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2628028392791748,
+      "learning_rate": 9.238043455894293e-05,
+      "loss": 0.0987,
+      "step": 310
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.318097859621048,
+      "learning_rate": 9.23308951233202e-05,
+      "loss": 0.1108,
+      "step": 311
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2207389920949936,
+      "learning_rate": 9.228120853871571e-05,
+      "loss": 0.0826,
+      "step": 312
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34375905990600586,
+      "learning_rate": 9.223137497784797e-05,
+      "loss": 0.1174,
+      "step": 313
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.16714760661125183,
+      "learning_rate": 9.218139461394644e-05,
+      "loss": 0.0883,
+      "step": 314
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.24213539063930511,
+      "learning_rate": 9.213126762075088e-05,
+      "loss": 0.0686,
+      "step": 315
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2654499113559723,
+      "learning_rate": 9.208099417251077e-05,
+      "loss": 0.1185,
+      "step": 316
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1353083997964859,
+      "learning_rate": 9.203057444398469e-05,
+      "loss": 0.0806,
+      "step": 317
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.30304938554763794,
+      "learning_rate": 9.198000861043967e-05,
+      "loss": 0.0817,
+      "step": 318
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14495517313480377,
+      "learning_rate": 9.192929684765067e-05,
+      "loss": 0.0436,
+      "step": 319
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2180556207895279,
+      "learning_rate": 9.187843933189995e-05,
+      "loss": 0.1255,
+      "step": 320
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3051697611808777,
+      "learning_rate": 9.182743623997634e-05,
+      "loss": 0.1241,
+      "step": 321
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42936787009239197,
+      "learning_rate": 9.17762877491748e-05,
+      "loss": 0.1847,
+      "step": 322
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.20895107090473175,
+      "learning_rate": 9.172499403729566e-05,
+      "loss": 0.0939,
+      "step": 323
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.15273532271385193,
+      "learning_rate": 9.167355528264414e-05,
+      "loss": 0.1012,
+      "step": 324
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.15428248047828674,
+      "learning_rate": 9.162197166402956e-05,
+      "loss": 0.061,
+      "step": 325
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13089029490947723,
+      "learning_rate": 9.157024336076487e-05,
+      "loss": 0.089,
+      "step": 326
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12000248581171036,
+      "learning_rate": 9.151837055266594e-05,
+      "loss": 0.0813,
+      "step": 327
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12965545058250427,
+      "learning_rate": 9.146635342005099e-05,
+      "loss": 0.113,
+      "step": 328
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12225235253572464,
+      "learning_rate": 9.14141921437399e-05,
+      "loss": 0.0968,
+      "step": 329
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.11869696527719498,
+      "learning_rate": 9.136188690505363e-05,
+      "loss": 0.0752,
+      "step": 330
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.22600843012332916,
+      "learning_rate": 9.130943788581359e-05,
+      "loss": 0.1049,
+      "step": 331
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.13381795585155487,
+      "learning_rate": 9.125684526834099e-05,
+      "loss": 0.0917,
+      "step": 332
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.12936879694461823,
+      "learning_rate": 9.120410923545619e-05,
+      "loss": 0.0782,
+      "step": 333
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14804388582706451,
+      "learning_rate": 9.115122997047811e-05,
+      "loss": 0.0959,
+      "step": 334
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.18504676222801208,
+      "learning_rate": 9.109820765722357e-05,
+      "loss": 0.1126,
+      "step": 335
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.1957363486289978,
+      "learning_rate": 9.10450424800066e-05,
+      "loss": 0.101,
+      "step": 336
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.15677915513515472,
+      "learning_rate": 9.099173462363792e-05,
+      "loss": 0.0775,
+      "step": 337
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.124906025826931,
+      "learning_rate": 9.093828427342418e-05,
+      "loss": 0.07,
+      "step": 338
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12451624125242233,
+      "learning_rate": 9.088469161516735e-05,
+      "loss": 0.0588,
+      "step": 339
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.509678304195404,
+      "learning_rate": 9.083095683516414e-05,
+      "loss": 0.1563,
+      "step": 340
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2245551496744156,
+      "learning_rate": 9.077708012020524e-05,
+      "loss": 0.1029,
+      "step": 341
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2735763192176819,
+      "learning_rate": 9.072306165757476e-05,
+      "loss": 0.0958,
+      "step": 342
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2062731385231018,
+      "learning_rate": 9.066890163504955e-05,
+      "loss": 0.0638,
+      "step": 343
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1664024293422699,
+      "learning_rate": 9.061460024089853e-05,
+      "loss": 0.0555,
+      "step": 344
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.15788845717906952,
+      "learning_rate": 9.056015766388205e-05,
+      "loss": 0.0509,
+      "step": 345
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.185616135597229,
+      "learning_rate": 9.050557409325125e-05,
+      "loss": 0.1196,
+      "step": 346
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.24650661647319794,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 0.0723,
+      "step": 347
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.23959776759147644,
+      "learning_rate": 9.039598473060113e-05,
+      "loss": 0.1139,
+      "step": 348
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.24370582401752472,
+      "learning_rate": 9.034097931953201e-05,
+      "loss": 0.0559,
+      "step": 349
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.11590461432933807,
+      "learning_rate": 9.028583367674765e-05,
+      "loss": 0.0285,
+      "step": 350
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.21419131755828857,
+      "learning_rate": 9.023054799394316e-05,
+      "loss": 0.0686,
+      "step": 351
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2115790992975235,
+      "learning_rate": 9.017512246330042e-05,
+      "loss": 0.071,
+      "step": 352
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2025454044342041,
+      "learning_rate": 9.011955727748748e-05,
+      "loss": 0.0993,
+      "step": 353
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.22743502259254456,
+      "learning_rate": 9.006385262965786e-05,
+      "loss": 0.0705,
+      "step": 354
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.16963045299053192,
+      "learning_rate": 9.00080087134498e-05,
+      "loss": 0.0569,
+      "step": 355
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.12319042533636093,
+      "learning_rate": 8.995202572298576e-05,
+      "loss": 0.0427,
+      "step": 356
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.15424852073192596,
+      "learning_rate": 8.989590385287155e-05,
+      "loss": 0.0564,
+      "step": 357
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.306594580411911,
+      "learning_rate": 8.983964329819583e-05,
+      "loss": 0.095,
+      "step": 358
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13876177370548248,
+      "learning_rate": 8.978324425452931e-05,
+      "loss": 0.0641,
+      "step": 359
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.19561870396137238,
+      "learning_rate": 8.972670691792409e-05,
+      "loss": 0.0635,
+      "step": 360
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3458711504936218,
+      "learning_rate": 8.967003148491304e-05,
+      "loss": 0.1328,
+      "step": 361
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1129189059138298,
+      "learning_rate": 8.961321815250905e-05,
+      "loss": 0.0205,
+      "step": 362
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3680332601070404,
+      "learning_rate": 8.955626711820438e-05,
+      "loss": 0.1302,
+      "step": 363
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2695287764072418,
+      "learning_rate": 8.949917857996996e-05,
+      "loss": 0.0511,
+      "step": 364
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.17332953214645386,
+      "learning_rate": 8.94419527362547e-05,
+      "loss": 0.0494,
+      "step": 365
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.11610284447669983,
+      "learning_rate": 8.938458978598483e-05,
+      "loss": 0.0381,
+      "step": 366
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.23595061898231506,
+      "learning_rate": 8.932708992856315e-05,
+      "loss": 0.0802,
+      "step": 367
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.26876452565193176,
+      "learning_rate": 8.926945336386838e-05,
+      "loss": 0.0461,
+      "step": 368
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.19504375755786896,
+      "learning_rate": 8.921168029225448e-05,
+      "loss": 0.0317,
+      "step": 369
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4416268467903137,
+      "learning_rate": 8.915377091454992e-05,
+      "loss": 0.0952,
+      "step": 370
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.32519325613975525,
+      "learning_rate": 8.909572543205698e-05,
+      "loss": 0.1027,
+      "step": 371
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3939536511898041,
+      "learning_rate": 8.903754404655106e-05,
+      "loss": 0.1718,
+      "step": 372
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.20514678955078125,
+      "learning_rate": 8.897922696027999e-05,
+      "loss": 0.06,
+      "step": 373
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3049127459526062,
+      "learning_rate": 8.892077437596332e-05,
+      "loss": 0.1014,
+      "step": 374
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.19251297414302826,
+      "learning_rate": 8.88621864967916e-05,
+      "loss": 0.048,
+      "step": 375
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.21226820349693298,
+      "learning_rate": 8.880346352642575e-05,
+      "loss": 0.0652,
+      "step": 376
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.42633509635925293,
+      "learning_rate": 8.874460566899616e-05,
+      "loss": 0.1083,
+      "step": 377
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.24073313176631927,
+      "learning_rate": 8.868561312910221e-05,
+      "loss": 0.0851,
+      "step": 378
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3047339618206024,
+      "learning_rate": 8.862648611181145e-05,
+      "loss": 0.086,
+      "step": 379
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2227114737033844,
+      "learning_rate": 8.856722482265886e-05,
+      "loss": 0.1196,
+      "step": 380
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1860799938440323,
+      "learning_rate": 8.850782946764619e-05,
+      "loss": 0.0779,
+      "step": 381
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2109043002128601,
+      "learning_rate": 8.844830025324122e-05,
+      "loss": 0.076,
+      "step": 382
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1998620331287384,
+      "learning_rate": 8.838863738637706e-05,
+      "loss": 0.1027,
+      "step": 383
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.12607474625110626,
+      "learning_rate": 8.832884107445139e-05,
+      "loss": 0.0436,
+      "step": 384
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2890150845050812,
+      "learning_rate": 8.826891152532579e-05,
+      "loss": 0.0966,
+      "step": 385
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4496447443962097,
+      "learning_rate": 8.820884894732497e-05,
+      "loss": 0.1575,
+      "step": 386
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.19411596655845642,
+      "learning_rate": 8.814865354923613e-05,
+      "loss": 0.1201,
+      "step": 387
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.19513021409511566,
+      "learning_rate": 8.808832554030808e-05,
+      "loss": 0.0747,
+      "step": 388
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14038780331611633,
+      "learning_rate": 8.802786513025068e-05,
+      "loss": 0.0608,
+      "step": 389
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14907363057136536,
+      "learning_rate": 8.796727252923402e-05,
+      "loss": 0.0843,
+      "step": 390
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.18512780964374542,
+      "learning_rate": 8.790654794788769e-05,
+      "loss": 0.0988,
+      "step": 391
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.17880797386169434,
+      "learning_rate": 8.784569159730007e-05,
+      "loss": 0.079,
+      "step": 392
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.16263402998447418,
+      "learning_rate": 8.778470368901762e-05,
+      "loss": 0.0704,
+      "step": 393
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.27071598172187805,
+      "learning_rate": 8.772358443504405e-05,
+      "loss": 0.0983,
+      "step": 394
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.23446398973464966,
+      "learning_rate": 8.766233404783974e-05,
+      "loss": 0.0577,
+      "step": 395
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2932927906513214,
+      "learning_rate": 8.760095274032083e-05,
+      "loss": 0.0946,
+      "step": 396
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.29224956035614014,
+      "learning_rate": 8.75394407258586e-05,
+      "loss": 0.078,
+      "step": 397
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.15467233955860138,
+      "learning_rate": 8.747779821827868e-05,
+      "loss": 0.0779,
+      "step": 398
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1883499026298523,
+      "learning_rate": 8.741602543186032e-05,
+      "loss": 0.0721,
+      "step": 399
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.25179481506347656,
+      "learning_rate": 8.735412258133562e-05,
+      "loss": 0.0875,
+      "step": 400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.23851999640464783,
+      "learning_rate": 8.729208988188881e-05,
+      "loss": 0.0959,
+      "step": 401
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2662704885005951,
+      "learning_rate": 8.722992754915554e-05,
+      "loss": 0.1025,
+      "step": 402
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.17909982800483704,
+      "learning_rate": 8.716763579922204e-05,
+      "loss": 0.0504,
+      "step": 403
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.17002324759960175,
+      "learning_rate": 8.710521484862439e-05,
+      "loss": 0.0856,
+      "step": 404
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2229025810956955,
+      "learning_rate": 8.704266491434788e-05,
+      "loss": 0.0591,
+      "step": 405
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1444559544324875,
+      "learning_rate": 8.697998621382607e-05,
+      "loss": 0.0297,
+      "step": 406
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2677093744277954,
+      "learning_rate": 8.69171789649402e-05,
+      "loss": 0.0543,
+      "step": 407
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5875506401062012,
+      "learning_rate": 8.685424338601834e-05,
+      "loss": 0.1199,
+      "step": 408
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.34576529264450073,
+      "learning_rate": 8.679117969583464e-05,
+      "loss": 0.1003,
+      "step": 409
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2765222489833832,
+      "learning_rate": 8.672798811360863e-05,
+      "loss": 0.0358,
+      "step": 410
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.17154745757579803,
+      "learning_rate": 8.666466885900438e-05,
+      "loss": 0.0736,
+      "step": 411
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1607416719198227,
+      "learning_rate": 8.660122215212977e-05,
+      "loss": 0.0678,
+      "step": 412
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1216413602232933,
+      "learning_rate": 8.653764821353573e-05,
+      "loss": 0.0341,
+      "step": 413
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3713608980178833,
+      "learning_rate": 8.647394726421547e-05,
+      "loss": 0.118,
+      "step": 414
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.48576387763023376,
+      "learning_rate": 8.641011952560371e-05,
+      "loss": 0.0931,
+      "step": 415
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14704179763793945,
+      "learning_rate": 8.63461652195759e-05,
+      "loss": 0.0285,
+      "step": 416
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3306657671928406,
+      "learning_rate": 8.628208456844747e-05,
+      "loss": 0.0737,
+      "step": 417
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39767885208129883,
+      "learning_rate": 8.621787779497305e-05,
+      "loss": 0.097,
+      "step": 418
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3198534846305847,
+      "learning_rate": 8.615354512234569e-05,
+      "loss": 0.0731,
+      "step": 419
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.19191338121891022,
+      "learning_rate": 8.608908677419606e-05,
+      "loss": 0.0697,
+      "step": 420
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.26490989327430725,
+      "learning_rate": 8.602450297459172e-05,
+      "loss": 0.1214,
+      "step": 421
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3545917868614197,
+      "learning_rate": 8.595979394803634e-05,
+      "loss": 0.0933,
+      "step": 422
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.28239014744758606,
+      "learning_rate": 8.589495991946885e-05,
+      "loss": 0.0707,
+      "step": 423
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.48272502422332764,
+      "learning_rate": 8.583000111426276e-05,
+      "loss": 0.0831,
+      "step": 424
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.15938633680343628,
+      "learning_rate": 8.576491775822527e-05,
+      "loss": 0.0899,
+      "step": 425
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2616162896156311,
+      "learning_rate": 8.569971007759657e-05,
+      "loss": 0.118,
+      "step": 426
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.0783080980181694,
+      "eval_runtime": 14.6414,
+      "eval_samples_per_second": 32.579,
+      "eval_steps_per_second": 8.196,
+      "step": 426
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.19521737098693848,
+      "learning_rate": 8.563437829904903e-05,
+      "loss": 0.0814,
+      "step": 427
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.194011390209198,
+      "learning_rate": 8.55689226496864e-05,
+      "loss": 0.0799,
+      "step": 428
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2743787169456482,
+      "learning_rate": 8.550334335704298e-05,
+      "loss": 0.0869,
+      "step": 429
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1369010955095291,
+      "learning_rate": 8.543764064908295e-05,
+      "loss": 0.0435,
+      "step": 430
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.24237819015979767,
+      "learning_rate": 8.537181475419944e-05,
+      "loss": 0.1148,
+      "step": 431
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.14511409401893616,
+      "learning_rate": 8.530586590121383e-05,
+      "loss": 0.0764,
+      "step": 432
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15356196463108063,
+      "learning_rate": 8.523979431937492e-05,
+      "loss": 0.05,
+      "step": 433
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.21860916912555695,
+      "learning_rate": 8.51736002383581e-05,
+      "loss": 0.0971,
+      "step": 434
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.23724305629730225,
+      "learning_rate": 8.510728388826463e-05,
+      "loss": 0.1049,
+      "step": 435
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.25301918387413025,
+      "learning_rate": 8.50408454996208e-05,
+      "loss": 0.0848,
+      "step": 436
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.22409550845623016,
+      "learning_rate": 8.497428530337706e-05,
+      "loss": 0.101,
+      "step": 437
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1324710100889206,
+      "learning_rate": 8.490760353090737e-05,
+      "loss": 0.0723,
+      "step": 438
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1362515389919281,
+      "learning_rate": 8.484080041400826e-05,
+      "loss": 0.0709,
+      "step": 439
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.16375669836997986,
+      "learning_rate": 8.477387618489807e-05,
+      "loss": 0.0405,
+      "step": 440
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21752700209617615,
+      "learning_rate": 8.470683107621616e-05,
+      "loss": 0.0455,
+      "step": 441
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1989530771970749,
+      "learning_rate": 8.463966532102207e-05,
+      "loss": 0.0704,
+      "step": 442
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.192123144865036,
+      "learning_rate": 8.457237915279476e-05,
+      "loss": 0.063,
+      "step": 443
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1221012994647026,
+      "learning_rate": 8.450497280543174e-05,
+      "loss": 0.0302,
+      "step": 444
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5705539584159851,
+      "learning_rate": 8.443744651324827e-05,
+      "loss": 0.1531,
+      "step": 445
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.21490426361560822,
+      "learning_rate": 8.436980051097659e-05,
+      "loss": 0.0626,
+      "step": 446
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2654309570789337,
+      "learning_rate": 8.430203503376505e-05,
+      "loss": 0.0838,
+      "step": 447
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.22457195818424225,
+      "learning_rate": 8.423415031717733e-05,
+      "loss": 0.0309,
+      "step": 448
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.10934180021286011,
+      "learning_rate": 8.416614659719157e-05,
+      "loss": 0.0132,
+      "step": 449
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.32622861862182617,
+      "learning_rate": 8.409802411019963e-05,
+      "loss": 0.107,
+      "step": 450
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.24298590421676636,
+      "learning_rate": 8.40297830930062e-05,
+      "loss": 0.1268,
+      "step": 451
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.29994437098503113,
+      "learning_rate": 8.396142378282798e-05,
+      "loss": 0.0747,
+      "step": 452
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.16668649017810822,
+      "learning_rate": 8.389294641729293e-05,
+      "loss": 0.0479,
+      "step": 453
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.26706060767173767,
+      "learning_rate": 8.382435123443934e-05,
+      "loss": 0.1116,
+      "step": 454
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1750030219554901,
+      "learning_rate": 8.375563847271506e-05,
+      "loss": 0.0597,
+      "step": 455
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2318125069141388,
+      "learning_rate": 8.36868083709767e-05,
+      "loss": 0.0909,
+      "step": 456
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1834569126367569,
+      "learning_rate": 8.361786116848872e-05,
+      "loss": 0.0813,
+      "step": 457
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.27685895562171936,
+      "learning_rate": 8.354879710492264e-05,
+      "loss": 0.1301,
+      "step": 458
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.16120545566082,
+      "learning_rate": 8.347961642035624e-05,
+      "loss": 0.0717,
+      "step": 459
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.17625439167022705,
+      "learning_rate": 8.341031935527267e-05,
+      "loss": 0.0867,
+      "step": 460
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2321135252714157,
+      "learning_rate": 8.334090615055966e-05,
+      "loss": 0.1122,
+      "step": 461
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.12720270454883575,
+      "learning_rate": 8.327137704750862e-05,
+      "loss": 0.0375,
+      "step": 462
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2046743929386139,
+      "learning_rate": 8.320173228781389e-05,
+      "loss": 0.0808,
+      "step": 463
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.16513489186763763,
+      "learning_rate": 8.313197211357181e-05,
+      "loss": 0.0825,
+      "step": 464
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.12492749094963074,
+      "learning_rate": 8.306209676727994e-05,
+      "loss": 0.0876,
+      "step": 465
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1343008577823639,
+      "learning_rate": 8.299210649183619e-05,
+      "loss": 0.0852,
+      "step": 466
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.13951613008975983,
+      "learning_rate": 8.2922001530538e-05,
+      "loss": 0.1003,
+      "step": 467
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.16553768515586853,
+      "learning_rate": 8.285178212708143e-05,
+      "loss": 0.0662,
+      "step": 468
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.15311822295188904,
+      "learning_rate": 8.278144852556042e-05,
+      "loss": 0.0785,
+      "step": 469
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1948017179965973,
+      "learning_rate": 8.271100097046584e-05,
+      "loss": 0.0898,
+      "step": 470
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.11078551411628723,
+      "learning_rate": 8.264043970668469e-05,
+      "loss": 0.0386,
+      "step": 471
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1605585515499115,
+      "learning_rate": 8.256976497949924e-05,
+      "loss": 0.0497,
+      "step": 472
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1617887318134308,
+      "learning_rate": 8.249897703458619e-05,
+      "loss": 0.0624,
+      "step": 473
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1274091750383377,
+      "learning_rate": 8.242807611801578e-05,
+      "loss": 0.0578,
+      "step": 474
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.15953154861927032,
+      "learning_rate": 8.235706247625098e-05,
+      "loss": 0.042,
+      "step": 475
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.27984094619750977,
+      "learning_rate": 8.228593635614659e-05,
+      "loss": 0.1037,
+      "step": 476
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1895013153553009,
+      "learning_rate": 8.22146980049484e-05,
+      "loss": 0.0728,
+      "step": 477
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1580246388912201,
+      "learning_rate": 8.214334767029239e-05,
+      "loss": 0.0398,
+      "step": 478
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2391231805086136,
+      "learning_rate": 8.207188560020373e-05,
+      "loss": 0.0707,
+      "step": 479
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.25975751876831055,
+      "learning_rate": 8.200031204309603e-05,
+      "loss": 0.1369,
+      "step": 480
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.32591861486434937,
+      "learning_rate": 8.192862724777051e-05,
+      "loss": 0.0878,
+      "step": 481
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.32488566637039185,
+      "learning_rate": 8.185683146341496e-05,
+      "loss": 0.0692,
+      "step": 482
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1918002963066101,
+      "learning_rate": 8.178492493960309e-05,
+      "loss": 0.0942,
+      "step": 483
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.43140751123428345,
+      "learning_rate": 8.171290792629347e-05,
+      "loss": 0.0979,
+      "step": 484
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1771157830953598,
+      "learning_rate": 8.164078067382882e-05,
+      "loss": 0.0894,
+      "step": 485
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.25012728571891785,
+      "learning_rate": 8.1568543432935e-05,
+      "loss": 0.0734,
+      "step": 486
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19328337907791138,
+      "learning_rate": 8.149619645472031e-05,
+      "loss": 0.0869,
+      "step": 487
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.18180640041828156,
+      "learning_rate": 8.142373999067439e-05,
+      "loss": 0.0897,
+      "step": 488
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.25753355026245117,
+      "learning_rate": 8.135117429266757e-05,
+      "loss": 0.0883,
+      "step": 489
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.23837833106517792,
+      "learning_rate": 8.127849961294984e-05,
+      "loss": 0.0549,
+      "step": 490
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.25032365322113037,
+      "learning_rate": 8.120571620415006e-05,
+      "loss": 0.0976,
+      "step": 491
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.30728307366371155,
+      "learning_rate": 8.113282431927502e-05,
+      "loss": 0.0709,
+      "step": 492
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1391928791999817,
+      "learning_rate": 8.10598242117086e-05,
+      "loss": 0.0378,
+      "step": 493
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1786775141954422,
+      "learning_rate": 8.098671613521089e-05,
+      "loss": 0.0671,
+      "step": 494
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.21963584423065186,
+      "learning_rate": 8.091350034391732e-05,
+      "loss": 0.0936,
+      "step": 495
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.13954299688339233,
+      "learning_rate": 8.084017709233767e-05,
+      "loss": 0.052,
+      "step": 496
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19656923413276672,
+      "learning_rate": 8.076674663535537e-05,
+      "loss": 0.0584,
+      "step": 497
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14330637454986572,
+      "learning_rate": 8.069320922822643e-05,
+      "loss": 0.0786,
+      "step": 498
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19019991159439087,
+      "learning_rate": 8.061956512657871e-05,
+      "loss": 0.0837,
+      "step": 499
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2079285979270935,
+      "learning_rate": 8.05458145864109e-05,
+      "loss": 0.0459,
+      "step": 500
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3516862392425537,
+      "learning_rate": 8.047195786409172e-05,
+      "loss": 0.191,
+      "step": 501
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.192392036318779,
+      "learning_rate": 8.039799521635896e-05,
+      "loss": 0.1072,
+      "step": 502
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1787678301334381,
+      "learning_rate": 8.032392690031867e-05,
+      "loss": 0.0649,
+      "step": 503
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.21046535670757294,
+      "learning_rate": 8.024975317344421e-05,
+      "loss": 0.1065,
+      "step": 504
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1215684562921524,
+      "learning_rate": 8.017547429357532e-05,
+      "loss": 0.0433,
+      "step": 505
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1342051476240158,
+      "learning_rate": 8.010109051891731e-05,
+      "loss": 0.0774,
+      "step": 506
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.10215850174427032,
+      "learning_rate": 8.002660210804011e-05,
+      "loss": 0.0338,
+      "step": 507
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.23539598286151886,
+      "learning_rate": 7.995200931987743e-05,
+      "loss": 0.0516,
+      "step": 508
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.15601155161857605,
+      "learning_rate": 7.987731241372572e-05,
+      "loss": 0.0559,
+      "step": 509
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.165851429104805,
+      "learning_rate": 7.98025116492434e-05,
+      "loss": 0.0372,
+      "step": 510
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.21045421063899994,
+      "learning_rate": 7.972760728644996e-05,
+      "loss": 0.086,
+      "step": 511
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14814500510692596,
+      "learning_rate": 7.965259958572496e-05,
+      "loss": 0.0587,
+      "step": 512
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.15543898940086365,
+      "learning_rate": 7.95774888078072e-05,
+      "loss": 0.0682,
+      "step": 513
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.13527697324752808,
+      "learning_rate": 7.950227521379382e-05,
+      "loss": 0.0468,
+      "step": 514
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35773295164108276,
+      "learning_rate": 7.94269590651393e-05,
+      "loss": 0.1273,
+      "step": 515
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.22433511912822723,
+      "learning_rate": 7.935154062365467e-05,
+      "loss": 0.0438,
+      "step": 516
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1453983038663864,
+      "learning_rate": 7.927602015150655e-05,
+      "loss": 0.0367,
+      "step": 517
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2343645691871643,
+      "learning_rate": 7.920039791121617e-05,
+      "loss": 0.128,
+      "step": 518
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.25178173184394836,
+      "learning_rate": 7.912467416565861e-05,
+      "loss": 0.1094,
+      "step": 519
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3170076012611389,
+      "learning_rate": 7.904884917806174e-05,
+      "loss": 0.1323,
+      "step": 520
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.15917453169822693,
+      "learning_rate": 7.897292321200538e-05,
+      "loss": 0.036,
+      "step": 521
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.31578320264816284,
+      "learning_rate": 7.889689653142036e-05,
+      "loss": 0.0909,
+      "step": 522
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.16602741181850433,
+      "learning_rate": 7.882076940058764e-05,
+      "loss": 0.0371,
+      "step": 523
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3235325813293457,
+      "learning_rate": 7.874454208413731e-05,
+      "loss": 0.1561,
+      "step": 524
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1256486028432846,
+      "learning_rate": 7.866821484704776e-05,
+      "loss": 0.0364,
+      "step": 525
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2234162986278534,
+      "learning_rate": 7.859178795464472e-05,
+      "loss": 0.0883,
+      "step": 526
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1564294993877411,
+      "learning_rate": 7.851526167260034e-05,
+      "loss": 0.0679,
+      "step": 527
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.16309525072574615,
+      "learning_rate": 7.84386362669322e-05,
+      "loss": 0.0912,
+      "step": 528
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.21584004163742065,
+      "learning_rate": 7.836191200400255e-05,
+      "loss": 0.0695,
+      "step": 529
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15948422253131866,
+      "learning_rate": 7.828508915051724e-05,
+      "loss": 0.0459,
+      "step": 530
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.24016940593719482,
+      "learning_rate": 7.82081679735248e-05,
+      "loss": 0.1127,
+      "step": 531
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2894397974014282,
+      "learning_rate": 7.813114874041557e-05,
+      "loss": 0.0584,
+      "step": 532
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.20707662403583527,
+      "learning_rate": 7.805403171892079e-05,
+      "loss": 0.1045,
+      "step": 533
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.23427248001098633,
+      "learning_rate": 7.797681717711161e-05,
+      "loss": 0.1345,
+      "step": 534
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.13141866028308868,
+      "learning_rate": 7.789950538339812e-05,
+      "loss": 0.052,
+      "step": 535
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.21118536591529846,
+      "learning_rate": 7.782209660652855e-05,
+      "loss": 0.1272,
+      "step": 536
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.15485352277755737,
+      "learning_rate": 7.77445911155882e-05,
+      "loss": 0.0686,
+      "step": 537
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.11380946636199951,
+      "learning_rate": 7.766698917999861e-05,
+      "loss": 0.0735,
+      "step": 538
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.26798170804977417,
+      "learning_rate": 7.758929106951656e-05,
+      "loss": 0.0934,
+      "step": 539
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23003587126731873,
+      "learning_rate": 7.751149705423312e-05,
+      "loss": 0.0816,
+      "step": 540
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2122953236103058,
+      "learning_rate": 7.743360740457278e-05,
+      "loss": 0.0827,
+      "step": 541
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.22673499584197998,
+      "learning_rate": 7.735562239129247e-05,
+      "loss": 0.1232,
+      "step": 542
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.24960415065288544,
+      "learning_rate": 7.727754228548058e-05,
+      "loss": 0.1124,
+      "step": 543
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.10405872017145157,
+      "learning_rate": 7.719936735855611e-05,
+      "loss": 0.0687,
+      "step": 544
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.16980154812335968,
+      "learning_rate": 7.712109788226762e-05,
+      "loss": 0.0874,
+      "step": 545
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1486412137746811,
+      "learning_rate": 7.704273412869238e-05,
+      "loss": 0.0815,
+      "step": 546
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14432762563228607,
+      "learning_rate": 7.696427637023538e-05,
+      "loss": 0.0752,
+      "step": 547
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2627028822898865,
+      "learning_rate": 7.688572487962835e-05,
+      "loss": 0.0982,
+      "step": 548
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.16832011938095093,
+      "learning_rate": 7.680707992992888e-05,
+      "loss": 0.0895,
+      "step": 549
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14999301731586456,
+      "learning_rate": 7.672834179451942e-05,
+      "loss": 0.0544,
+      "step": 550
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14237482845783234,
+      "learning_rate": 7.664951074710638e-05,
+      "loss": 0.0623,
+      "step": 551
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1694159060716629,
+      "learning_rate": 7.657058706171911e-05,
+      "loss": 0.0784,
+      "step": 552
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1470886617898941,
+      "learning_rate": 7.649157101270902e-05,
+      "loss": 0.0635,
+      "step": 553
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.16492018103599548,
+      "learning_rate": 7.641246287474855e-05,
+      "loss": 0.0669,
+      "step": 554
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.195392444729805,
+      "learning_rate": 7.633326292283028e-05,
+      "loss": 0.0387,
+      "step": 555
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.17653177678585052,
+      "learning_rate": 7.625397143226596e-05,
+      "loss": 0.0592,
+      "step": 556
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.23455718159675598,
+      "learning_rate": 7.617458867868553e-05,
+      "loss": 0.0882,
+      "step": 557
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3588998317718506,
+      "learning_rate": 7.609511493803616e-05,
+      "loss": 0.107,
+      "step": 558
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2767946720123291,
+      "learning_rate": 7.601555048658134e-05,
+      "loss": 0.1609,
+      "step": 559
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.22181196510791779,
+      "learning_rate": 7.593589560089985e-05,
+      "loss": 0.0598,
+      "step": 560
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.30335313081741333,
+      "learning_rate": 7.585615055788484e-05,
+      "loss": 0.0825,
+      "step": 561
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.19477833807468414,
+      "learning_rate": 7.577631563474291e-05,
+      "loss": 0.0446,
+      "step": 562
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.11036123335361481,
+      "learning_rate": 7.569639110899303e-05,
+      "loss": 0.025,
+      "step": 563
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.19955220818519592,
+      "learning_rate": 7.561637725846568e-05,
+      "loss": 0.0484,
+      "step": 564
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.20293684303760529,
+      "learning_rate": 7.553627436130183e-05,
+      "loss": 0.0689,
+      "step": 565
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.18100765347480774,
+      "learning_rate": 7.545608269595202e-05,
+      "loss": 0.0371,
+      "step": 566
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.43053922057151794,
+      "learning_rate": 7.537580254117531e-05,
+      "loss": 0.0901,
+      "step": 567
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3413926959037781,
+      "learning_rate": 7.529543417603844e-05,
+      "loss": 0.1088,
+      "step": 568
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.09266742318868637,
+      "eval_runtime": 14.642,
+      "eval_samples_per_second": 32.578,
+      "eval_steps_per_second": 8.196,
+      "step": 568
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.1796027272939682,
+      "learning_rate": 7.521497787991471e-05,
+      "loss": 0.0244,
+      "step": 569
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.30515041947364807,
+      "learning_rate": 7.513443393248312e-05,
+      "loss": 0.0682,
+      "step": 570
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3712550103664398,
+      "learning_rate": 7.505380261372734e-05,
+      "loss": 0.0921,
+      "step": 571
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.4219339191913605,
+      "learning_rate": 7.497308420393477e-05,
+      "loss": 0.0785,
+      "step": 572
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.24129725992679596,
+      "learning_rate": 7.489227898369559e-05,
+      "loss": 0.0851,
+      "step": 573
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.24595998227596283,
+      "learning_rate": 7.481138723390164e-05,
+      "loss": 0.1143,
+      "step": 574
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.13906948268413544,
+      "learning_rate": 7.473040923574567e-05,
+      "loss": 0.0402,
+      "step": 575
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.1885530650615692,
+      "learning_rate": 7.464934527072016e-05,
+      "loss": 0.0384,
+      "step": 576
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.13116823136806488,
+      "learning_rate": 7.456819562061649e-05,
+      "loss": 0.0447,
+      "step": 577
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.23953841626644135,
+      "learning_rate": 7.448696056752383e-05,
+      "loss": 0.0602,
+      "step": 578
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.17374739050865173,
+      "learning_rate": 7.440564039382827e-05,
+      "loss": 0.0657,
+      "step": 579
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.20921552181243896,
+      "learning_rate": 7.432423538221178e-05,
+      "loss": 0.0757,
+      "step": 580
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.10258325189352036,
+      "learning_rate": 7.424274581565123e-05,
+      "loss": 0.0237,
+      "step": 581
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.31752172112464905,
+      "learning_rate": 7.416117197741742e-05,
+      "loss": 0.0625,
+      "step": 582
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.229179248213768,
+      "learning_rate": 7.407951415107413e-05,
+      "loss": 0.0792,
+      "step": 583
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.16059361398220062,
+      "learning_rate": 7.3997772620477e-05,
+      "loss": 0.0718,
+      "step": 584
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.1626499593257904,
+      "learning_rate": 7.391594766977277e-05,
+      "loss": 0.0457,
+      "step": 585
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.1549261212348938,
+      "learning_rate": 7.383403958339807e-05,
+      "loss": 0.0544,
+      "step": 586
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.1588374525308609,
+      "learning_rate": 7.375204864607852e-05,
+      "loss": 0.0342,
+      "step": 587
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.09883646667003632,
+      "learning_rate": 7.366997514282782e-05,
+      "loss": 0.0292,
+      "step": 588
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.3421178460121155,
+      "learning_rate": 7.358781935894659e-05,
+      "loss": 0.0999,
+      "step": 589
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.12105683982372284,
+      "learning_rate": 7.350558158002154e-05,
+      "loss": 0.023,
+      "step": 590
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.15255074203014374,
+      "learning_rate": 7.342326209192435e-05,
+      "loss": 0.0423,
+      "step": 591
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.18337713181972504,
+      "learning_rate": 7.33408611808108e-05,
+      "loss": 0.0305,
+      "step": 592
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.12969495356082916,
+      "learning_rate": 7.325837913311966e-05,
+      "loss": 0.0175,
+      "step": 593
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2849477231502533,
+      "learning_rate": 7.317581623557177e-05,
+      "loss": 0.0878,
+      "step": 594
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.24307942390441895,
+      "learning_rate": 7.3093172775169e-05,
+      "loss": 0.037,
+      "step": 595
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2612784206867218,
+      "learning_rate": 7.301044903919325e-05,
+      "loss": 0.097,
+      "step": 596
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.2548207640647888,
+      "learning_rate": 7.292764531520553e-05,
+      "loss": 0.0928,
+      "step": 597
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.3221377432346344,
+      "learning_rate": 7.284476189104485e-05,
+      "loss": 0.0874,
+      "step": 598
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.20441681146621704,
+      "learning_rate": 7.27617990548273e-05,
+      "loss": 0.035,
+      "step": 599
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.26378926634788513,
+      "learning_rate": 7.267875709494499e-05,
+      "loss": 0.0494,
+      "step": 600
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5505862832069397,
+      "learning_rate": 7.259563630006512e-05,
+      "loss": 0.1241,
+      "step": 601
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.13375498354434967,
+      "learning_rate": 7.251243695912886e-05,
+      "loss": 0.0241,
+      "step": 602
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.19569019973278046,
+      "learning_rate": 7.242915936135051e-05,
+      "loss": 0.0698,
+      "step": 603
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.28432735800743103,
+      "learning_rate": 7.234580379621637e-05,
+      "loss": 0.0641,
+      "step": 604
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.13998962938785553,
+      "learning_rate": 7.22623705534837e-05,
+      "loss": 0.0365,
+      "step": 605
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.1772097498178482,
+      "learning_rate": 7.217885992317985e-05,
+      "loss": 0.081,
+      "step": 606
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4318295121192932,
+      "learning_rate": 7.209527219560119e-05,
+      "loss": 0.0532,
+      "step": 607
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.2505156695842743,
+      "learning_rate": 7.201160766131207e-05,
+      "loss": 0.0667,
+      "step": 608
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.13024090230464935,
+      "learning_rate": 7.192786661114384e-05,
+      "loss": 0.0234,
+      "step": 609
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.2824789583683014,
+      "learning_rate": 7.184404933619377e-05,
+      "loss": 0.095,
+      "step": 610
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.18059489130973816,
+      "learning_rate": 7.17601561278242e-05,
+      "loss": 0.0471,
+      "step": 611
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.2839769124984741,
+      "learning_rate": 7.167618727766138e-05,
+      "loss": 0.0783,
+      "step": 612
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.1342955082654953,
+      "learning_rate": 7.159214307759448e-05,
+      "loss": 0.0453,
+      "step": 613
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.1336507499217987,
+      "learning_rate": 7.150802381977464e-05,
+      "loss": 0.0431,
+      "step": 614
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2953212559223175,
+      "learning_rate": 7.142382979661386e-05,
+      "loss": 0.0705,
+      "step": 615
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.17532870173454285,
+      "learning_rate": 7.133956130078412e-05,
+      "loss": 0.0666,
+      "step": 616
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.17404836416244507,
+      "learning_rate": 7.12552186252162e-05,
+      "loss": 0.0522,
+      "step": 617
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.25231000781059265,
+      "learning_rate": 7.117080206309878e-05,
+      "loss": 0.0854,
+      "step": 618
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.2264215499162674,
+      "learning_rate": 7.108631190787735e-05,
+      "loss": 0.0692,
+      "step": 619
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3555202782154083,
+      "learning_rate": 7.100174845325327e-05,
+      "loss": 0.074,
+      "step": 620
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.34550729393959045,
+      "learning_rate": 7.091711199318264e-05,
+      "loss": 0.0831,
+      "step": 621
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.14560338854789734,
+      "learning_rate": 7.083240282187543e-05,
+      "loss": 0.0404,
+      "step": 622
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.23464788496494293,
+      "learning_rate": 7.074762123379423e-05,
+      "loss": 0.0699,
+      "step": 623
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.22587832808494568,
+      "learning_rate": 7.066276752365352e-05,
+      "loss": 0.0887,
+      "step": 624
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.17183855175971985,
+      "learning_rate": 7.057784198641834e-05,
+      "loss": 0.0373,
+      "step": 625
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.19148162007331848,
+      "learning_rate": 7.049284491730354e-05,
+      "loss": 0.0289,
+      "step": 626
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.26134082674980164,
+      "learning_rate": 7.040777661177251e-05,
+      "loss": 0.0367,
+      "step": 627
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5379131436347961,
+      "learning_rate": 7.032263736553635e-05,
+      "loss": 0.1049,
+      "step": 628
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.13634662330150604,
+      "learning_rate": 7.023742747455276e-05,
+      "loss": 0.018,
+      "step": 629
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.28767991065979004,
+      "learning_rate": 7.015214723502496e-05,
+      "loss": 0.06,
+      "step": 630
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.2551933825016022,
+      "learning_rate": 7.006679694340073e-05,
+      "loss": 0.0407,
+      "step": 631
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.41325151920318604,
+      "learning_rate": 6.998137689637142e-05,
+      "loss": 0.046,
+      "step": 632
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.25655174255371094,
+      "learning_rate": 6.989588739087078e-05,
+      "loss": 0.0398,
+      "step": 633
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3917771279811859,
+      "learning_rate": 6.981032872407405e-05,
+      "loss": 0.1072,
+      "step": 634
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.20738206803798676,
+      "learning_rate": 6.972470119339691e-05,
+      "loss": 0.0457,
+      "step": 635
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.1865154653787613,
+      "learning_rate": 6.963900509649434e-05,
+      "loss": 0.0258,
+      "step": 636
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.282071590423584,
+      "learning_rate": 6.955324073125979e-05,
+      "loss": 0.07,
+      "step": 637
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.27442115545272827,
+      "learning_rate": 6.946740839582388e-05,
+      "loss": 0.0875,
+      "step": 638
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.2635151445865631,
+      "learning_rate": 6.938150838855359e-05,
+      "loss": 0.0332,
+      "step": 639
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.16783182322978973,
+      "learning_rate": 6.929554100805118e-05,
+      "loss": 0.0405,
+      "step": 640
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.3328685760498047,
+      "learning_rate": 6.920950655315297e-05,
+      "loss": 0.1076,
+      "step": 641
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.20146729052066803,
+      "learning_rate": 6.91234053229286e-05,
+      "loss": 0.0481,
+      "step": 642
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.21599121391773224,
+      "learning_rate": 6.903723761667973e-05,
+      "loss": 0.0502,
+      "step": 643
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.16269706189632416,
+      "learning_rate": 6.895100373393913e-05,
+      "loss": 0.0652,
+      "step": 644
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3716180622577667,
+      "learning_rate": 6.886470397446958e-05,
+      "loss": 0.0914,
+      "step": 645
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.18003414571285248,
+      "learning_rate": 6.877833863826295e-05,
+      "loss": 0.0484,
+      "step": 646
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.2745915949344635,
+      "learning_rate": 6.869190802553894e-05,
+      "loss": 0.1057,
+      "step": 647
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.2507147490978241,
+      "learning_rate": 6.860541243674426e-05,
+      "loss": 0.0587,
+      "step": 648
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.19874247908592224,
+      "learning_rate": 6.851885217255145e-05,
+      "loss": 0.0452,
+      "step": 649
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.21256215870380402,
+      "learning_rate": 6.843222753385786e-05,
+      "loss": 0.0434,
+      "step": 650
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.13569054007530212,
+      "learning_rate": 6.834553882178463e-05,
+      "loss": 0.0275,
+      "step": 651
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.26869267225265503,
+      "learning_rate": 6.825878633767563e-05,
+      "loss": 0.1006,
+      "step": 652
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.21900776028633118,
+      "learning_rate": 6.817197038309644e-05,
+      "loss": 0.0564,
+      "step": 653
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.13306765258312225,
+      "learning_rate": 6.80850912598332e-05,
+      "loss": 0.0304,
+      "step": 654
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.09863998740911484,
+      "learning_rate": 6.79981492698917e-05,
+      "loss": 0.0178,
+      "step": 655
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.2170545756816864,
+      "learning_rate": 6.791114471549627e-05,
+      "loss": 0.0754,
+      "step": 656
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.26661446690559387,
+      "learning_rate": 6.782407789908863e-05,
+      "loss": 0.1083,
+      "step": 657
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.2747049629688263,
+      "learning_rate": 6.773694912332707e-05,
+      "loss": 0.0758,
+      "step": 658
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.252560019493103,
+      "learning_rate": 6.764975869108514e-05,
+      "loss": 0.0681,
+      "step": 659
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.2867240905761719,
+      "learning_rate": 6.756250690545079e-05,
+      "loss": 0.095,
+      "step": 660
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.09760677814483643,
+      "learning_rate": 6.747519406972524e-05,
+      "loss": 0.0123,
+      "step": 661
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.17589041590690613,
+      "learning_rate": 6.738782048742187e-05,
+      "loss": 0.0437,
+      "step": 662
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.29266613721847534,
+      "learning_rate": 6.730038646226532e-05,
+      "loss": 0.0706,
+      "step": 663
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.1055804044008255,
+      "learning_rate": 6.721289229819024e-05,
+      "loss": 0.0343,
+      "step": 664
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.2909635305404663,
+      "learning_rate": 6.712533829934042e-05,
+      "loss": 0.0817,
+      "step": 665
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.2604895532131195,
+      "learning_rate": 6.703772477006757e-05,
+      "loss": 0.0452,
+      "step": 666
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.10520771890878677,
+      "learning_rate": 6.695005201493038e-05,
+      "loss": 0.0215,
+      "step": 667
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.10080817341804504,
+      "learning_rate": 6.686232033869344e-05,
+      "loss": 0.0188,
+      "step": 668
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3340647220611572,
+      "learning_rate": 6.677453004632608e-05,
+      "loss": 0.0612,
+      "step": 669
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.29719796776771545,
+      "learning_rate": 6.668668144300149e-05,
+      "loss": 0.1014,
+      "step": 670
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.2131602168083191,
+      "learning_rate": 6.659877483409545e-05,
+      "loss": 0.0621,
+      "step": 671
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.1867963820695877,
+      "learning_rate": 6.65108105251855e-05,
+      "loss": 0.0312,
+      "step": 672
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.4250008463859558,
+      "learning_rate": 6.642278882204963e-05,
+      "loss": 0.0684,
+      "step": 673
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.20828047394752502,
+      "learning_rate": 6.633471003066543e-05,
+      "loss": 0.0421,
+      "step": 674
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.23356445133686066,
+      "learning_rate": 6.62465744572089e-05,
+      "loss": 0.0277,
+      "step": 675
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.42427390813827515,
+      "learning_rate": 6.615838240805344e-05,
+      "loss": 0.0745,
+      "step": 676
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.23298533260822296,
+      "learning_rate": 6.607013418976874e-05,
+      "loss": 0.047,
+      "step": 677
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5681192278862,
+      "learning_rate": 6.598183010911978e-05,
+      "loss": 0.1032,
+      "step": 678
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.15370431542396545,
+      "learning_rate": 6.589347047306571e-05,
+      "loss": 0.0224,
+      "step": 679
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.2974132001399994,
+      "learning_rate": 6.580505558875877e-05,
+      "loss": 0.0908,
+      "step": 680
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.12158460170030594,
+      "learning_rate": 6.571658576354333e-05,
+      "loss": 0.0212,
+      "step": 681
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.32594335079193115,
+      "learning_rate": 6.562806130495467e-05,
+      "loss": 0.1016,
+      "step": 682
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.3316996097564697,
+      "learning_rate": 6.5539482520718e-05,
+      "loss": 0.0639,
+      "step": 683
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.21660655736923218,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 0.043,
+      "step": 684
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.350033164024353,
+      "learning_rate": 6.536216320714466e-05,
+      "loss": 0.0752,
+      "step": 685
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.30745336413383484,
+      "learning_rate": 6.527342329419837e-05,
+      "loss": 0.0927,
+      "step": 686
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.24984771013259888,
+      "learning_rate": 6.51846302883827e-05,
+      "loss": 0.0685,
+      "step": 687
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.07773179560899734,
+      "learning_rate": 6.509578449835636e-05,
+      "loss": 0.0152,
+      "step": 688
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.1620987057685852,
+      "learning_rate": 6.500688623296159e-05,
+      "loss": 0.0514,
+      "step": 689
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.1917831003665924,
+      "learning_rate": 6.491793580122301e-05,
+      "loss": 0.066,
+      "step": 690
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.21920029819011688,
+      "learning_rate": 6.482893351234658e-05,
+      "loss": 0.0547,
+      "step": 691
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.29076483845710754,
+      "learning_rate": 6.473987967571856e-05,
+      "loss": 0.079,
+      "step": 692
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.30292215943336487,
+      "learning_rate": 6.46507746009043e-05,
+      "loss": 0.0957,
+      "step": 693
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.14139439165592194,
+      "learning_rate": 6.456161859764744e-05,
+      "loss": 0.0346,
+      "step": 694
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.22850438952445984,
+      "learning_rate": 6.447241197586847e-05,
+      "loss": 0.0744,
+      "step": 695
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.48915836215019226,
+      "learning_rate": 6.438315504566397e-05,
+      "loss": 0.0953,
+      "step": 696
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.17644958198070526,
+      "learning_rate": 6.429384811730528e-05,
+      "loss": 0.046,
+      "step": 697
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.2039819210767746,
+      "learning_rate": 6.420449150123767e-05,
+      "loss": 0.1052,
+      "step": 698
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.17715586721897125,
+      "learning_rate": 6.411508550807906e-05,
+      "loss": 0.0447,
+      "step": 699
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.16100600361824036,
+      "learning_rate": 6.4025630448619e-05,
+      "loss": 0.0344,
+      "step": 700
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.22480256855487823,
+      "learning_rate": 6.393612663381763e-05,
+      "loss": 0.0495,
+      "step": 701
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.12992677092552185,
+      "learning_rate": 6.384657437480458e-05,
+      "loss": 0.0409,
+      "step": 702
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.1325366348028183,
+      "learning_rate": 6.375697398287787e-05,
+      "loss": 0.0257,
+      "step": 703
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.16241514682769775,
+      "learning_rate": 6.366732576950284e-05,
+      "loss": 0.0427,
+      "step": 704
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.21476183831691742,
+      "learning_rate": 6.357763004631104e-05,
+      "loss": 0.0451,
+      "step": 705
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.32039332389831543,
+      "learning_rate": 6.34878871250992e-05,
+      "loss": 0.0545,
+      "step": 706
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3203076124191284,
+      "learning_rate": 6.33980973178281e-05,
+      "loss": 0.0917,
+      "step": 707
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.25006967782974243,
+      "learning_rate": 6.330826093662156e-05,
+      "loss": 0.1028,
+      "step": 708
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.52630215883255,
+      "learning_rate": 6.32183782937652e-05,
+      "loss": 0.0889,
+      "step": 709
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.33741331100463867,
+      "learning_rate": 6.31284497017055e-05,
+      "loss": 0.0725,
+      "step": 710
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 0.07285241782665253,
+      "eval_runtime": 14.6756,
+      "eval_samples_per_second": 32.503,
+      "eval_steps_per_second": 8.177,
+      "step": 710
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.40746867656707764,
+      "learning_rate": 6.303847547304873e-05,
+      "loss": 0.0945,
+      "step": 711
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.22757941484451294,
+      "learning_rate": 6.294845592055967e-05,
+      "loss": 0.0532,
+      "step": 712
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.19006334245204926,
+      "learning_rate": 6.285839135716079e-05,
+      "loss": 0.0484,
+      "step": 713
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.48126357793807983,
+      "learning_rate": 6.27682820959309e-05,
+      "loss": 0.0967,
+      "step": 714
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.23766569793224335,
+      "learning_rate": 6.26781284501043e-05,
+      "loss": 0.106,
+      "step": 715
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.16818860173225403,
+      "learning_rate": 6.258793073306949e-05,
+      "loss": 0.0494,
+      "step": 716
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.28579115867614746,
+      "learning_rate": 6.249768925836822e-05,
+      "loss": 0.0937,
+      "step": 717
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.16623319685459137,
+      "learning_rate": 6.240740433969432e-05,
+      "loss": 0.0301,
+      "step": 718
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.1560198813676834,
+      "learning_rate": 6.231707629089262e-05,
+      "loss": 0.0384,
+      "step": 719
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.2002251148223877,
+      "learning_rate": 6.2226705425958e-05,
+      "loss": 0.0545,
+      "step": 720
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.16213096678256989,
+      "learning_rate": 6.2136292059034e-05,
+      "loss": 0.0433,
+      "step": 721
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.27064821124076843,
+      "learning_rate": 6.204583650441201e-05,
+      "loss": 0.0796,
+      "step": 722
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.11131159216165543,
+      "learning_rate": 6.195533907653004e-05,
+      "loss": 0.0229,
+      "step": 723
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.22354401648044586,
+      "learning_rate": 6.18648000899717e-05,
+      "loss": 0.0475,
+      "step": 724
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.19944117963314056,
+      "learning_rate": 6.177421985946499e-05,
+      "loss": 0.0413,
+      "step": 725
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.32458746433258057,
+      "learning_rate": 6.168359869988134e-05,
+      "loss": 0.1205,
+      "step": 726
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.19088833034038544,
+      "learning_rate": 6.159293692623443e-05,
+      "loss": 0.0626,
+      "step": 727
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.2114744633436203,
+      "learning_rate": 6.150223485367914e-05,
+      "loss": 0.048,
+      "step": 728
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.11308068782091141,
+      "learning_rate": 6.141149279751043e-05,
+      "loss": 0.0286,
+      "step": 729
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.22453975677490234,
+      "learning_rate": 6.13207110731622e-05,
+      "loss": 0.0279,
+      "step": 730
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.274513840675354,
+      "learning_rate": 6.122988999620634e-05,
+      "loss": 0.0553,
+      "step": 731
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.2700372636318207,
+      "learning_rate": 6.113902988235145e-05,
+      "loss": 0.0973,
+      "step": 732
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.3287579119205475,
+      "learning_rate": 6.104813104744188e-05,
+      "loss": 0.0853,
+      "step": 733
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.37582048773765564,
+      "learning_rate": 6.095719380745654e-05,
+      "loss": 0.088,
+      "step": 734
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.2009502649307251,
+      "learning_rate": 6.086621847850788e-05,
+      "loss": 0.0525,
+      "step": 735
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.1417909413576126,
+      "learning_rate": 6.077520537684072e-05,
+      "loss": 0.0311,
+      "step": 736
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.18328174948692322,
+      "learning_rate": 6.068415481883122e-05,
+      "loss": 0.0379,
+      "step": 737
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.1513252556324005,
+      "learning_rate": 6.059306712098571e-05,
+      "loss": 0.0319,
+      "step": 738
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.2544059753417969,
+      "learning_rate": 6.0501942599939666e-05,
+      "loss": 0.0593,
+      "step": 739
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.4964008331298828,
+      "learning_rate": 6.0410781572456486e-05,
+      "loss": 0.0367,
+      "step": 740
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.14776591956615448,
+      "learning_rate": 6.031958435542659e-05,
+      "loss": 0.0284,
+      "step": 741
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.20590472221374512,
+      "learning_rate": 6.022835126586609e-05,
+      "loss": 0.0359,
+      "step": 742
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.2493211179971695,
+      "learning_rate": 6.0137082620915863e-05,
+      "loss": 0.0424,
+      "step": 743
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.31557443737983704,
+      "learning_rate": 6.0045778737840344e-05,
+      "loss": 0.0563,
+      "step": 744
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.14257828891277313,
+      "learning_rate": 5.995443993402647e-05,
+      "loss": 0.024,
+      "step": 745
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.21385452151298523,
+      "learning_rate": 5.9863066526982605e-05,
+      "loss": 0.0721,
+      "step": 746
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.17539048194885254,
+      "learning_rate": 5.977165883433734e-05,
+      "loss": 0.025,
+      "step": 747
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.28508231043815613,
+      "learning_rate": 5.9680217173838494e-05,
+      "loss": 0.0595,
+      "step": 748
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.30929744243621826,
+      "learning_rate": 5.9588741863351924e-05,
+      "loss": 0.112,
+      "step": 749
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.439656525850296,
+      "learning_rate": 5.949723322086053e-05,
+      "loss": 0.0427,
+      "step": 750
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.2300054430961609,
+      "learning_rate": 5.940569156446298e-05,
+      "loss": 0.0437,
+      "step": 751
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4155109226703644,
+      "learning_rate": 5.931411721237279e-05,
+      "loss": 0.0569,
+      "step": 752
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.25196224451065063,
+      "learning_rate": 5.922251048291707e-05,
+      "loss": 0.0413,
+      "step": 753
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5078486204147339,
+      "learning_rate": 5.913087169453554e-05,
+      "loss": 0.0988,
+      "step": 754
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.26931652426719666,
+      "learning_rate": 5.9039201165779315e-05,
+      "loss": 0.0578,
+      "step": 755
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.2641213834285736,
+      "learning_rate": 5.8947499215309834e-05,
+      "loss": 0.0362,
+      "step": 756
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.23865339159965515,
+      "learning_rate": 5.8855766161897805e-05,
+      "loss": 0.0375,
+      "step": 757
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.2594137191772461,
+      "learning_rate": 5.876400232442205e-05,
+      "loss": 0.0489,
+      "step": 758
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.2721590995788574,
+      "learning_rate": 5.867220802186837e-05,
+      "loss": 0.0407,
+      "step": 759
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3681499660015106,
+      "learning_rate": 5.85803835733285e-05,
+      "loss": 0.0554,
+      "step": 760
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.3132595121860504,
+      "learning_rate": 5.848852929799894e-05,
+      "loss": 0.0486,
+      "step": 761
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.16972127556800842,
+      "learning_rate": 5.8396645515179884e-05,
+      "loss": 0.0473,
+      "step": 762
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.30628886818885803,
+      "learning_rate": 5.83047325442741e-05,
+      "loss": 0.0664,
+      "step": 763
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.3327179551124573,
+      "learning_rate": 5.8212790704785824e-05,
+      "loss": 0.0605,
+      "step": 764
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.3301398754119873,
+      "learning_rate": 5.812082031631966e-05,
+      "loss": 0.0477,
+      "step": 765
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.23960134387016296,
+      "learning_rate": 5.8028821698579385e-05,
+      "loss": 0.0376,
+      "step": 766
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.2526357173919678,
+      "learning_rate": 5.7936795171367e-05,
+      "loss": 0.0712,
+      "step": 767
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.32746273279190063,
+      "learning_rate": 5.784474105458143e-05,
+      "loss": 0.0542,
+      "step": 768
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.10859230905771255,
+      "learning_rate": 5.77526596682176e-05,
+      "loss": 0.019,
+      "step": 769
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.2908915877342224,
+      "learning_rate": 5.766055133236513e-05,
+      "loss": 0.142,
+      "step": 770
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.26869770884513855,
+      "learning_rate": 5.7568416367207404e-05,
+      "loss": 0.0774,
+      "step": 771
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.39681994915008545,
+      "learning_rate": 5.7476255093020326e-05,
+      "loss": 0.0632,
+      "step": 772
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.14335761964321136,
+      "learning_rate": 5.7384067830171274e-05,
+      "loss": 0.03,
+      "step": 773
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.1377771943807602,
+      "learning_rate": 5.729185489911797e-05,
+      "loss": 0.0263,
+      "step": 774
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.19834232330322266,
+      "learning_rate": 5.719961662040733e-05,
+      "loss": 0.0506,
+      "step": 775
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.14378659427165985,
+      "learning_rate": 5.710735331467444e-05,
+      "loss": 0.0285,
+      "step": 776
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.25368401408195496,
+      "learning_rate": 5.701506530264132e-05,
+      "loss": 0.0584,
+      "step": 777
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.12339203804731369,
+      "learning_rate": 5.692275290511592e-05,
+      "loss": 0.0282,
+      "step": 778
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.203715518116951,
+      "learning_rate": 5.683041644299093e-05,
+      "loss": 0.0849,
+      "step": 779
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.1526814103126526,
+      "learning_rate": 5.673805623724272e-05,
+      "loss": 0.0256,
+      "step": 780
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.18840323388576508,
+      "learning_rate": 5.664567260893019e-05,
+      "loss": 0.048,
+      "step": 781
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.15979206562042236,
+      "learning_rate": 5.6553265879193606e-05,
+      "loss": 0.0237,
+      "step": 782
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.1128401905298233,
+      "learning_rate": 5.6460836369253624e-05,
+      "loss": 0.0213,
+      "step": 783
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.1648949831724167,
+      "learning_rate": 5.6368384400410035e-05,
+      "loss": 0.0348,
+      "step": 784
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.31091129779815674,
+      "learning_rate": 5.627591029404071e-05,
+      "loss": 0.0685,
+      "step": 785
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.2921251654624939,
+      "learning_rate": 5.6183414371600496e-05,
+      "loss": 0.045,
+      "step": 786
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.3398689925670624,
+      "learning_rate": 5.609089695462002e-05,
+      "loss": 0.0546,
+      "step": 787
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.21610289812088013,
+      "learning_rate": 5.599835836470469e-05,
+      "loss": 0.0322,
+      "step": 788
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3218781054019928,
+      "learning_rate": 5.5905798923533484e-05,
+      "loss": 0.0331,
+      "step": 789
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5338783860206604,
+      "learning_rate": 5.581321895285787e-05,
+      "loss": 0.0764,
+      "step": 790
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.2539553940296173,
+      "learning_rate": 5.5720618774500675e-05,
+      "loss": 0.0553,
+      "step": 791
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.22199298441410065,
+      "learning_rate": 5.5627998710354957e-05,
+      "loss": 0.0304,
+      "step": 792
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.13213643431663513,
+      "learning_rate": 5.5535359082382944e-05,
+      "loss": 0.0115,
+      "step": 793
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3686007857322693,
+      "learning_rate": 5.544270021261483e-05,
+      "loss": 0.0371,
+      "step": 794
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.08815140277147293,
+      "learning_rate": 5.535002242314772e-05,
+      "loss": 0.0089,
+      "step": 795
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.40321916341781616,
+      "learning_rate": 5.525732603614444e-05,
+      "loss": 0.0653,
+      "step": 796
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.44097116589546204,
+      "learning_rate": 5.5164611373832544e-05,
+      "loss": 0.0555,
+      "step": 797
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5565125942230225,
+      "learning_rate": 5.5071878758503046e-05,
+      "loss": 0.0646,
+      "step": 798
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.8579866290092468,
+      "learning_rate": 5.49791285125094e-05,
+      "loss": 0.1532,
+      "step": 799
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.550639271736145,
+      "learning_rate": 5.488636095826636e-05,
+      "loss": 0.0574,
+      "step": 800
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.07725897431373596,
+      "learning_rate": 5.479357641824877e-05,
+      "loss": 0.0087,
+      "step": 801
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.25981655716896057,
+      "learning_rate": 5.470077521499063e-05,
+      "loss": 0.0328,
+      "step": 802
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.07465404272079468,
+      "learning_rate": 5.4607957671083786e-05,
+      "loss": 0.0117,
+      "step": 803
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.22613628208637238,
+      "learning_rate": 5.4515124109176904e-05,
+      "loss": 0.0596,
+      "step": 804
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.20493067800998688,
+      "learning_rate": 5.442227485197435e-05,
+      "loss": 0.0394,
+      "step": 805
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.2182394117116928,
+      "learning_rate": 5.4329410222235034e-05,
+      "loss": 0.0491,
+      "step": 806
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.16270771622657776,
+      "learning_rate": 5.42365305427713e-05,
+      "loss": 0.0333,
+      "step": 807
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3527982234954834,
+      "learning_rate": 5.414363613644782e-05,
+      "loss": 0.1369,
+      "step": 808
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.38832610845565796,
+      "learning_rate": 5.405072732618043e-05,
+      "loss": 0.0719,
+      "step": 809
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.18581318855285645,
+      "learning_rate": 5.395780443493508e-05,
+      "loss": 0.0305,
+      "step": 810
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.298115611076355,
+      "learning_rate": 5.386486778572665e-05,
+      "loss": 0.0676,
+      "step": 811
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.15877433121204376,
+      "learning_rate": 5.3771917701617827e-05,
+      "loss": 0.0343,
+      "step": 812
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.14393776655197144,
+      "learning_rate": 5.367895450571801e-05,
+      "loss": 0.0395,
+      "step": 813
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.25177934765815735,
+      "learning_rate": 5.358597852118219e-05,
+      "loss": 0.0757,
+      "step": 814
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.16125288605690002,
+      "learning_rate": 5.3492990071209806e-05,
+      "loss": 0.0432,
+      "step": 815
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.21766537427902222,
+      "learning_rate": 5.3399989479043624e-05,
+      "loss": 0.087,
+      "step": 816
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.18221743404865265,
+      "learning_rate": 5.3306977067968614e-05,
+      "loss": 0.0403,
+      "step": 817
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.2517869472503662,
+      "learning_rate": 5.3213953161310825e-05,
+      "loss": 0.0666,
+      "step": 818
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.20983122289180756,
+      "learning_rate": 5.3120918082436314e-05,
+      "loss": 0.0664,
+      "step": 819
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.17914025485515594,
+      "learning_rate": 5.3027872154749915e-05,
+      "loss": 0.0382,
+      "step": 820
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.12290598452091217,
+      "learning_rate": 5.2934815701694204e-05,
+      "loss": 0.0282,
+      "step": 821
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.17699532210826874,
+      "learning_rate": 5.2841749046748345e-05,
+      "loss": 0.0413,
+      "step": 822
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.3889511227607727,
+      "learning_rate": 5.274867251342694e-05,
+      "loss": 0.0758,
+      "step": 823
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.24286973476409912,
+      "learning_rate": 5.2655586425278966e-05,
+      "loss": 0.0532,
+      "step": 824
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.14911137521266937,
+      "learning_rate": 5.256249110588659e-05,
+      "loss": 0.0277,
+      "step": 825
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.31271466612815857,
+      "learning_rate": 5.246938687886409e-05,
+      "loss": 0.0726,
+      "step": 826
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.2684333920478821,
+      "learning_rate": 5.237627406785667e-05,
+      "loss": 0.0993,
+      "step": 827
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.14797139167785645,
+      "learning_rate": 5.228315299653942e-05,
+      "loss": 0.0198,
+      "step": 828
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.29548555612564087,
+      "learning_rate": 5.2190023988616113e-05,
+      "loss": 0.0562,
+      "step": 829
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.29321712255477905,
+      "learning_rate": 5.2096887367818105e-05,
+      "loss": 0.1208,
+      "step": 830
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.18331380188465118,
+      "learning_rate": 5.2003743457903256e-05,
+      "loss": 0.0256,
+      "step": 831
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.21740898489952087,
+      "learning_rate": 5.1910592582654715e-05,
+      "loss": 0.057,
+      "step": 832
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.2625051736831665,
+      "learning_rate": 5.181743506587989e-05,
+      "loss": 0.0667,
+      "step": 833
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2670525908470154,
+      "learning_rate": 5.172427123140923e-05,
+      "loss": 0.0883,
+      "step": 834
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33282265067100525,
+      "learning_rate": 5.1631101403095184e-05,
+      "loss": 0.0424,
+      "step": 835
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.21608753502368927,
+      "learning_rate": 5.1537925904811004e-05,
+      "loss": 0.049,
+      "step": 836
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.10450909286737442,
+      "learning_rate": 5.144474506044968e-05,
+      "loss": 0.0158,
+      "step": 837
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.3188491761684418,
+      "learning_rate": 5.135155919392279e-05,
+      "loss": 0.0547,
+      "step": 838
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.24398969113826752,
+      "learning_rate": 5.125836862915934e-05,
+      "loss": 0.053,
+      "step": 839
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.1743936687707901,
+      "learning_rate": 5.116517369010466e-05,
+      "loss": 0.0239,
+      "step": 840
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.180791437625885,
+      "learning_rate": 5.1071974700719326e-05,
+      "loss": 0.0864,
+      "step": 841
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.19678902626037598,
+      "learning_rate": 5.0978771984978003e-05,
+      "loss": 0.0376,
+      "step": 842
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.230797678232193,
+      "learning_rate": 5.0885565866868227e-05,
+      "loss": 0.0597,
+      "step": 843
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4890972971916199,
+      "learning_rate": 5.079235667038944e-05,
+      "loss": 0.0832,
+      "step": 844
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.20508797466754913,
+      "learning_rate": 5.069914471955178e-05,
+      "loss": 0.0349,
+      "step": 845
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.21593628823757172,
+      "learning_rate": 5.060593033837493e-05,
+      "loss": 0.0354,
+      "step": 846
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.2712628245353699,
+      "learning_rate": 5.051271385088702e-05,
+      "loss": 0.0311,
+      "step": 847
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.11844774335622787,
+      "learning_rate": 5.041949558112351e-05,
+      "loss": 0.0109,
+      "step": 848
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.1798882633447647,
+      "learning_rate": 5.032627585312608e-05,
+      "loss": 0.0196,
+      "step": 849
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.290019690990448,
+      "learning_rate": 5.023305499094144e-05,
+      "loss": 0.0667,
+      "step": 850
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.24924272298812866,
+      "learning_rate": 5.013983331862027e-05,
+      "loss": 0.0556,
+      "step": 851
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.22597135603427887,
+      "learning_rate": 5.004661116021605e-05,
+      "loss": 0.0495,
+      "step": 852
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 0.07545028626918793,
+      "eval_runtime": 14.6561,
+      "eval_samples_per_second": 32.546,
+      "eval_steps_per_second": 8.188,
+      "step": 852
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.15348747372627258,
+      "learning_rate": 4.9953388839783954e-05,
+      "loss": 0.0204,
+      "step": 853
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.6507572531700134,
+      "learning_rate": 4.9860166681379745e-05,
+      "loss": 0.076,
+      "step": 854
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.13102935254573822,
+      "learning_rate": 4.976694500905857e-05,
+      "loss": 0.0143,
+      "step": 855
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.43004414439201355,
+      "learning_rate": 4.967372414687393e-05,
+      "loss": 0.0675,
+      "step": 856
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.38339918851852417,
+      "learning_rate": 4.95805044188765e-05,
+      "loss": 0.0747,
+      "step": 857
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.4646240472793579,
+      "learning_rate": 4.9487286149112986e-05,
+      "loss": 0.0883,
+      "step": 858
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.2721651792526245,
+      "learning_rate": 4.9394069661625076e-05,
+      "loss": 0.062,
+      "step": 859
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.356275349855423,
+      "learning_rate": 4.930085528044823e-05,
+      "loss": 0.0321,
+      "step": 860
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3066048324108124,
+      "learning_rate": 4.9207643329610556e-05,
+      "loss": 0.0525,
+      "step": 861
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.12254035472869873,
+      "learning_rate": 4.911443413313179e-05,
+      "loss": 0.0106,
+      "step": 862
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.19596797227859497,
+      "learning_rate": 4.9021228015022015e-05,
+      "loss": 0.0242,
+      "step": 863
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.39066678285598755,
+      "learning_rate": 4.892802529928067e-05,
+      "loss": 0.0558,
+      "step": 864
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.17401085793972015,
+      "learning_rate": 4.883482630989535e-05,
+      "loss": 0.0203,
+      "step": 865
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.28903472423553467,
+      "learning_rate": 4.874163137084068e-05,
+      "loss": 0.0819,
+      "step": 866
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.2746363580226898,
+      "learning_rate": 4.8648440806077226e-05,
+      "loss": 0.0696,
+      "step": 867
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.40534642338752747,
+      "learning_rate": 4.8555254939550324e-05,
+      "loss": 0.111,
+      "step": 868
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.41272208094596863,
+      "learning_rate": 4.8462074095188994e-05,
+      "loss": 0.1089,
+      "step": 869
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.40718454122543335,
+      "learning_rate": 4.8368898596904834e-05,
+      "loss": 0.1339,
+      "step": 870
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.14493143558502197,
+      "learning_rate": 4.827572876859078e-05,
+      "loss": 0.0227,
+      "step": 871
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.20390640199184418,
+      "learning_rate": 4.8182564934120115e-05,
+      "loss": 0.0464,
+      "step": 872
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.12927311658859253,
+      "learning_rate": 4.80894074173453e-05,
+      "loss": 0.0253,
+      "step": 873
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.17653903365135193,
+      "learning_rate": 4.799625654209675e-05,
+      "loss": 0.0509,
+      "step": 874
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34687289595603943,
+      "learning_rate": 4.790311263218191e-05,
+      "loss": 0.0916,
+      "step": 875
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.22851605713367462,
+      "learning_rate": 4.7809976011383905e-05,
+      "loss": 0.0857,
+      "step": 876
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.27485382556915283,
+      "learning_rate": 4.771684700346059e-05,
+      "loss": 0.0623,
+      "step": 877
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.21887461841106415,
+      "learning_rate": 4.762372593214335e-05,
+      "loss": 0.0573,
+      "step": 878
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.2466115951538086,
+      "learning_rate": 4.753061312113592e-05,
+      "loss": 0.1039,
+      "step": 879
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.344625860452652,
+      "learning_rate": 4.743750889411342e-05,
+      "loss": 0.0637,
+      "step": 880
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.1676146686077118,
+      "learning_rate": 4.7344413574721046e-05,
+      "loss": 0.0372,
+      "step": 881
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.14225785434246063,
+      "learning_rate": 4.725132748657307e-05,
+      "loss": 0.0506,
+      "step": 882
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.19915729761123657,
+      "learning_rate": 4.715825095325168e-05,
+      "loss": 0.0459,
+      "step": 883
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.20955249667167664,
+      "learning_rate": 4.70651842983058e-05,
+      "loss": 0.0539,
+      "step": 884
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.171535924077034,
+      "learning_rate": 4.697212784525008e-05,
+      "loss": 0.0346,
+      "step": 885
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.11981090158224106,
+      "learning_rate": 4.687908191756369e-05,
+      "loss": 0.0378,
+      "step": 886
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.18210795521736145,
+      "learning_rate": 4.678604683868918e-05,
+      "loss": 0.0563,
+      "step": 887
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.18884742259979248,
+      "learning_rate": 4.669302293203142e-05,
+      "loss": 0.0393,
+      "step": 888
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.21338443458080292,
+      "learning_rate": 4.660001052095639e-05,
+      "loss": 0.054,
+      "step": 889
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.16022799909114838,
+      "learning_rate": 4.65070099287902e-05,
+      "loss": 0.0497,
+      "step": 890
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.30642077326774597,
+      "learning_rate": 4.641402147881782e-05,
+      "loss": 0.0702,
+      "step": 891
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.24659690260887146,
+      "learning_rate": 4.6321045494282e-05,
+      "loss": 0.0986,
+      "step": 892
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4151371419429779,
+      "learning_rate": 4.62280822983822e-05,
+      "loss": 0.1064,
+      "step": 893
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.19555744528770447,
+      "learning_rate": 4.613513221427337e-05,
+      "loss": 0.034,
+      "step": 894
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.3575385510921478,
+      "learning_rate": 4.604219556506492e-05,
+      "loss": 0.0563,
+      "step": 895
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.33982524275779724,
+      "learning_rate": 4.594927267381958e-05,
+      "loss": 0.1152,
+      "step": 896
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.40054503083229065,
+      "learning_rate": 4.58563638635522e-05,
+      "loss": 0.0684,
+      "step": 897
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.16741478443145752,
+      "learning_rate": 4.5763469457228695e-05,
+      "loss": 0.0221,
+      "step": 898
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.30603042244911194,
+      "learning_rate": 4.5670589777764984e-05,
+      "loss": 0.0725,
+      "step": 899
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.345217227935791,
+      "learning_rate": 4.5577725148025646e-05,
+      "loss": 0.062,
+      "step": 900
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4248473048210144,
+      "learning_rate": 4.54848758908231e-05,
+      "loss": 0.1482,
+      "step": 901
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.25596097111701965,
+      "learning_rate": 4.5392042328916226e-05,
+      "loss": 0.0417,
+      "step": 902
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.14021873474121094,
+      "learning_rate": 4.5299224785009374e-05,
+      "loss": 0.0242,
+      "step": 903
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.16665437817573547,
+      "learning_rate": 4.5206423581751245e-05,
+      "loss": 0.0569,
+      "step": 904
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.29362550377845764,
+      "learning_rate": 4.511363904173366e-05,
+      "loss": 0.068,
+      "step": 905
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.44577184319496155,
+      "learning_rate": 4.5020871487490604e-05,
+      "loss": 0.0787,
+      "step": 906
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.20594125986099243,
+      "learning_rate": 4.492812124149696e-05,
+      "loss": 0.0868,
+      "step": 907
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.2906559109687805,
+      "learning_rate": 4.483538862616747e-05,
+      "loss": 0.0592,
+      "step": 908
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.17545486986637115,
+      "learning_rate": 4.4742673963855576e-05,
+      "loss": 0.0225,
+      "step": 909
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.18305286765098572,
+      "learning_rate": 4.46499775768523e-05,
+      "loss": 0.0483,
+      "step": 910
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.2249644249677658,
+      "learning_rate": 4.455729978738517e-05,
+      "loss": 0.0383,
+      "step": 911
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.3094448149204254,
+      "learning_rate": 4.446464091761706e-05,
+      "loss": 0.0533,
+      "step": 912
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.22453933954238892,
+      "learning_rate": 4.437200128964504e-05,
+      "loss": 0.0435,
+      "step": 913
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.1814616322517395,
+      "learning_rate": 4.4279381225499344e-05,
+      "loss": 0.0245,
+      "step": 914
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.20599542558193207,
+      "learning_rate": 4.418678104714214e-05,
+      "loss": 0.0321,
+      "step": 915
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.27197298407554626,
+      "learning_rate": 4.409420107646652e-05,
+      "loss": 0.0512,
+      "step": 916
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.35009968280792236,
+      "learning_rate": 4.400164163529532e-05,
+      "loss": 0.0717,
+      "step": 917
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.17196977138519287,
+      "learning_rate": 4.390910304537999e-05,
+      "loss": 0.033,
+      "step": 918
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.1884760707616806,
+      "learning_rate": 4.381658562839953e-05,
+      "loss": 0.0526,
+      "step": 919
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.4165942966938019,
+      "learning_rate": 4.3724089705959305e-05,
+      "loss": 0.0824,
+      "step": 920
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.36213231086730957,
+      "learning_rate": 4.363161559958996e-05,
+      "loss": 0.0524,
+      "step": 921
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.22675907611846924,
+      "learning_rate": 4.353916363074638e-05,
+      "loss": 0.0367,
+      "step": 922
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.29561713337898254,
+      "learning_rate": 4.34467341208064e-05,
+      "loss": 0.0364,
+      "step": 923
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.22054970264434814,
+      "learning_rate": 4.3354327391069826e-05,
+      "loss": 0.025,
+      "step": 924
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.06693907827138901,
+      "learning_rate": 4.3261943762757287e-05,
+      "loss": 0.0104,
+      "step": 925
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.339631050825119,
+      "learning_rate": 4.3169583557009064e-05,
+      "loss": 0.0732,
+      "step": 926
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.2175193578004837,
+      "learning_rate": 4.307724709488409e-05,
+      "loss": 0.0464,
+      "step": 927
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.23093104362487793,
+      "learning_rate": 4.298493469735869e-05,
+      "loss": 0.0335,
+      "step": 928
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.38875579833984375,
+      "learning_rate": 4.289264668532557e-05,
+      "loss": 0.0327,
+      "step": 929
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.05294647812843323,
+      "learning_rate": 4.280038337959268e-05,
+      "loss": 0.007,
+      "step": 930
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.36184802651405334,
+      "learning_rate": 4.270814510088203e-05,
+      "loss": 0.0688,
+      "step": 931
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.531517744064331,
+      "learning_rate": 4.2615932169828744e-05,
+      "loss": 0.1305,
+      "step": 932
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.3471108376979828,
+      "learning_rate": 4.2523744906979686e-05,
+      "loss": 0.0236,
+      "step": 933
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.2624709904193878,
+      "learning_rate": 4.24315836327926e-05,
+      "loss": 0.0272,
+      "step": 934
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.3648707866668701,
+      "learning_rate": 4.233944866763489e-05,
+      "loss": 0.0384,
+      "step": 935
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.2199166864156723,
+      "learning_rate": 4.224734033178241e-05,
+      "loss": 0.0347,
+      "step": 936
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.44493308663368225,
+      "learning_rate": 4.2155258945418566e-05,
+      "loss": 0.0405,
+      "step": 937
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.4102453291416168,
+      "learning_rate": 4.206320482863301e-05,
+      "loss": 0.0849,
+      "step": 938
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.33510318398475647,
+      "learning_rate": 4.1971178301420613e-05,
+      "loss": 0.052,
+      "step": 939
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.40965744853019714,
+      "learning_rate": 4.187917968368036e-05,
+      "loss": 0.0848,
+      "step": 940
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.2755095064640045,
+      "learning_rate": 4.178720929521418e-05,
+      "loss": 0.0391,
+      "step": 941
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.32818931341171265,
+      "learning_rate": 4.16952674557259e-05,
+      "loss": 0.0327,
+      "step": 942
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.39538365602493286,
+      "learning_rate": 4.1603354484820134e-05,
+      "loss": 0.043,
+      "step": 943
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.2221785932779312,
+      "learning_rate": 4.1511470702001074e-05,
+      "loss": 0.0288,
+      "step": 944
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.2112448811531067,
+      "learning_rate": 4.141961642667152e-05,
+      "loss": 0.025,
+      "step": 945
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.15548011660575867,
+      "learning_rate": 4.132779197813164e-05,
+      "loss": 0.0694,
+      "step": 946
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.10790842771530151,
+      "learning_rate": 4.1235997675577956e-05,
+      "loss": 0.0124,
+      "step": 947
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.42269936203956604,
+      "learning_rate": 4.11442338381022e-05,
+      "loss": 0.0937,
+      "step": 948
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.36002832651138306,
+      "learning_rate": 4.105250078469018e-05,
+      "loss": 0.1242,
+      "step": 949
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.22437407076358795,
+      "learning_rate": 4.0960798834220704e-05,
+      "loss": 0.0638,
+      "step": 950
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.3967968225479126,
+      "learning_rate": 4.086912830546448e-05,
+      "loss": 0.0439,
+      "step": 951
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.20550177991390228,
+      "learning_rate": 4.077748951708292e-05,
+      "loss": 0.0347,
+      "step": 952
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2020653337240219,
+      "learning_rate": 4.068588278762723e-05,
+      "loss": 0.0376,
+      "step": 953
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.19614431262016296,
+      "learning_rate": 4.0594308435537024e-05,
+      "loss": 0.032,
+      "step": 954
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.08721073716878891,
+      "learning_rate": 4.0502766779139484e-05,
+      "loss": 0.012,
+      "step": 955
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2303171306848526,
+      "learning_rate": 4.041125813664808e-05,
+      "loss": 0.0268,
+      "step": 956
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2725672721862793,
+      "learning_rate": 4.031978282616151e-05,
+      "loss": 0.0413,
+      "step": 957
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.28184202313423157,
+      "learning_rate": 4.0228341165662685e-05,
+      "loss": 0.0383,
+      "step": 958
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.28514358401298523,
+      "learning_rate": 4.0136933473017407e-05,
+      "loss": 0.044,
+      "step": 959
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.374714732170105,
+      "learning_rate": 4.004556006597353e-05,
+      "loss": 0.044,
+      "step": 960
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.19936969876289368,
+      "learning_rate": 3.9954221262159674e-05,
+      "loss": 0.0334,
+      "step": 961
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.3212338984012604,
+      "learning_rate": 3.986291737908414e-05,
+      "loss": 0.0473,
+      "step": 962
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.28045183420181274,
+      "learning_rate": 3.9771648734133906e-05,
+      "loss": 0.0321,
+      "step": 963
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.34167036414146423,
+      "learning_rate": 3.968041564457342e-05,
+      "loss": 0.0696,
+      "step": 964
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5529135465621948,
+      "learning_rate": 3.958921842754351e-05,
+      "loss": 0.131,
+      "step": 965
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.275803804397583,
+      "learning_rate": 3.949805740006036e-05,
+      "loss": 0.0436,
+      "step": 966
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.3003288209438324,
+      "learning_rate": 3.94069328790143e-05,
+      "loss": 0.073,
+      "step": 967
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.20078504085540771,
+      "learning_rate": 3.9315845181168784e-05,
+      "loss": 0.0425,
+      "step": 968
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.3904169797897339,
+      "learning_rate": 3.9224794623159294e-05,
+      "loss": 0.0668,
+      "step": 969
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.212997168302536,
+      "learning_rate": 3.913378152149214e-05,
+      "loss": 0.0436,
+      "step": 970
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.04039880260825157,
+      "learning_rate": 3.904280619254348e-05,
+      "loss": 0.0077,
+      "step": 971
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.21076536178588867,
+      "learning_rate": 3.895186895255814e-05,
+      "loss": 0.0677,
+      "step": 972
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.37436169385910034,
+      "learning_rate": 3.886097011764856e-05,
+      "loss": 0.0294,
+      "step": 973
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.26611942052841187,
+      "learning_rate": 3.877011000379367e-05,
+      "loss": 0.057,
+      "step": 974
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.32198566198349,
+      "learning_rate": 3.8679288926837804e-05,
+      "loss": 0.0583,
+      "step": 975
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.2785477340221405,
+      "learning_rate": 3.8588507202489586e-05,
+      "loss": 0.0913,
+      "step": 976
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.20920749008655548,
+      "learning_rate": 3.8497765146320876e-05,
+      "loss": 0.0454,
+      "step": 977
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.31738653779029846,
+      "learning_rate": 3.840706307376557e-05,
+      "loss": 0.0464,
+      "step": 978
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.1887190192937851,
+      "learning_rate": 3.8316401300118675e-05,
+      "loss": 0.026,
+      "step": 979
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.22016988694667816,
+      "learning_rate": 3.8225780140535025e-05,
+      "loss": 0.0375,
+      "step": 980
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.2261650264263153,
+      "learning_rate": 3.813519991002831e-05,
+      "loss": 0.0368,
+      "step": 981
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.3108493983745575,
+      "learning_rate": 3.804466092346997e-05,
+      "loss": 0.0539,
+      "step": 982
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.23392857611179352,
+      "learning_rate": 3.7954163495587995e-05,
+      "loss": 0.0363,
+      "step": 983
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.23699642717838287,
+      "learning_rate": 3.786370794096603e-05,
+      "loss": 0.0362,
+      "step": 984
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.29672032594680786,
+      "learning_rate": 3.777329457404202e-05,
+      "loss": 0.0388,
+      "step": 985
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.25258907675743103,
+      "learning_rate": 3.768292370910737e-05,
+      "loss": 0.0278,
+      "step": 986
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.40179169178009033,
+      "learning_rate": 3.759259566030571e-05,
+      "loss": 0.1118,
+      "step": 987
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.3777885437011719,
+      "learning_rate": 3.750231074163179e-05,
+      "loss": 0.073,
+      "step": 988
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.39896661043167114,
+      "learning_rate": 3.7412069266930516e-05,
+      "loss": 0.0428,
+      "step": 989
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.26577284932136536,
+      "learning_rate": 3.7321871549895714e-05,
+      "loss": 0.0335,
+      "step": 990
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5360684990882874,
+      "learning_rate": 3.7231717904069094e-05,
+      "loss": 0.0979,
+      "step": 991
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.21000511944293976,
+      "learning_rate": 3.714160864283923e-05,
+      "loss": 0.0156,
+      "step": 992
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.11425631493330002,
+      "learning_rate": 3.7051544079440336e-05,
+      "loss": 0.0143,
+      "step": 993
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.20283763110637665,
+      "learning_rate": 3.696152452695128e-05,
+      "loss": 0.0627,
+      "step": 994
+    },
+    {
+      "epoch": 1.76,
+      "eval_loss": 0.07780980318784714,
+      "eval_runtime": 14.6775,
+      "eval_samples_per_second": 32.499,
+      "eval_steps_per_second": 8.176,
+      "step": 994
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.13721764087677002,
+      "learning_rate": 3.68715502982945e-05,
+      "loss": 0.0154,
+      "step": 995
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.397158145904541,
+      "learning_rate": 3.678162170623481e-05,
+      "loss": 0.0494,
+      "step": 996
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.47730910778045654,
+      "learning_rate": 3.669173906337846e-05,
+      "loss": 0.0897,
+      "step": 997
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.19200514256954193,
+      "learning_rate": 3.6601902682171894e-05,
+      "loss": 0.0145,
+      "step": 998
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.39415818452835083,
+      "learning_rate": 3.65121128749008e-05,
+      "loss": 0.0778,
+      "step": 999
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.17673304677009583,
+      "learning_rate": 3.642236995368897e-05,
+      "loss": 0.0211,
+      "step": 1000
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.3190731108188629,
+      "learning_rate": 3.633267423049717e-05,
+      "loss": 0.0856,
+      "step": 1001
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.3833164572715759,
+      "learning_rate": 3.624302601712213e-05,
+      "loss": 0.0687,
+      "step": 1002
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2938999831676483,
+      "learning_rate": 3.6153425625195425e-05,
+      "loss": 0.0717,
+      "step": 1003
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.2645472586154938,
+      "learning_rate": 3.606387336618237e-05,
+      "loss": 0.0341,
+      "step": 1004
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.1330021619796753,
+      "learning_rate": 3.597436955138102e-05,
+      "loss": 0.0244,
+      "step": 1005
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.34308159351348877,
+      "learning_rate": 3.588491449192096e-05,
+      "loss": 0.091,
+      "step": 1006
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.2861696481704712,
+      "learning_rate": 3.579550849876233e-05,
+      "loss": 0.0601,
+      "step": 1007
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.32016104459762573,
+      "learning_rate": 3.570615188269473e-05,
+      "loss": 0.0699,
+      "step": 1008
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2865599989891052,
+      "learning_rate": 3.561684495433605e-05,
+      "loss": 0.0742,
+      "step": 1009
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2045123279094696,
+      "learning_rate": 3.5527588024131544e-05,
+      "loss": 0.0323,
+      "step": 1010
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.13562075793743134,
+      "learning_rate": 3.5438381402352574e-05,
+      "loss": 0.0167,
+      "step": 1011
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2536921799182892,
+      "learning_rate": 3.534922539909569e-05,
+      "loss": 0.047,
+      "step": 1012
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.193417489528656,
+      "learning_rate": 3.5260120324281474e-05,
+      "loss": 0.0349,
+      "step": 1013
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.18863309919834137,
+      "learning_rate": 3.517106648765343e-05,
+      "loss": 0.0261,
+      "step": 1014
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5035936832427979,
+      "learning_rate": 3.5082064198777e-05,
+      "loss": 0.0963,
+      "step": 1015
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.33512285351753235,
+      "learning_rate": 3.499311376703842e-05,
+      "loss": 0.0534,
+      "step": 1016
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.14217574894428253,
+      "learning_rate": 3.4904215501643646e-05,
+      "loss": 0.0246,
+      "step": 1017
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3124421536922455,
+      "learning_rate": 3.4815369711617316e-05,
+      "loss": 0.0498,
+      "step": 1018
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.22936655580997467,
+      "learning_rate": 3.4726576705801636e-05,
+      "loss": 0.0249,
+      "step": 1019
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5534436106681824,
+      "learning_rate": 3.463783679285535e-05,
+      "loss": 0.1696,
+      "step": 1020
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3127197027206421,
+      "learning_rate": 3.4549150281252636e-05,
+      "loss": 0.045,
+      "step": 1021
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.13819609582424164,
+      "learning_rate": 3.446051747928202e-05,
+      "loss": 0.0203,
+      "step": 1022
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.35015806555747986,
+      "learning_rate": 3.4371938695045346e-05,
+      "loss": 0.0608,
+      "step": 1023
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3767643868923187,
+      "learning_rate": 3.428341423645668e-05,
+      "loss": 0.0686,
+      "step": 1024
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3282710611820221,
+      "learning_rate": 3.419494441124121e-05,
+      "loss": 0.0697,
+      "step": 1025
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.24474768340587616,
+      "learning_rate": 3.4106529526934306e-05,
+      "loss": 0.0583,
+      "step": 1026
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.24781620502471924,
+      "learning_rate": 3.4018169890880225e-05,
+      "loss": 0.0327,
+      "step": 1027
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.2474932074546814,
+      "learning_rate": 3.392986581023126e-05,
+      "loss": 0.0679,
+      "step": 1028
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.39474180340766907,
+      "learning_rate": 3.384161759194658e-05,
+      "loss": 0.0713,
+      "step": 1029
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.13963012397289276,
+      "learning_rate": 3.375342554279111e-05,
+      "loss": 0.0179,
+      "step": 1030
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.32144859433174133,
+      "learning_rate": 3.3665289969334585e-05,
+      "loss": 0.0447,
+      "step": 1031
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.14767055213451385,
+      "learning_rate": 3.3577211177950385e-05,
+      "loss": 0.017,
+      "step": 1032
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.3088414967060089,
+      "learning_rate": 3.348918947481452e-05,
+      "loss": 0.0483,
+      "step": 1033
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.33189231157302856,
+      "learning_rate": 3.340122516590456e-05,
+      "loss": 0.0382,
+      "step": 1034
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.4593893885612488,
+      "learning_rate": 3.3313318556998526e-05,
+      "loss": 0.0523,
+      "step": 1035
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.253412127494812,
+      "learning_rate": 3.322546995367394e-05,
+      "loss": 0.0212,
+      "step": 1036
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.42397648096084595,
+      "learning_rate": 3.3137679661306576e-05,
+      "loss": 0.087,
+      "step": 1037
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.40718910098075867,
+      "learning_rate": 3.3049947985069616e-05,
+      "loss": 0.0965,
+      "step": 1038
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.3604757487773895,
+      "learning_rate": 3.2962275229932446e-05,
+      "loss": 0.0973,
+      "step": 1039
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.2129022628068924,
+      "learning_rate": 3.287466170065959e-05,
+      "loss": 0.0384,
+      "step": 1040
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.20552971959114075,
+      "learning_rate": 3.2787107701809754e-05,
+      "loss": 0.0305,
+      "step": 1041
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.41909754276275635,
+      "learning_rate": 3.269961353773469e-05,
+      "loss": 0.0725,
+      "step": 1042
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.2766873240470886,
+      "learning_rate": 3.261217951257813e-05,
+      "loss": 0.0649,
+      "step": 1043
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.1571783572435379,
+      "learning_rate": 3.252480593027478e-05,
+      "loss": 0.0238,
+      "step": 1044
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.2809221148490906,
+      "learning_rate": 3.243749309454922e-05,
+      "loss": 0.0613,
+      "step": 1045
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.4999224841594696,
+      "learning_rate": 3.235024130891487e-05,
+      "loss": 0.0806,
+      "step": 1046
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.1895889937877655,
+      "learning_rate": 3.226305087667295e-05,
+      "loss": 0.026,
+      "step": 1047
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.220509871840477,
+      "learning_rate": 3.217592210091137e-05,
+      "loss": 0.0681,
+      "step": 1048
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.19226831197738647,
+      "learning_rate": 3.208885528450376e-05,
+      "loss": 0.0232,
+      "step": 1049
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.30779534578323364,
+      "learning_rate": 3.200185073010831e-05,
+      "loss": 0.0547,
+      "step": 1050
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.16252338886260986,
+      "learning_rate": 3.1914908740166795e-05,
+      "loss": 0.0237,
+      "step": 1051
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4130539000034332,
+      "learning_rate": 3.182802961690357e-05,
+      "loss": 0.0437,
+      "step": 1052
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.20889542996883392,
+      "learning_rate": 3.1741213662324365e-05,
+      "loss": 0.0493,
+      "step": 1053
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.27447709441185,
+      "learning_rate": 3.165446117821538e-05,
+      "loss": 0.0859,
+      "step": 1054
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.07740370184183121,
+      "learning_rate": 3.1567772466142156e-05,
+      "loss": 0.011,
+      "step": 1055
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.1462429016828537,
+      "learning_rate": 3.148114782744855e-05,
+      "loss": 0.0228,
+      "step": 1056
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.34325939416885376,
+      "learning_rate": 3.139458756325576e-05,
+      "loss": 0.0928,
+      "step": 1057
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.2712673246860504,
+      "learning_rate": 3.130809197446106e-05,
+      "loss": 0.0408,
+      "step": 1058
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.24180567264556885,
+      "learning_rate": 3.122166136173706e-05,
+      "loss": 0.0309,
+      "step": 1059
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.2045336663722992,
+      "learning_rate": 3.113529602553042e-05,
+      "loss": 0.0391,
+      "step": 1060
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.09178590029478073,
+      "learning_rate": 3.104899626606088e-05,
+      "loss": 0.0132,
+      "step": 1061
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.300592303276062,
+      "learning_rate": 3.0962762383320285e-05,
+      "loss": 0.0787,
+      "step": 1062
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.31034108996391296,
+      "learning_rate": 3.08765946770714e-05,
+      "loss": 0.0307,
+      "step": 1063
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.1603342443704605,
+      "learning_rate": 3.0790493446847024e-05,
+      "loss": 0.0209,
+      "step": 1064
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.23990066349506378,
+      "learning_rate": 3.070445899194885e-05,
+      "loss": 0.0361,
+      "step": 1065
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.15933218598365784,
+      "learning_rate": 3.061849161144641e-05,
+      "loss": 0.0137,
+      "step": 1066
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4810096025466919,
+      "learning_rate": 3.053259160417613e-05,
+      "loss": 0.1255,
+      "step": 1067
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4099353849887848,
+      "learning_rate": 3.0446759268740233e-05,
+      "loss": 0.0718,
+      "step": 1068
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.39221569895744324,
+      "learning_rate": 3.0360994903505653e-05,
+      "loss": 0.0721,
+      "step": 1069
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.48040878772735596,
+      "learning_rate": 3.02752988066031e-05,
+      "loss": 0.1077,
+      "step": 1070
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.1548435539007187,
+      "learning_rate": 3.018967127592595e-05,
+      "loss": 0.0273,
+      "step": 1071
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.37816908955574036,
+      "learning_rate": 3.010411260912922e-05,
+      "loss": 0.0421,
+      "step": 1072
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.2902359366416931,
+      "learning_rate": 3.0018623103628596e-05,
+      "loss": 0.0645,
+      "step": 1073
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.32010090351104736,
+      "learning_rate": 2.9933203056599275e-05,
+      "loss": 0.0444,
+      "step": 1074
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.47379517555236816,
+      "learning_rate": 2.984785276497507e-05,
+      "loss": 0.0423,
+      "step": 1075
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.3198046088218689,
+      "learning_rate": 2.9762572525447262e-05,
+      "loss": 0.0678,
+      "step": 1076
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.33741793036460876,
+      "learning_rate": 2.9677362634463647e-05,
+      "loss": 0.0643,
+      "step": 1077
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.2264060080051422,
+      "learning_rate": 2.9592223388227503e-05,
+      "loss": 0.0275,
+      "step": 1078
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3069595694541931,
+      "learning_rate": 2.9507155082696482e-05,
+      "loss": 0.0481,
+      "step": 1079
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.26178085803985596,
+      "learning_rate": 2.9422158013581658e-05,
+      "loss": 0.0298,
+      "step": 1080
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.4191998243331909,
+      "learning_rate": 2.93372324763465e-05,
+      "loss": 0.1156,
+      "step": 1081
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.21308496594429016,
+      "learning_rate": 2.9252378766205758e-05,
+      "loss": 0.0478,
+      "step": 1082
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.20033082365989685,
+      "learning_rate": 2.9167597178124585e-05,
+      "loss": 0.0262,
+      "step": 1083
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.21688255667686462,
+      "learning_rate": 2.9082888006817365e-05,
+      "loss": 0.0767,
+      "step": 1084
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.300791472196579,
+      "learning_rate": 2.899825154674674e-05,
+      "loss": 0.0464,
+      "step": 1085
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.2416476011276245,
+      "learning_rate": 2.8913688092122664e-05,
+      "loss": 0.0605,
+      "step": 1086
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.2521096169948578,
+      "learning_rate": 2.8829197936901232e-05,
+      "loss": 0.0293,
+      "step": 1087
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3229115605354309,
+      "learning_rate": 2.8744781374783813e-05,
+      "loss": 0.0435,
+      "step": 1088
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.0780615508556366,
+      "learning_rate": 2.8660438699215898e-05,
+      "loss": 0.0109,
+      "step": 1089
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.2879962623119354,
+      "learning_rate": 2.8576170203386143e-05,
+      "loss": 0.0665,
+      "step": 1090
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.19985683262348175,
+      "learning_rate": 2.8491976180225388e-05,
+      "loss": 0.0378,
+      "step": 1091
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.15360887348651886,
+      "learning_rate": 2.840785692240553e-05,
+      "loss": 0.0257,
+      "step": 1092
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.19224387407302856,
+      "learning_rate": 2.832381272233864e-05,
+      "loss": 0.0529,
+      "step": 1093
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.2037738561630249,
+      "learning_rate": 2.8239843872175814e-05,
+      "loss": 0.0228,
+      "step": 1094
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.380874902009964,
+      "learning_rate": 2.8155950663806235e-05,
+      "loss": 0.0525,
+      "step": 1095
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.28522253036499023,
+      "learning_rate": 2.8072133388856192e-05,
+      "loss": 0.0615,
+      "step": 1096
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.1098146066069603,
+      "learning_rate": 2.7988392338687926e-05,
+      "loss": 0.0159,
+      "step": 1097
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.2615334987640381,
+      "learning_rate": 2.7904727804398812e-05,
+      "loss": 0.0353,
+      "step": 1098
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.2041955441236496,
+      "learning_rate": 2.7821140076820162e-05,
+      "loss": 0.0187,
+      "step": 1099
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.1846192330121994,
+      "learning_rate": 2.773762944651632e-05,
+      "loss": 0.0554,
+      "step": 1100
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.17711102962493896,
+      "learning_rate": 2.765419620378366e-05,
+      "loss": 0.0342,
+      "step": 1101
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.3703756332397461,
+      "learning_rate": 2.7570840638649486e-05,
+      "loss": 0.0378,
+      "step": 1102
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4282096326351166,
+      "learning_rate": 2.7487563040871145e-05,
+      "loss": 0.0789,
+      "step": 1103
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.32506605982780457,
+      "learning_rate": 2.740436369993491e-05,
+      "loss": 0.0337,
+      "step": 1104
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.250688374042511,
+      "learning_rate": 2.7321242905055013e-05,
+      "loss": 0.0554,
+      "step": 1105
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3557257354259491,
+      "learning_rate": 2.7238200945172698e-05,
+      "loss": 0.0356,
+      "step": 1106
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3472774028778076,
+      "learning_rate": 2.715523810895515e-05,
+      "loss": 0.0348,
+      "step": 1107
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.07373315095901489,
+      "learning_rate": 2.707235468479449e-05,
+      "loss": 0.0088,
+      "step": 1108
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.21439437568187714,
+      "learning_rate": 2.6989550960806768e-05,
+      "loss": 0.0222,
+      "step": 1109
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.2730681002140045,
+      "learning_rate": 2.690682722483102e-05,
+      "loss": 0.068,
+      "step": 1110
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.41124334931373596,
+      "learning_rate": 2.6824183764428224e-05,
+      "loss": 0.086,
+      "step": 1111
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6637737154960632,
+      "learning_rate": 2.6741620866880335e-05,
+      "loss": 0.0365,
+      "step": 1112
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.425441712141037,
+      "learning_rate": 2.665913881918921e-05,
+      "loss": 0.095,
+      "step": 1113
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5520187020301819,
+      "learning_rate": 2.6576737908075668e-05,
+      "loss": 0.0514,
+      "step": 1114
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5284621119499207,
+      "learning_rate": 2.6494418419978482e-05,
+      "loss": 0.0593,
+      "step": 1115
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.08148845285177231,
+      "learning_rate": 2.641218064105341e-05,
+      "loss": 0.0084,
+      "step": 1116
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.8642109036445618,
+      "learning_rate": 2.6330024857172192e-05,
+      "loss": 0.0766,
+      "step": 1117
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.40509146451950073,
+      "learning_rate": 2.6247951353921485e-05,
+      "loss": 0.1148,
+      "step": 1118
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.2981242537498474,
+      "learning_rate": 2.616596041660194e-05,
+      "loss": 0.0666,
+      "step": 1119
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.21514151990413666,
+      "learning_rate": 2.6084052330227238e-05,
+      "loss": 0.0363,
+      "step": 1120
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.10281267762184143,
+      "learning_rate": 2.6002227379522992e-05,
+      "loss": 0.0169,
+      "step": 1121
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3236760199069977,
+      "learning_rate": 2.5920485848925913e-05,
+      "loss": 0.0296,
+      "step": 1122
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.22741632163524628,
+      "learning_rate": 2.5838828022582594e-05,
+      "loss": 0.023,
+      "step": 1123
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.3826078772544861,
+      "learning_rate": 2.5757254184348778e-05,
+      "loss": 0.0744,
+      "step": 1124
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.226307213306427,
+      "learning_rate": 2.5675764617788234e-05,
+      "loss": 0.0297,
+      "step": 1125
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.31913021206855774,
+      "learning_rate": 2.5594359606171724e-05,
+      "loss": 0.0793,
+      "step": 1126
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.2947479486465454,
+      "learning_rate": 2.5513039432476193e-05,
+      "loss": 0.1363,
+      "step": 1127
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.26046791672706604,
+      "learning_rate": 2.5431804379383523e-05,
+      "loss": 0.0727,
+      "step": 1128
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.1183793917298317,
+      "learning_rate": 2.535065472927983e-05,
+      "loss": 0.0139,
+      "step": 1129
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.23370495438575745,
+      "learning_rate": 2.526959076425434e-05,
+      "loss": 0.0503,
+      "step": 1130
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1695,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 565,
+  "total_flos": 1.0339891388035891e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}