diff --git "a/checkpoint-586/trainer_state.json" "b/checkpoint-586/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-586/trainer_state.json"
@@ -0,0 +1,4295 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0017094017094017,
+  "eval_steps": 30,
+  "global_step": 586,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0017094017094017094,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.5e-06,
+      "loss": 2.3076,
+      "step": 1
+    },
+    {
+      "epoch": 0.0017094017094017094,
+      "eval_loss": 2.248321533203125,
+      "eval_runtime": 152.6229,
+      "eval_samples_per_second": 1.677,
+      "eval_steps_per_second": 0.282,
+      "step": 1
+    },
+    {
+      "epoch": 0.003418803418803419,
+      "grad_norm": 0.671875,
+      "learning_rate": 3e-06,
+      "loss": 2.3377,
+      "step": 2
+    },
+    {
+      "epoch": 0.005128205128205128,
+      "grad_norm": 0.65625,
+      "learning_rate": 4.5e-06,
+      "loss": 2.3576,
+      "step": 3
+    },
+    {
+      "epoch": 0.006837606837606838,
+      "grad_norm": 0.6796875,
+      "learning_rate": 6e-06,
+      "loss": 2.2635,
+      "step": 4
+    },
+    {
+      "epoch": 0.008547008547008548,
+      "grad_norm": 0.7265625,
+      "learning_rate": 7.5e-06,
+      "loss": 2.4173,
+      "step": 5
+    },
+    {
+      "epoch": 0.010256410256410256,
+      "grad_norm": 0.64453125,
+      "learning_rate": 9e-06,
+      "loss": 2.2549,
+      "step": 6
+    },
+    {
+      "epoch": 0.011965811965811967,
+      "grad_norm": 0.65234375,
+      "learning_rate": 1.05e-05,
+      "loss": 2.4147,
+      "step": 7
+    },
+    {
+      "epoch": 0.013675213675213675,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.2e-05,
+      "loss": 2.3563,
+      "step": 8
+    },
+    {
+      "epoch": 0.015384615384615385,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 2.3558,
+      "step": 9
+    },
+    {
+      "epoch": 0.017094017094017096,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.5e-05,
+      "loss": 2.3426,
+      "step": 10
+    },
+    {
+      "epoch": 0.018803418803418803,
+      "grad_norm": 0.498046875,
+      "learning_rate": 1.65e-05,
+      "loss": 2.3699,
+      "step": 11
+    },
+    {
+      "epoch": 0.020512820512820513,
+      "grad_norm": 0.515625,
+      "learning_rate": 1.8e-05,
+      "loss": 2.2644,
+      "step": 12
+    },
+    {
+      "epoch": 0.022222222222222223,
+      "grad_norm": 0.5859375,
+      "learning_rate": 1.95e-05,
+      "loss": 2.2842,
+      "step": 13
+    },
+    {
+      "epoch": 0.023931623931623933,
+      "grad_norm": 0.640625,
+      "learning_rate": 2.1e-05,
+      "loss": 2.2977,
+      "step": 14
+    },
+    {
+      "epoch": 0.02564102564102564,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.25e-05,
+      "loss": 2.2834,
+      "step": 15
+    },
+    {
+      "epoch": 0.02735042735042735,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.4e-05,
+      "loss": 2.3457,
+      "step": 16
+    },
+    {
+      "epoch": 0.02905982905982906,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.55e-05,
+      "loss": 2.1915,
+      "step": 17
+    },
+    {
+      "epoch": 0.03076923076923077,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 2.264,
+      "step": 18
+    },
+    {
+      "epoch": 0.03247863247863248,
+      "grad_norm": 0.55078125,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 2.2251,
+      "step": 19
+    },
+    {
+      "epoch": 0.03418803418803419,
+      "grad_norm": 0.55859375,
+      "learning_rate": 3e-05,
+      "loss": 2.2482,
+      "step": 20
+    },
+    {
+      "epoch": 0.035897435897435895,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.9999946827306146e-05,
+      "loss": 2.3187,
+      "step": 21
+    },
+    {
+      "epoch": 0.037606837606837605,
+      "grad_norm": 0.57421875,
+      "learning_rate": 2.9999787309621396e-05,
+      "loss": 2.329,
+      "step": 22
+    },
+    {
+      "epoch": 0.039316239316239315,
+      "grad_norm": 0.55859375,
+      "learning_rate": 2.9999521448136205e-05,
+      "loss": 2.1736,
+      "step": 23
+    },
+    {
+      "epoch": 0.041025641025641026,
+      "grad_norm": 0.53515625,
+      "learning_rate": 2.999914924483466e-05,
+      "loss": 2.2928,
+      "step": 24
+    },
+    {
+      "epoch": 0.042735042735042736,
+      "grad_norm": 0.5390625,
+      "learning_rate": 2.999867070249444e-05,
+      "loss": 2.3343,
+      "step": 25
+    },
+    {
+      "epoch": 0.044444444444444446,
+      "grad_norm": 0.5078125,
+      "learning_rate": 2.9998085824686838e-05,
+      "loss": 2.3226,
+      "step": 26
+    },
+    {
+      "epoch": 0.046153846153846156,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.9997394615776692e-05,
+      "loss": 2.2437,
+      "step": 27
+    },
+    {
+      "epoch": 0.04786324786324787,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.999659708092239e-05,
+      "loss": 2.2302,
+      "step": 28
+    },
+    {
+      "epoch": 0.04957264957264957,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.9995693226075796e-05,
+      "loss": 2.1686,
+      "step": 29
+    },
+    {
+      "epoch": 0.05128205128205128,
+      "grad_norm": 0.5,
+      "learning_rate": 2.9994683057982227e-05,
+      "loss": 2.2699,
+      "step": 30
+    },
+    {
+      "epoch": 0.05128205128205128,
+      "eval_loss": 2.2213175296783447,
+      "eval_runtime": 153.4952,
+      "eval_samples_per_second": 1.668,
+      "eval_steps_per_second": 0.28,
+      "step": 30
+    },
+    {
+      "epoch": 0.05299145299145299,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.9993566584180403e-05,
+      "loss": 2.2601,
+      "step": 31
+    },
+    {
+      "epoch": 0.0547008547008547,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.9992343813002384e-05,
+      "loss": 2.3567,
+      "step": 32
+    },
+    {
+      "epoch": 0.05641025641025641,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.99910147535735e-05,
+      "loss": 2.3461,
+      "step": 33
+    },
+    {
+      "epoch": 0.05811965811965812,
+      "grad_norm": 0.5,
+      "learning_rate": 2.9989579415812313e-05,
+      "loss": 2.2736,
+      "step": 34
+    },
+    {
+      "epoch": 0.05982905982905983,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.998803781043051e-05,
+      "loss": 2.4475,
+      "step": 35
+    },
+    {
+      "epoch": 0.06153846153846154,
+      "grad_norm": 0.48828125,
+      "learning_rate": 2.9986389948932825e-05,
+      "loss": 2.2864,
+      "step": 36
+    },
+    {
+      "epoch": 0.06324786324786325,
+      "grad_norm": 0.53515625,
+      "learning_rate": 2.998463584361699e-05,
+      "loss": 2.2971,
+      "step": 37
+    },
+    {
+      "epoch": 0.06495726495726496,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.9982775507573586e-05,
+      "loss": 2.1686,
+      "step": 38
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.9980808954686005e-05,
+      "loss": 2.2225,
+      "step": 39
+    },
+    {
+      "epoch": 0.06837606837606838,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.9978736199630297e-05,
+      "loss": 2.1414,
+      "step": 40
+    },
+    {
+      "epoch": 0.07008547008547009,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.99765572578751e-05,
+      "loss": 2.2085,
+      "step": 41
+    },
+    {
+      "epoch": 0.07179487179487179,
+      "grad_norm": 0.625,
+      "learning_rate": 2.9974272145681487e-05,
+      "loss": 2.3185,
+      "step": 42
+    },
+    {
+      "epoch": 0.0735042735042735,
+      "grad_norm": 0.494140625,
+      "learning_rate": 2.997188088010287e-05,
+      "loss": 2.1781,
+      "step": 43
+    },
+    {
+      "epoch": 0.07521367521367521,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.9969383478984875e-05,
+      "loss": 2.1979,
+      "step": 44
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.9966779960965186e-05,
+      "loss": 2.1786,
+      "step": 45
+    },
+    {
+      "epoch": 0.07863247863247863,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.9964070345473427e-05,
+      "loss": 2.2664,
+      "step": 46
+    },
+    {
+      "epoch": 0.08034188034188035,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.9961254652731014e-05,
+      "loss": 2.1936,
+      "step": 47
+    },
+    {
+      "epoch": 0.08205128205128205,
+      "grad_norm": 0.6171875,
+      "learning_rate": 2.9958332903750987e-05,
+      "loss": 2.3133,
+      "step": 48
+    },
+    {
+      "epoch": 0.08376068376068375,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.9955305120337887e-05,
+      "loss": 2.2156,
+      "step": 49
+    },
+    {
+      "epoch": 0.08547008547008547,
+      "grad_norm": 0.51953125,
+      "learning_rate": 2.9952171325087547e-05,
+      "loss": 2.4135,
+      "step": 50
+    },
+    {
+      "epoch": 0.08717948717948718,
+      "grad_norm": 0.921875,
+      "learning_rate": 2.994893154138697e-05,
+      "loss": 2.2793,
+      "step": 51
+    },
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.9945585793414117e-05,
+      "loss": 2.2301,
+      "step": 52
+    },
+    {
+      "epoch": 0.0905982905982906,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.9942134106137754e-05,
+      "loss": 2.2878,
+      "step": 53
+    },
+    {
+      "epoch": 0.09230769230769231,
+      "grad_norm": 1.1328125,
+      "learning_rate": 2.9938576505317253e-05,
+      "loss": 2.3196,
+      "step": 54
+    },
+    {
+      "epoch": 0.09401709401709402,
+      "grad_norm": 0.5078125,
+      "learning_rate": 2.99349130175024e-05,
+      "loss": 2.3061,
+      "step": 55
+    },
+    {
+      "epoch": 0.09572649572649573,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.9931143670033195e-05,
+      "loss": 2.1972,
+      "step": 56
+    },
+    {
+      "epoch": 0.09743589743589744,
+      "grad_norm": 0.490234375,
+      "learning_rate": 2.9927268491039653e-05,
+      "loss": 2.3351,
+      "step": 57
+    },
+    {
+      "epoch": 0.09914529914529914,
+      "grad_norm": 0.482421875,
+      "learning_rate": 2.992328750944159e-05,
+      "loss": 2.2107,
+      "step": 58
+    },
+    {
+      "epoch": 0.10085470085470086,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.9919200754948417e-05,
+      "loss": 2.2458,
+      "step": 59
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.99150082580589e-05,
+      "loss": 2.1415,
+      "step": 60
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "eval_loss": 2.194425106048584,
+      "eval_runtime": 153.5382,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 60
+    },
+    {
+      "epoch": 0.10427350427350428,
+      "grad_norm": 0.486328125,
+      "learning_rate": 2.9910710050060954e-05,
+      "loss": 2.23,
+      "step": 61
+    },
+    {
+      "epoch": 0.10598290598290598,
+      "grad_norm": 0.5078125,
+      "learning_rate": 2.9906306163031388e-05,
+      "loss": 2.1581,
+      "step": 62
+    },
+    {
+      "epoch": 0.1076923076923077,
+      "grad_norm": 0.54296875,
+      "learning_rate": 2.9901796629835686e-05,
+      "loss": 2.1723,
+      "step": 63
+    },
+    {
+      "epoch": 0.1094017094017094,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.9897181484127743e-05,
+      "loss": 2.2641,
+      "step": 64
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 0.55078125,
+      "learning_rate": 2.9892460760349624e-05,
+      "loss": 2.2183,
+      "step": 65
+    },
+    {
+      "epoch": 0.11282051282051282,
+      "grad_norm": 0.53515625,
+      "learning_rate": 2.9887634493731312e-05,
+      "loss": 2.2151,
+      "step": 66
+    },
+    {
+      "epoch": 0.11452991452991453,
+      "grad_norm": 0.54296875,
+      "learning_rate": 2.988270272029043e-05,
+      "loss": 2.206,
+      "step": 67
+    },
+    {
+      "epoch": 0.11623931623931624,
+      "grad_norm": 0.51953125,
+      "learning_rate": 2.987766547683199e-05,
+      "loss": 2.1754,
+      "step": 68
+    },
+    {
+      "epoch": 0.11794871794871795,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.9872522800948106e-05,
+      "loss": 2.2236,
+      "step": 69
+    },
+    {
+      "epoch": 0.11965811965811966,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.9867274731017714e-05,
+      "loss": 2.222,
+      "step": 70
+    },
+    {
+      "epoch": 0.12136752136752137,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.9861921306206287e-05,
+      "loss": 2.0767,
+      "step": 71
+    },
+    {
+      "epoch": 0.12307692307692308,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.9856462566465556e-05,
+      "loss": 2.2204,
+      "step": 72
+    },
+    {
+      "epoch": 0.12478632478632479,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.9850898552533175e-05,
+      "loss": 2.3901,
+      "step": 73
+    },
+    {
+      "epoch": 0.1264957264957265,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.984522930593247e-05,
+      "loss": 2.1304,
+      "step": 74
+    },
+    {
+      "epoch": 0.1282051282051282,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.9839454868972085e-05,
+      "loss": 2.1427,
+      "step": 75
+    },
+    {
+      "epoch": 0.12991452991452992,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.9833575284745687e-05,
+      "loss": 2.235,
+      "step": 76
+    },
+    {
+      "epoch": 0.13162393162393163,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.9827590597131636e-05,
+      "loss": 2.316,
+      "step": 77
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.9821500850792668e-05,
+      "loss": 2.2714,
+      "step": 78
+    },
+    {
+      "epoch": 0.13504273504273503,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.981530609117554e-05,
+      "loss": 2.2249,
+      "step": 79
+    },
+    {
+      "epoch": 0.13675213675213677,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.980900636451073e-05,
+      "loss": 2.1681,
+      "step": 80
+    },
+    {
+      "epoch": 0.13846153846153847,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.9802601717812045e-05,
+      "loss": 2.3345,
+      "step": 81
+    },
+    {
+      "epoch": 0.14017094017094017,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.9796092198876305e-05,
+      "loss": 2.1506,
+      "step": 82
+    },
+    {
+      "epoch": 0.14188034188034188,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.9789477856282968e-05,
+      "loss": 2.2908,
+      "step": 83
+    },
+    {
+      "epoch": 0.14358974358974358,
+      "grad_norm": 0.57421875,
+      "learning_rate": 2.978275873939378e-05,
+      "loss": 2.2495,
+      "step": 84
+    },
+    {
+      "epoch": 0.1452991452991453,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.977593489835239e-05,
+      "loss": 2.186,
+      "step": 85
+    },
+    {
+      "epoch": 0.147008547008547,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.9769006384084e-05,
+      "loss": 2.1442,
+      "step": 86
+    },
+    {
+      "epoch": 0.14871794871794872,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.976197324829496e-05,
+      "loss": 2.1663,
+      "step": 87
+    },
+    {
+      "epoch": 0.15042735042735042,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.9754835543472396e-05,
+      "loss": 2.3458,
+      "step": 88
+    },
+    {
+      "epoch": 0.15213675213675212,
+      "grad_norm": 0.55078125,
+      "learning_rate": 2.9747593322883817e-05,
+      "loss": 2.1568,
+      "step": 89
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.9740246640576713e-05,
+      "loss": 2.2196,
+      "step": 90
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "eval_loss": 2.1906561851501465,
+      "eval_runtime": 153.5179,
+      "eval_samples_per_second": 1.668,
+      "eval_steps_per_second": 0.28,
+      "step": 90
+    },
+    {
+      "epoch": 0.15555555555555556,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.9732795551378154e-05,
+      "loss": 2.1843,
+      "step": 91
+    },
+    {
+      "epoch": 0.15726495726495726,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.972524011089439e-05,
+      "loss": 2.2099,
+      "step": 92
+    },
+    {
+      "epoch": 0.15897435897435896,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.971758037551042e-05,
+      "loss": 2.1464,
+      "step": 93
+    },
+    {
+      "epoch": 0.1606837606837607,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.9709816402389573e-05,
+      "loss": 2.2127,
+      "step": 94
+    },
+    {
+      "epoch": 0.1623931623931624,
+      "grad_norm": 0.55859375,
+      "learning_rate": 2.9701948249473103e-05,
+      "loss": 2.2087,
+      "step": 95
+    },
+    {
+      "epoch": 0.1641025641025641,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.9693975975479723e-05,
+      "loss": 2.297,
+      "step": 96
+    },
+    {
+      "epoch": 0.1658119658119658,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.968589963990519e-05,
+      "loss": 2.1285,
+      "step": 97
+    },
+    {
+      "epoch": 0.1675213675213675,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.9677719303021867e-05,
+      "loss": 2.1603,
+      "step": 98
+    },
+    {
+      "epoch": 0.16923076923076924,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.9669435025878246e-05,
+      "loss": 2.1093,
+      "step": 99
+    },
+    {
+      "epoch": 0.17094017094017094,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.966104687029851e-05,
+      "loss": 2.325,
+      "step": 100
+    },
+    {
+      "epoch": 0.17264957264957265,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.9652554898882082e-05,
+      "loss": 2.0521,
+      "step": 101
+    },
+    {
+      "epoch": 0.17435897435897435,
+      "grad_norm": 0.58203125,
+      "learning_rate": 2.9643959175003137e-05,
+      "loss": 2.3031,
+      "step": 102
+    },
+    {
+      "epoch": 0.17606837606837608,
+      "grad_norm": 0.51953125,
+      "learning_rate": 2.963525976281013e-05,
+      "loss": 2.1553,
+      "step": 103
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.9626456727225336e-05,
+      "loss": 2.1681,
+      "step": 104
+    },
+    {
+      "epoch": 0.1794871794871795,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.961755013394435e-05,
+      "loss": 2.2048,
+      "step": 105
+    },
+    {
+      "epoch": 0.1811965811965812,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.9608540049435603e-05,
+      "loss": 2.2213,
+      "step": 106
+    },
+    {
+      "epoch": 0.1829059829059829,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.959942654093986e-05,
+      "loss": 2.1712,
+      "step": 107
+    },
+    {
+      "epoch": 0.18461538461538463,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.9590209676469725e-05,
+      "loss": 2.1191,
+      "step": 108
+    },
+    {
+      "epoch": 0.18632478632478633,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.958088952480913e-05,
+      "loss": 2.1237,
+      "step": 109
+    },
+    {
+      "epoch": 0.18803418803418803,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.9571466155512816e-05,
+      "loss": 2.1009,
+      "step": 110
+    },
+    {
+      "epoch": 0.18974358974358974,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.956193963890583e-05,
+      "loss": 2.1874,
+      "step": 111
+    },
+    {
+      "epoch": 0.19145299145299147,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.9552310046082996e-05,
+      "loss": 2.1853,
+      "step": 112
+    },
+    {
+      "epoch": 0.19316239316239317,
+      "grad_norm": 0.78515625,
+      "learning_rate": 2.954257744890835e-05,
+      "loss": 2.1964,
+      "step": 113
+    },
+    {
+      "epoch": 0.19487179487179487,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.9532741920014666e-05,
+      "loss": 2.1841,
+      "step": 114
+    },
+    {
+      "epoch": 0.19658119658119658,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.9522803532802863e-05,
+      "loss": 2.1514,
+      "step": 115
+    },
+    {
+      "epoch": 0.19829059829059828,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.9512762361441473e-05,
+      "loss": 2.1976,
+      "step": 116
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.9502618480866093e-05,
+      "loss": 2.2175,
+      "step": 117
+    },
+    {
+      "epoch": 0.20170940170940171,
+      "grad_norm": 0.5390625,
+      "learning_rate": 2.9492371966778828e-05,
+      "loss": 2.3224,
+      "step": 118
+    },
+    {
+      "epoch": 0.20341880341880342,
+      "grad_norm": 0.6171875,
+      "learning_rate": 2.948202289564771e-05,
+      "loss": 2.0922,
+      "step": 119
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.947157134470614e-05,
+      "loss": 2.1286,
+      "step": 120
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "eval_loss": 2.1840789318084717,
+      "eval_runtime": 153.5679,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 120
+    },
+    {
+      "epoch": 0.20683760683760682,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.946101739195231e-05,
+      "loss": 2.3025,
+      "step": 121
+    },
+    {
+      "epoch": 0.20854700854700856,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.945036111614863e-05,
+      "loss": 2.2483,
+      "step": 122
+    },
+    {
+      "epoch": 0.21025641025641026,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.9439602596821103e-05,
+      "loss": 2.1793,
+      "step": 123
+    },
+    {
+      "epoch": 0.21196581196581196,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.9428741914258788e-05,
+      "loss": 2.1217,
+      "step": 124
+    },
+    {
+      "epoch": 0.21367521367521367,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.941777914951315e-05,
+      "loss": 2.2512,
+      "step": 125
+    },
+    {
+      "epoch": 0.2153846153846154,
+      "grad_norm": 0.58203125,
+      "learning_rate": 2.9406714384397493e-05,
+      "loss": 2.1602,
+      "step": 126
+    },
+    {
+      "epoch": 0.2170940170940171,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.9395547701486322e-05,
+      "loss": 2.1697,
+      "step": 127
+    },
+    {
+      "epoch": 0.2188034188034188,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.938427918411474e-05,
+      "loss": 2.1518,
+      "step": 128
+    },
+    {
+      "epoch": 0.2205128205128205,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.937290891637782e-05,
+      "loss": 2.1246,
+      "step": 129
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.9361436983129985e-05,
+      "loss": 2.1706,
+      "step": 130
+    },
+    {
+      "epoch": 0.22393162393162394,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.934986346998438e-05,
+      "loss": 2.1758,
+      "step": 131
+    },
+    {
+      "epoch": 0.22564102564102564,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.933818846331221e-05,
+      "loss": 2.1348,
+      "step": 132
+    },
+    {
+      "epoch": 0.22735042735042735,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.9326412050242115e-05,
+      "loss": 2.133,
+      "step": 133
+    },
+    {
+      "epoch": 0.22905982905982905,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.9314534318659512e-05,
+      "loss": 2.1673,
+      "step": 134
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.62890625,
+      "learning_rate": 2.930255535720594e-05,
+      "loss": 2.1968,
+      "step": 135
+    },
+    {
+      "epoch": 0.23247863247863249,
+      "grad_norm": 0.55859375,
+      "learning_rate": 2.929047525527841e-05,
+      "loss": 2.1598,
+      "step": 136
+    },
+    {
+      "epoch": 0.2341880341880342,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.927829410302871e-05,
+      "loss": 2.2926,
+      "step": 137
+    },
+    {
+      "epoch": 0.2358974358974359,
+      "grad_norm": 0.625,
+      "learning_rate": 2.926601199136277e-05,
+      "loss": 2.2654,
+      "step": 138
+    },
+    {
+      "epoch": 0.2376068376068376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.9253629011939953e-05,
+      "loss": 2.1086,
+      "step": 139
+    },
+    {
+      "epoch": 0.23931623931623933,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.9241145257172376e-05,
+      "loss": 2.1793,
+      "step": 140
+    },
+    {
+      "epoch": 0.24102564102564103,
+      "grad_norm": 0.58203125,
+      "learning_rate": 2.9228560820224233e-05,
+      "loss": 2.1496,
+      "step": 141
+    },
+    {
+      "epoch": 0.24273504273504273,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.9215875795011088e-05,
+      "loss": 2.1907,
+      "step": 142
+    },
+    {
+      "epoch": 0.24444444444444444,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.9203090276199183e-05,
+      "loss": 2.1389,
+      "step": 143
+    },
+    {
+      "epoch": 0.24615384615384617,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.9190204359204736e-05,
+      "loss": 2.1267,
+      "step": 144
+    },
+    {
+      "epoch": 0.24786324786324787,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.9177218140193194e-05,
+      "loss": 2.2647,
+      "step": 145
+    },
+    {
+      "epoch": 0.24957264957264957,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.916413171607856e-05,
+      "loss": 2.1353,
+      "step": 146
+    },
+    {
+      "epoch": 0.2512820512820513,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.9150945184522657e-05,
+      "loss": 2.1386,
+      "step": 147
+    },
+    {
+      "epoch": 0.252991452991453,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.9137658643934374e-05,
+      "loss": 2.13,
+      "step": 148
+    },
+    {
+      "epoch": 0.2547008547008547,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.9124272193468964e-05,
+      "loss": 2.2023,
+      "step": 149
+    },
+    {
+      "epoch": 0.2564102564102564,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.911078593302729e-05,
+      "loss": 2.1163,
+      "step": 150
+    },
+    {
+      "epoch": 0.2564102564102564,
+      "eval_loss": 2.1831252574920654,
+      "eval_runtime": 153.5476,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 150
+    },
+    {
+      "epoch": 0.25811965811965815,
+      "grad_norm": 0.55859375,
+      "learning_rate": 2.9097199963255074e-05,
+      "loss": 2.1386,
+      "step": 151
+    },
+    {
+      "epoch": 0.25982905982905985,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.9083514385542162e-05,
+      "loss": 2.1316,
+      "step": 152
+    },
+    {
+      "epoch": 0.26153846153846155,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.9069729302021747e-05,
+      "loss": 2.2437,
+      "step": 153
+    },
+    {
+      "epoch": 0.26324786324786326,
+      "grad_norm": 0.62890625,
+      "learning_rate": 2.905584481556962e-05,
+      "loss": 2.1822,
+      "step": 154
+    },
+    {
+      "epoch": 0.26495726495726496,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.9041861029803407e-05,
+      "loss": 2.2614,
+      "step": 155
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.55859375,
+      "learning_rate": 2.9027778049081785e-05,
+      "loss": 2.2387,
+      "step": 156
+    },
+    {
+      "epoch": 0.26837606837606837,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.9013595978503695e-05,
+      "loss": 2.124,
+      "step": 157
+    },
+    {
+      "epoch": 0.27008547008547007,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.8999314923907584e-05,
+      "loss": 2.0662,
+      "step": 158
+    },
+    {
+      "epoch": 0.2717948717948718,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.8984934991870598e-05,
+      "loss": 2.1483,
+      "step": 159
+    },
+    {
+      "epoch": 0.27350427350427353,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.8970456289707778e-05,
+      "loss": 2.1428,
+      "step": 160
+    },
+    {
+      "epoch": 0.27521367521367524,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.895587892547129e-05,
+      "loss": 2.1845,
+      "step": 161
+    },
+    {
+      "epoch": 0.27692307692307694,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.8941203007949575e-05,
+      "loss": 1.8986,
+      "step": 162
+    },
+    {
+      "epoch": 0.27863247863247864,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.8926428646666576e-05,
+      "loss": 2.1513,
+      "step": 163
+    },
+    {
+      "epoch": 0.28034188034188035,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.891155595188091e-05,
+      "loss": 2.1334,
+      "step": 164
+    },
+    {
+      "epoch": 0.28205128205128205,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.8896585034585022e-05,
+      "loss": 2.08,
+      "step": 165
+    },
+    {
+      "epoch": 0.28376068376068375,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.8881516006504388e-05,
+      "loss": 2.0936,
+      "step": 166
+    },
+    {
+      "epoch": 0.28547008547008546,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.886634898009668e-05,
+      "loss": 2.1244,
+      "step": 167
+    },
+    {
+      "epoch": 0.28717948717948716,
+      "grad_norm": 0.55859375,
+      "learning_rate": 2.885108406855088e-05,
+      "loss": 2.1053,
+      "step": 168
+    },
+    {
+      "epoch": 0.28888888888888886,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.8835721385786506e-05,
+      "loss": 2.067,
+      "step": 169
+    },
+    {
+      "epoch": 0.2905982905982906,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.882026104645271e-05,
+      "loss": 2.1022,
+      "step": 170
+    },
+    {
+      "epoch": 0.2923076923076923,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.8804703165927433e-05,
+      "loss": 2.1366,
+      "step": 171
+    },
+    {
+      "epoch": 0.294017094017094,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.878904786031658e-05,
+      "loss": 1.9787,
+      "step": 172
+    },
+    {
+      "epoch": 0.29572649572649573,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.877329524645308e-05,
+      "loss": 2.1275,
+      "step": 173
+    },
+    {
+      "epoch": 0.29743589743589743,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.8757445441896086e-05,
+      "loss": 2.0948,
+      "step": 174
+    },
+    {
+      "epoch": 0.29914529914529914,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.8741498564930065e-05,
+      "loss": 2.0375,
+      "step": 175
+    },
+    {
+      "epoch": 0.30085470085470084,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.8725454734563908e-05,
+      "loss": 2.0992,
+      "step": 176
+    },
+    {
+      "epoch": 0.30256410256410254,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.8709314070530077e-05,
+      "loss": 2.0009,
+      "step": 177
+    },
+    {
+      "epoch": 0.30427350427350425,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.8693076693283656e-05,
+      "loss": 2.0853,
+      "step": 178
+    },
+    {
+      "epoch": 0.305982905982906,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.8676742724001504e-05,
+      "loss": 2.1654,
+      "step": 179
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.8660312284581332e-05,
+      "loss": 2.0999,
+      "step": 180
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "eval_loss": 2.1848390102386475,
+      "eval_runtime": 153.653,
+      "eval_samples_per_second": 1.666,
+      "eval_steps_per_second": 0.28,
+      "step": 180
+    },
+    {
+      "epoch": 0.3094017094017094,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.8643785497640778e-05,
+      "loss": 2.0686,
+      "step": 181
+    },
+    {
+      "epoch": 0.3111111111111111,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.8627162486516516e-05,
+      "loss": 2.1621,
+      "step": 182
+    },
+    {
+      "epoch": 0.3128205128205128,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.8610443375263328e-05,
+      "loss": 2.2083,
+      "step": 183
+    },
+    {
+      "epoch": 0.3145299145299145,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.8593628288653162e-05,
+      "loss": 2.1462,
+      "step": 184
+    },
+    {
+      "epoch": 0.3162393162393162,
+      "grad_norm": 0.58203125,
+      "learning_rate": 2.857671735217423e-05,
+      "loss": 2.0359,
+      "step": 185
+    },
+    {
+      "epoch": 0.31794871794871793,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.8559710692030032e-05,
+      "loss": 2.06,
+      "step": 186
+    },
+    {
+      "epoch": 0.31965811965811963,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.8542608435138472e-05,
+      "loss": 2.0161,
+      "step": 187
+    },
+    {
+      "epoch": 0.3213675213675214,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.8525410709130844e-05,
+      "loss": 2.0891,
+      "step": 188
+    },
+    {
+      "epoch": 0.3230769230769231,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.8508117642350934e-05,
+      "loss": 2.161,
+      "step": 189
+    },
+    {
+      "epoch": 0.3247863247863248,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.849072936385403e-05,
+      "loss": 2.114,
+      "step": 190
+    },
+    {
+      "epoch": 0.3264957264957265,
+      "grad_norm": 0.625,
+      "learning_rate": 2.8473246003405977e-05,
+      "loss": 2.0737,
+      "step": 191
+    },
+    {
+      "epoch": 0.3282051282051282,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.8455667691482186e-05,
+      "loss": 2.1467,
+      "step": 192
+    },
+    {
+      "epoch": 0.3299145299145299,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.8437994559266685e-05,
+      "loss": 2.3981,
+      "step": 193
+    },
+    {
+      "epoch": 0.3316239316239316,
+      "grad_norm": 0.6796875,
+      "learning_rate": 2.8420226738651142e-05,
+      "loss": 2.1642,
+      "step": 194
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.8402364362233846e-05,
+      "loss": 2.0835,
+      "step": 195
+    },
+    {
+      "epoch": 0.335042735042735,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.8384407563318758e-05,
+      "loss": 2.1733,
+      "step": 196
+    },
+    {
+      "epoch": 0.3367521367521368,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.8366356475914487e-05,
+      "loss": 2.1804,
+      "step": 197
+    },
+    {
+      "epoch": 0.3384615384615385,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.834821123473331e-05,
+      "loss": 2.0218,
+      "step": 198
+    },
+    {
+      "epoch": 0.3401709401709402,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.8329971975190146e-05,
+      "loss": 2.137,
+      "step": 199
+    },
+    {
+      "epoch": 0.3418803418803419,
+      "grad_norm": 0.640625,
+      "learning_rate": 2.831163883340157e-05,
+      "loss": 2.1711,
+      "step": 200
+    },
+    {
+      "epoch": 0.3435897435897436,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.8293211946184787e-05,
+      "loss": 2.2119,
+      "step": 201
+    },
+    {
+      "epoch": 0.3452991452991453,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.827469145105659e-05,
+      "loss": 2.1208,
+      "step": 202
+    },
+    {
+      "epoch": 0.347008547008547,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.8256077486232368e-05,
+      "loss": 2.1377,
+      "step": 203
+    },
+    {
+      "epoch": 0.3487179487179487,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.8237370190625056e-05,
+      "loss": 2.0869,
+      "step": 204
+    },
+    {
+      "epoch": 0.3504273504273504,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.8218569703844104e-05,
+      "loss": 2.066,
+      "step": 205
+    },
+    {
+      "epoch": 0.35213675213675216,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.819967616619442e-05,
+      "loss": 2.1472,
+      "step": 206
+    },
+    {
+      "epoch": 0.35384615384615387,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.818068971867535e-05,
+      "loss": 2.3903,
+      "step": 207
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.8161610502979598e-05,
+      "loss": 2.154,
+      "step": 208
+    },
+    {
+      "epoch": 0.3572649572649573,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.8142438661492194e-05,
+      "loss": 2.143,
+      "step": 209
+    },
+    {
+      "epoch": 0.358974358974359,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.8123174337289405e-05,
+      "loss": 2.0984,
+      "step": 210
+    },
+    {
+      "epoch": 0.358974358974359,
+      "eval_loss": 2.177856683731079,
+      "eval_runtime": 153.5529,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 210
+    },
+    {
+      "epoch": 0.3606837606837607,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.8103817674137697e-05,
+      "loss": 2.0901,
+      "step": 211
+    },
+    {
+      "epoch": 0.3623931623931624,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.8084368816492643e-05,
+      "loss": 2.1651,
+      "step": 212
+    },
+    {
+      "epoch": 0.3641025641025641,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.8064827909497826e-05,
+      "loss": 2.0036,
+      "step": 213
+    },
+    {
+      "epoch": 0.3658119658119658,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.8045195098983813e-05,
+      "loss": 2.0947,
+      "step": 214
+    },
+    {
+      "epoch": 0.36752136752136755,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.8025470531467e-05,
+      "loss": 2.124,
+      "step": 215
+    },
+    {
+      "epoch": 0.36923076923076925,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.8005654354148578e-05,
+      "loss": 2.0957,
+      "step": 216
+    },
+    {
+      "epoch": 0.37094017094017095,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.7985746714913383e-05,
+      "loss": 2.1268,
+      "step": 217
+    },
+    {
+      "epoch": 0.37264957264957266,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.796574776232883e-05,
+      "loss": 2.1524,
+      "step": 218
+    },
+    {
+      "epoch": 0.37435897435897436,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.7945657645643788e-05,
+      "loss": 2.0742,
+      "step": 219
+    },
+    {
+      "epoch": 0.37606837606837606,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.7925476514787462e-05,
+      "loss": 2.0809,
+      "step": 220
+    },
+    {
+      "epoch": 0.37777777777777777,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.7905204520368297e-05,
+      "loss": 2.0084,
+      "step": 221
+    },
+    {
+      "epoch": 0.37948717948717947,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.7884841813672812e-05,
+      "loss": 2.1647,
+      "step": 222
+    },
+    {
+      "epoch": 0.3811965811965812,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.7864388546664528e-05,
+      "loss": 2.101,
+      "step": 223
+    },
+    {
+      "epoch": 0.38290598290598293,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.7843844871982773e-05,
+      "loss": 2.2433,
+      "step": 224
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.62890625,
+      "learning_rate": 2.7823210942941592e-05,
+      "loss": 2.3129,
+      "step": 225
+    },
+    {
+      "epoch": 0.38632478632478634,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.7802486913528573e-05,
+      "loss": 2.069,
+      "step": 226
+    },
+    {
+      "epoch": 0.38803418803418804,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.7781672938403715e-05,
+      "loss": 2.1515,
+      "step": 227
+    },
+    {
+      "epoch": 0.38974358974358975,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.776076917289826e-05,
+      "loss": 2.1224,
+      "step": 228
+    },
+    {
+      "epoch": 0.39145299145299145,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.7739775773013553e-05,
+      "loss": 2.0911,
+      "step": 229
+    },
+    {
+      "epoch": 0.39316239316239315,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.771869289541985e-05,
+      "loss": 2.03,
+      "step": 230
+    },
+    {
+      "epoch": 0.39487179487179486,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.7697520697455174e-05,
+      "loss": 2.1555,
+      "step": 231
+    },
+    {
+      "epoch": 0.39658119658119656,
+      "grad_norm": 0.87890625,
+      "learning_rate": 2.7676259337124127e-05,
+      "loss": 2.1044,
+      "step": 232
+    },
+    {
+      "epoch": 0.39829059829059826,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.765490897309672e-05,
+      "loss": 2.0823,
+      "step": 233
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.7633469764707173e-05,
+      "loss": 2.22,
+      "step": 234
+    },
+    {
+      "epoch": 0.4017094017094017,
+      "grad_norm": 0.6171875,
+      "learning_rate": 2.761194187195275e-05,
+      "loss": 2.1114,
+      "step": 235
+    },
+    {
+      "epoch": 0.40341880341880343,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.7590325455492544e-05,
+      "loss": 2.1627,
+      "step": 236
+    },
+    {
+      "epoch": 0.40512820512820513,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.756862067664629e-05,
+      "loss": 2.0677,
+      "step": 237
+    },
+    {
+      "epoch": 0.40683760683760684,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.754682769739315e-05,
+      "loss": 2.2179,
+      "step": 238
+    },
+    {
+      "epoch": 0.40854700854700854,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.7524946680370525e-05,
+      "loss": 2.1264,
+      "step": 239
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 0.62890625,
+      "learning_rate": 2.7502977788872803e-05,
+      "loss": 2.0791,
+      "step": 240
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "eval_loss": 2.1732006072998047,
+      "eval_runtime": 153.5496,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 240
+    },
+    {
+      "epoch": 0.41196581196581195,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.748092118685019e-05,
+      "loss": 2.1464,
+      "step": 241
+    },
+    {
+      "epoch": 0.41367521367521365,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.7458777038907448e-05,
+      "loss": 2.1003,
+      "step": 242
+    },
+    {
+      "epoch": 0.4153846153846154,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.7436545510302684e-05,
+      "loss": 2.1369,
+      "step": 243
+    },
+    {
+      "epoch": 0.4170940170940171,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.7414226766946108e-05,
+      "loss": 2.1722,
+      "step": 244
+    },
+    {
+      "epoch": 0.4188034188034188,
+      "grad_norm": 0.58203125,
+      "learning_rate": 2.7391820975398797e-05,
+      "loss": 2.0298,
+      "step": 245
+    },
+    {
+      "epoch": 0.4205128205128205,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.7369328302871473e-05,
+      "loss": 2.2088,
+      "step": 246
+    },
+    {
+      "epoch": 0.4222222222222222,
+      "grad_norm": 0.640625,
+      "learning_rate": 2.7346748917223208e-05,
+      "loss": 1.8934,
+      "step": 247
+    },
+    {
+      "epoch": 0.4239316239316239,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.732408298696022e-05,
+      "loss": 2.0064,
+      "step": 248
+    },
+    {
+      "epoch": 0.4256410256410256,
+      "grad_norm": 0.625,
+      "learning_rate": 2.730133068123459e-05,
+      "loss": 2.3159,
+      "step": 249
+    },
+    {
+      "epoch": 0.42735042735042733,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.7278492169843002e-05,
+      "loss": 2.091,
+      "step": 250
+    },
+    {
+      "epoch": 0.42905982905982903,
+      "grad_norm": 0.625,
+      "learning_rate": 2.7255567623225484e-05,
+      "loss": 2.1577,
+      "step": 251
+    },
+    {
+      "epoch": 0.4307692307692308,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.723255721246412e-05,
+      "loss": 2.185,
+      "step": 252
+    },
+    {
+      "epoch": 0.4324786324786325,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.72094611092818e-05,
+      "loss": 2.1338,
+      "step": 253
+    },
+    {
+      "epoch": 0.4341880341880342,
+      "grad_norm": 0.62890625,
+      "learning_rate": 2.7186279486040902e-05,
+      "loss": 2.058,
+      "step": 254
+    },
+    {
+      "epoch": 0.4358974358974359,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.7163012515742044e-05,
+      "loss": 2.0945,
+      "step": 255
+    },
+    {
+      "epoch": 0.4376068376068376,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.713966037202277e-05,
+      "loss": 2.0251,
+      "step": 256
+    },
+    {
+      "epoch": 0.4393162393162393,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.711622322915626e-05,
+      "loss": 2.1224,
+      "step": 257
+    },
+    {
+      "epoch": 0.441025641025641,
+      "grad_norm": 0.6015625,
+      "learning_rate": 2.7092701262050017e-05,
+      "loss": 2.1081,
+      "step": 258
+    },
+    {
+      "epoch": 0.4427350427350427,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.706909464624459e-05,
+      "loss": 2.2308,
+      "step": 259
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.7045403557912232e-05,
+      "loss": 2.1526,
+      "step": 260
+    },
+    {
+      "epoch": 0.4461538461538462,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.7021628173855615e-05,
+      "loss": 2.0584,
+      "step": 261
+    },
+    {
+      "epoch": 0.4478632478632479,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.699776867150648e-05,
+      "loss": 2.2188,
+      "step": 262
+    },
+    {
+      "epoch": 0.4495726495726496,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.697382522892435e-05,
+      "loss": 2.2346,
+      "step": 263
+    },
+    {
+      "epoch": 0.4512820512820513,
+      "grad_norm": 0.625,
+      "learning_rate": 2.6949798024795157e-05,
+      "loss": 2.1916,
+      "step": 264
+    },
+    {
+      "epoch": 0.452991452991453,
+      "grad_norm": 3.375,
+      "learning_rate": 2.6925687238429946e-05,
+      "loss": 2.068,
+      "step": 265
+    },
+    {
+      "epoch": 0.4547008547008547,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.690149304976352e-05,
+      "loss": 2.3094,
+      "step": 266
+    },
+    {
+      "epoch": 0.4564102564102564,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.68772156393531e-05,
+      "loss": 2.138,
+      "step": 267
+    },
+    {
+      "epoch": 0.4581196581196581,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.685285518837697e-05,
+      "loss": 2.0336,
+      "step": 268
+    },
+    {
+      "epoch": 0.4598290598290598,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.6828411878633143e-05,
+      "loss": 2.1164,
+      "step": 269
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.6803885892537985e-05,
+      "loss": 2.1127,
+      "step": 270
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "eval_loss": 2.1717300415039062,
+      "eval_runtime": 153.5734,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 270
+    },
+    {
+      "epoch": 0.46324786324786327,
+      "grad_norm": 0.8984375,
+      "learning_rate": 2.677927741312486e-05,
+      "loss": 2.1996,
+      "step": 271
+    },
+    {
+      "epoch": 0.46495726495726497,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.6754586624042777e-05,
+      "loss": 2.0878,
+      "step": 272
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.6729813709555e-05,
+      "loss": 2.165,
+      "step": 273
+    },
+    {
+      "epoch": 0.4683760683760684,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.6704958854537677e-05,
+      "loss": 2.0755,
+      "step": 274
+    },
+    {
+      "epoch": 0.4700854700854701,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.6680022244478478e-05,
+      "loss": 2.1522,
+      "step": 275
+    },
+    {
+      "epoch": 0.4717948717948718,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.6655004065475192e-05,
+      "loss": 2.0565,
+      "step": 276
+    },
+    {
+      "epoch": 0.4735042735042735,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.6629904504234338e-05,
+      "loss": 2.0979,
+      "step": 277
+    },
+    {
+      "epoch": 0.4752136752136752,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.6604723748069774e-05,
+      "loss": 2.056,
+      "step": 278
+    },
+    {
+      "epoch": 0.47692307692307695,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.657946198490132e-05,
+      "loss": 2.087,
+      "step": 279
+    },
+    {
+      "epoch": 0.47863247863247865,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.655411940325332e-05,
+      "loss": 2.13,
+      "step": 280
+    },
+    {
+      "epoch": 0.48034188034188036,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.652869619225326e-05,
+      "loss": 2.1566,
+      "step": 281
+    },
+    {
+      "epoch": 0.48205128205128206,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.6503192541630355e-05,
+      "loss": 2.1701,
+      "step": 282
+    },
+    {
+      "epoch": 0.48376068376068376,
+      "grad_norm": 0.93359375,
+      "learning_rate": 2.6477608641714113e-05,
+      "loss": 2.0952,
+      "step": 283
+    },
+    {
+      "epoch": 0.48547008547008547,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.645194468343295e-05,
+      "loss": 1.9035,
+      "step": 284
+    },
+    {
+      "epoch": 0.48717948717948717,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.6426200858312717e-05,
+      "loss": 2.0392,
+      "step": 285
+    },
+    {
+      "epoch": 0.4888888888888889,
+      "grad_norm": 0.74609375,
+      "learning_rate": 2.6400377358475323e-05,
+      "loss": 2.0303,
+      "step": 286
+    },
+    {
+      "epoch": 0.4905982905982906,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.6374474376637264e-05,
+      "loss": 2.0651,
+      "step": 287
+    },
+    {
+      "epoch": 0.49230769230769234,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.63484921061082e-05,
+      "loss": 2.0559,
+      "step": 288
+    },
+    {
+      "epoch": 0.49401709401709404,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.6322430740789513e-05,
+      "loss": 2.0077,
+      "step": 289
+    },
+    {
+      "epoch": 0.49572649572649574,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.629629047517284e-05,
+      "loss": 2.1017,
+      "step": 290
+    },
+    {
+      "epoch": 0.49743589743589745,
+      "grad_norm": 0.78515625,
+      "learning_rate": 2.6270071504338662e-05,
+      "loss": 2.1471,
+      "step": 291
+    },
+    {
+      "epoch": 0.49914529914529915,
+      "grad_norm": 0.625,
+      "learning_rate": 2.6243774023954807e-05,
+      "loss": 2.0882,
+      "step": 292
+    },
+    {
+      "epoch": 0.5008547008547009,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.6217398230275006e-05,
+      "loss": 2.162,
+      "step": 293
+    },
+    {
+      "epoch": 0.5025641025641026,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.6190944320137448e-05,
+      "loss": 2.0303,
+      "step": 294
+    },
+    {
+      "epoch": 0.5042735042735043,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.616441249096327e-05,
+      "loss": 2.0479,
+      "step": 295
+    },
+    {
+      "epoch": 0.505982905982906,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.6137802940755115e-05,
+      "loss": 2.1011,
+      "step": 296
+    },
+    {
+      "epoch": 0.5076923076923077,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.6111115868095647e-05,
+      "loss": 2.1615,
+      "step": 297
+    },
+    {
+      "epoch": 0.5094017094017094,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.6084351472146068e-05,
+      "loss": 2.1579,
+      "step": 298
+    },
+    {
+      "epoch": 0.5111111111111111,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.605750995264462e-05,
+      "loss": 2.141,
+      "step": 299
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.6030591509905123e-05,
+      "loss": 2.1742,
+      "step": 300
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "eval_loss": 2.1740505695343018,
+      "eval_runtime": 153.6223,
+      "eval_samples_per_second": 1.666,
+      "eval_steps_per_second": 0.28,
+      "step": 300
+    },
+    {
+      "epoch": 0.5145299145299145,
+      "grad_norm": 0.6953125,
+      "learning_rate": 2.6003596344815443e-05,
+      "loss": 2.2403,
+      "step": 301
+    },
+    {
+      "epoch": 0.5162393162393163,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.5976524658836026e-05,
+      "loss": 2.182,
+      "step": 302
+    },
+    {
+      "epoch": 0.517948717948718,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.5949376653998384e-05,
+      "loss": 2.0574,
+      "step": 303
+    },
+    {
+      "epoch": 0.5196581196581197,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.5922152532903568e-05,
+      "loss": 2.0676,
+      "step": 304
+    },
+    {
+      "epoch": 0.5213675213675214,
+      "grad_norm": 0.625,
+      "learning_rate": 2.589485249872068e-05,
+      "loss": 2.0652,
+      "step": 305
+    },
+    {
+      "epoch": 0.5230769230769231,
+      "grad_norm": 0.640625,
+      "learning_rate": 2.5867476755185357e-05,
+      "loss": 2.116,
+      "step": 306
+    },
+    {
+      "epoch": 0.5247863247863248,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.584002550659823e-05,
+      "loss": 2.1012,
+      "step": 307
+    },
+    {
+      "epoch": 0.5264957264957265,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.5812498957823415e-05,
+      "loss": 2.0946,
+      "step": 308
+    },
+    {
+      "epoch": 0.5282051282051282,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.578489731428698e-05,
+      "loss": 2.1419,
+      "step": 309
+    },
+    {
+      "epoch": 0.5299145299145299,
+      "grad_norm": 0.6796875,
+      "learning_rate": 2.5757220781975422e-05,
+      "loss": 2.0806,
+      "step": 310
+    },
+    {
+      "epoch": 0.5316239316239316,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.572946956743411e-05,
+      "loss": 2.0881,
+      "step": 311
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.570164387776576e-05,
+      "loss": 2.1517,
+      "step": 312
+    },
+    {
+      "epoch": 0.535042735042735,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.5673743920628884e-05,
+      "loss": 1.9632,
+      "step": 313
+    },
+    {
+      "epoch": 0.5367521367521367,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.564576990423623e-05,
+      "loss": 2.085,
+      "step": 314
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.5617722037353246e-05,
+      "loss": 2.1078,
+      "step": 315
+    },
+    {
+      "epoch": 0.5401709401709401,
+      "grad_norm": 0.75,
+      "learning_rate": 2.558960052929651e-05,
+      "loss": 2.152,
+      "step": 316
+    },
+    {
+      "epoch": 0.5418803418803418,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.5561405589932182e-05,
+      "loss": 1.9695,
+      "step": 317
+    },
+    {
+      "epoch": 0.5435897435897435,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.5533137429674416e-05,
+      "loss": 2.0379,
+      "step": 318
+    },
+    {
+      "epoch": 0.5452991452991452,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.5504796259483816e-05,
+      "loss": 2.0513,
+      "step": 319
+    },
+    {
+      "epoch": 0.5470085470085471,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.547638229086582e-05,
+      "loss": 2.2728,
+      "step": 320
+    },
+    {
+      "epoch": 0.5487179487179488,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.5447895735869184e-05,
+      "loss": 2.2484,
+      "step": 321
+    },
+    {
+      "epoch": 0.5504273504273505,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.541933680708432e-05,
+      "loss": 2.0199,
+      "step": 322
+    },
+    {
+      "epoch": 0.5521367521367522,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.5390705717641808e-05,
+      "loss": 1.9719,
+      "step": 323
+    },
+    {
+      "epoch": 0.5538461538461539,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.53620026812107e-05,
+      "loss": 2.0675,
+      "step": 324
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.5333227911997014e-05,
+      "loss": 2.1001,
+      "step": 325
+    },
+    {
+      "epoch": 0.5572649572649573,
+      "grad_norm": 0.76171875,
+      "learning_rate": 2.5304381624742075e-05,
+      "loss": 2.0147,
+      "step": 326
+    },
+    {
+      "epoch": 0.558974358974359,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.5275464034720945e-05,
+      "loss": 1.9013,
+      "step": 327
+    },
+    {
+      "epoch": 0.5606837606837607,
+      "grad_norm": 0.6328125,
+      "learning_rate": 2.52464753577408e-05,
+      "loss": 2.1728,
+      "step": 328
+    },
+    {
+      "epoch": 0.5623931623931624,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.5217415810139334e-05,
+      "loss": 2.1316,
+      "step": 329
+    },
+    {
+      "epoch": 0.5641025641025641,
+      "grad_norm": 0.8671875,
+      "learning_rate": 2.5188285608783135e-05,
+      "loss": 2.1845,
+      "step": 330
+    },
+    {
+      "epoch": 0.5641025641025641,
+      "eval_loss": 2.1763150691986084,
+      "eval_runtime": 153.6214,
+      "eval_samples_per_second": 1.666,
+      "eval_steps_per_second": 0.28,
+      "step": 330
+    },
+    {
+      "epoch": 0.5658119658119658,
+      "grad_norm": 0.66015625,
+      "learning_rate": 2.5159084971066066e-05,
+      "loss": 2.0914,
+      "step": 331
+    },
+    {
+      "epoch": 0.5675213675213675,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.5129814114907645e-05,
+      "loss": 2.3016,
+      "step": 332
+    },
+    {
+      "epoch": 0.5692307692307692,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.5100473258751415e-05,
+      "loss": 2.1094,
+      "step": 333
+    },
+    {
+      "epoch": 0.5709401709401709,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.507106262156333e-05,
+      "loss": 2.119,
+      "step": 334
+    },
+    {
+      "epoch": 0.5726495726495726,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.5041582422830094e-05,
+      "loss": 2.0574,
+      "step": 335
+    },
+    {
+      "epoch": 0.5743589743589743,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.5012032882557544e-05,
+      "loss": 2.1085,
+      "step": 336
+    },
+    {
+      "epoch": 0.576068376068376,
+      "grad_norm": 1.1328125,
+      "learning_rate": 2.4982414221269002e-05,
+      "loss": 2.2168,
+      "step": 337
+    },
+    {
+      "epoch": 0.5777777777777777,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.4952726660003624e-05,
+      "loss": 2.0249,
+      "step": 338
+    },
+    {
+      "epoch": 0.5794871794871795,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.4922970420314763e-05,
+      "loss": 2.1327,
+      "step": 339
+    },
+    {
+      "epoch": 0.5811965811965812,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.4893145724268295e-05,
+      "loss": 2.143,
+      "step": 340
+    },
+    {
+      "epoch": 0.582905982905983,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.4863252794440995e-05,
+      "loss": 1.9559,
+      "step": 341
+    },
+    {
+      "epoch": 0.5846153846153846,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.483329185391883e-05,
+      "loss": 2.1754,
+      "step": 342
+    },
+    {
+      "epoch": 0.5863247863247864,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.480326312629534e-05,
+      "loss": 2.1291,
+      "step": 343
+    },
+    {
+      "epoch": 0.588034188034188,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.4773166835669938e-05,
+      "loss": 2.0804,
+      "step": 344
+    },
+    {
+      "epoch": 0.5897435897435898,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.4743003206646256e-05,
+      "loss": 2.1426,
+      "step": 345
+    },
+    {
+      "epoch": 0.5914529914529915,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.471277246433045e-05,
+      "loss": 2.0337,
+      "step": 346
+    },
+    {
+      "epoch": 0.5931623931623932,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.4682474834329545e-05,
+      "loss": 2.0721,
+      "step": 347
+    },
+    {
+      "epoch": 0.5948717948717949,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.4652110542749725e-05,
+      "loss": 1.9929,
+      "step": 348
+    },
+    {
+      "epoch": 0.5965811965811966,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.462167981619467e-05,
+      "loss": 1.9594,
+      "step": 349
+    },
+    {
+      "epoch": 0.5982905982905983,
+      "grad_norm": 0.7578125,
+      "learning_rate": 2.4591182881763846e-05,
+      "loss": 2.0625,
+      "step": 350
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.456061996705082e-05,
+      "loss": 2.2578,
+      "step": 351
+    },
+    {
+      "epoch": 0.6017094017094017,
+      "grad_norm": 0.75390625,
+      "learning_rate": 2.4529991300141557e-05,
+      "loss": 2.0662,
+      "step": 352
+    },
+    {
+      "epoch": 0.6034188034188034,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.4499297109612714e-05,
+      "loss": 2.0684,
+      "step": 353
+    },
+    {
+      "epoch": 0.6051282051282051,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.4468537624529954e-05,
+      "loss": 1.9513,
+      "step": 354
+    },
+    {
+      "epoch": 0.6068376068376068,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.443771307444622e-05,
+      "loss": 2.0398,
+      "step": 355
+    },
+    {
+      "epoch": 0.6085470085470085,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.440682368940001e-05,
+      "loss": 2.264,
+      "step": 356
+    },
+    {
+      "epoch": 0.6102564102564103,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.4375869699913682e-05,
+      "loss": 2.0756,
+      "step": 357
+    },
+    {
+      "epoch": 0.611965811965812,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.434485133699174e-05,
+      "loss": 2.1884,
+      "step": 358
+    },
+    {
+      "epoch": 0.6136752136752137,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.4313768832119076e-05,
+      "loss": 2.0069,
+      "step": 359
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.4282622417259272e-05,
+      "loss": 2.0416,
+      "step": 360
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "eval_loss": 2.1718642711639404,
+      "eval_runtime": 153.5557,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 360
+    },
+    {
+      "epoch": 0.6170940170940171,
+      "grad_norm": 0.8359375,
+      "learning_rate": 2.4251412324852862e-05,
+      "loss": 2.1642,
+      "step": 361
+    },
+    {
+      "epoch": 0.6188034188034188,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.422013878781559e-05,
+      "loss": 2.0486,
+      "step": 362
+    },
+    {
+      "epoch": 0.6205128205128205,
+      "grad_norm": 0.83203125,
+      "learning_rate": 2.4188802039536672e-05,
+      "loss": 2.0809,
+      "step": 363
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 0.6796875,
+      "learning_rate": 2.4157402313877073e-05,
+      "loss": 2.2191,
+      "step": 364
+    },
+    {
+      "epoch": 0.6239316239316239,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.4125939845167742e-05,
+      "loss": 2.1666,
+      "step": 365
+    },
+    {
+      "epoch": 0.6256410256410256,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.4094414868207865e-05,
+      "loss": 2.0827,
+      "step": 366
+    },
+    {
+      "epoch": 0.6273504273504273,
+      "grad_norm": 0.83203125,
+      "learning_rate": 2.4062827618263117e-05,
+      "loss": 2.1373,
+      "step": 367
+    },
+    {
+      "epoch": 0.629059829059829,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.403117833106391e-05,
+      "loss": 2.0808,
+      "step": 368
+    },
+    {
+      "epoch": 0.6307692307692307,
+      "grad_norm": 0.625,
+      "learning_rate": 2.3999467242803623e-05,
+      "loss": 2.1417,
+      "step": 369
+    },
+    {
+      "epoch": 0.6324786324786325,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.3967694590136858e-05,
+      "loss": 2.0922,
+      "step": 370
+    },
+    {
+      "epoch": 0.6341880341880342,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.3935860610177652e-05,
+      "loss": 1.9197,
+      "step": 371
+    },
+    {
+      "epoch": 0.6358974358974359,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.390396554049772e-05,
+      "loss": 2.1443,
+      "step": 372
+    },
+    {
+      "epoch": 0.6376068376068376,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.3872009619124675e-05,
+      "loss": 1.9295,
+      "step": 373
+    },
+    {
+      "epoch": 0.6393162393162393,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.3839993084540264e-05,
+      "loss": 2.1179,
+      "step": 374
+    },
+    {
+      "epoch": 0.6410256410256411,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.3807916175678587e-05,
+      "loss": 2.0328,
+      "step": 375
+    },
+    {
+      "epoch": 0.6427350427350428,
+      "grad_norm": 0.59765625,
+      "learning_rate": 2.377577913192428e-05,
+      "loss": 2.1486,
+      "step": 376
+    },
+    {
+      "epoch": 0.6444444444444445,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.3743582193110784e-05,
+      "loss": 2.133,
+      "step": 377
+    },
+    {
+      "epoch": 0.6461538461538462,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.371132559951852e-05,
+      "loss": 2.1737,
+      "step": 378
+    },
+    {
+      "epoch": 0.6478632478632479,
+      "grad_norm": 1.046875,
+      "learning_rate": 2.3679009591873084e-05,
+      "loss": 2.1135,
+      "step": 379
+    },
+    {
+      "epoch": 0.6495726495726496,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.3646634411343493e-05,
+      "loss": 1.9031,
+      "step": 380
+    },
+    {
+      "epoch": 0.6512820512820513,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.3614200299540352e-05,
+      "loss": 1.9672,
+      "step": 381
+    },
+    {
+      "epoch": 0.652991452991453,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.358170749851406e-05,
+      "loss": 2.1242,
+      "step": 382
+    },
+    {
+      "epoch": 0.6547008547008547,
+      "grad_norm": 0.75,
+      "learning_rate": 2.3549156250753006e-05,
+      "loss": 1.9981,
+      "step": 383
+    },
+    {
+      "epoch": 0.6564102564102564,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.351654679918175e-05,
+      "loss": 2.0963,
+      "step": 384
+    },
+    {
+      "epoch": 0.6581196581196581,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.348387938715923e-05,
+      "loss": 2.0915,
+      "step": 385
+    },
+    {
+      "epoch": 0.6598290598290598,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.3451154258476926e-05,
+      "loss": 2.0056,
+      "step": 386
+    },
+    {
+      "epoch": 0.6615384615384615,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.341837165735704e-05,
+      "loss": 2.0933,
+      "step": 387
+    },
+    {
+      "epoch": 0.6632478632478632,
+      "grad_norm": 0.6953125,
+      "learning_rate": 2.33855318284507e-05,
+      "loss": 2.1574,
+      "step": 388
+    },
+    {
+      "epoch": 0.6649572649572649,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.3352635016836108e-05,
+      "loss": 2.0333,
+      "step": 389
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.33196814680167e-05,
+      "loss": 2.051,
+      "step": 390
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "eval_loss": 2.173579692840576,
+      "eval_runtime": 153.5924,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 390
+    },
+    {
+      "epoch": 0.6683760683760683,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.3286671427919357e-05,
+      "loss": 2.0965,
+      "step": 391
+    },
+    {
+      "epoch": 0.67008547008547,
+      "grad_norm": 0.75,
+      "learning_rate": 2.3253605142892538e-05,
+      "loss": 2.0582,
+      "step": 392
+    },
+    {
+      "epoch": 0.6717948717948717,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.3220482859704445e-05,
+      "loss": 2.0485,
+      "step": 393
+    },
+    {
+      "epoch": 0.6735042735042736,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.3187304825541185e-05,
+      "loss": 2.0276,
+      "step": 394
+    },
+    {
+      "epoch": 0.6752136752136753,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.3154071288004927e-05,
+      "loss": 2.1515,
+      "step": 395
+    },
+    {
+      "epoch": 0.676923076923077,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.3120782495112057e-05,
+      "loss": 2.1662,
+      "step": 396
+    },
+    {
+      "epoch": 0.6786324786324787,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.3087438695291314e-05,
+      "loss": 2.2733,
+      "step": 397
+    },
+    {
+      "epoch": 0.6803418803418804,
+      "grad_norm": 0.65234375,
+      "learning_rate": 2.3054040137381947e-05,
+      "loss": 1.9993,
+      "step": 398
+    },
+    {
+      "epoch": 0.6820512820512821,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.3020587070631854e-05,
+      "loss": 2.0988,
+      "step": 399
+    },
+    {
+      "epoch": 0.6837606837606838,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.298707974469574e-05,
+      "loss": 2.18,
+      "step": 400
+    },
+    {
+      "epoch": 0.6854700854700855,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.2953518409633206e-05,
+      "loss": 2.2455,
+      "step": 401
+    },
+    {
+      "epoch": 0.6871794871794872,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.291990331590695e-05,
+      "loss": 2.2599,
+      "step": 402
+    },
+    {
+      "epoch": 0.6888888888888889,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.2886234714380826e-05,
+      "loss": 2.2356,
+      "step": 403
+    },
+    {
+      "epoch": 0.6905982905982906,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.285251285631804e-05,
+      "loss": 2.1153,
+      "step": 404
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.2818737993379224e-05,
+      "loss": 2.0149,
+      "step": 405
+    },
+    {
+      "epoch": 0.694017094017094,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.278491037762058e-05,
+      "loss": 2.1123,
+      "step": 406
+    },
+    {
+      "epoch": 0.6957264957264957,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.275103026149199e-05,
+      "loss": 2.0417,
+      "step": 407
+    },
+    {
+      "epoch": 0.6974358974358974,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.271709789783516e-05,
+      "loss": 2.1376,
+      "step": 408
+    },
+    {
+      "epoch": 0.6991452991452991,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.268311353988168e-05,
+      "loss": 1.9649,
+      "step": 409
+    },
+    {
+      "epoch": 0.7008547008547008,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.264907744125119e-05,
+      "loss": 2.0379,
+      "step": 410
+    },
+    {
+      "epoch": 0.7025641025641025,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.2614989855949442e-05,
+      "loss": 2.0158,
+      "step": 411
+    },
+    {
+      "epoch": 0.7042735042735043,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.2580851038366447e-05,
+      "loss": 2.196,
+      "step": 412
+    },
+    {
+      "epoch": 0.705982905982906,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.254666124327453e-05,
+      "loss": 2.1096,
+      "step": 413
+    },
+    {
+      "epoch": 0.7076923076923077,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.251242072582647e-05,
+      "loss": 2.1235,
+      "step": 414
+    },
+    {
+      "epoch": 0.7094017094017094,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.2478129741553568e-05,
+      "loss": 2.0278,
+      "step": 415
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.2443788546363757e-05,
+      "loss": 2.0743,
+      "step": 416
+    },
+    {
+      "epoch": 0.7128205128205128,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.240939739653968e-05,
+      "loss": 2.0366,
+      "step": 417
+    },
+    {
+      "epoch": 0.7145299145299145,
+      "grad_norm": 0.9375,
+      "learning_rate": 2.23749565487368e-05,
+      "loss": 2.2902,
+      "step": 418
+    },
+    {
+      "epoch": 0.7162393162393162,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.2340466259981443e-05,
+      "loss": 2.1473,
+      "step": 419
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 0.88671875,
+      "learning_rate": 2.2305926787668918e-05,
+      "loss": 1.9886,
+      "step": 420
+    },
+    {
+      "epoch": 0.717948717948718,
+      "eval_loss": 2.170067071914673,
+      "eval_runtime": 153.5839,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 420
+    },
+    {
+      "epoch": 0.7196581196581197,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.2271338389561588e-05,
+      "loss": 2.0913,
+      "step": 421
+    },
+    {
+      "epoch": 0.7213675213675214,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.223670132378692e-05,
+      "loss": 1.9863,
+      "step": 422
+    },
+    {
+      "epoch": 0.7230769230769231,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.2202015848835606e-05,
+      "loss": 2.0038,
+      "step": 423
+    },
+    {
+      "epoch": 0.7247863247863248,
+      "grad_norm": 0.7578125,
+      "learning_rate": 2.2167282223559596e-05,
+      "loss": 2.2693,
+      "step": 424
+    },
+    {
+      "epoch": 0.7264957264957265,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.213250070717017e-05,
+      "loss": 2.0616,
+      "step": 425
+    },
+    {
+      "epoch": 0.7282051282051282,
+      "grad_norm": 0.765625,
+      "learning_rate": 2.2097671559236018e-05,
+      "loss": 1.9759,
+      "step": 426
+    },
+    {
+      "epoch": 0.7299145299145299,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.2062795039681313e-05,
+      "loss": 1.9583,
+      "step": 427
+    },
+    {
+      "epoch": 0.7316239316239316,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.202787140878372e-05,
+      "loss": 2.1853,
+      "step": 428
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.1992900927172517e-05,
+      "loss": 2.073,
+      "step": 429
+    },
+    {
+      "epoch": 0.7350427350427351,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.1957883855826605e-05,
+      "loss": 2.0493,
+      "step": 430
+    },
+    {
+      "epoch": 0.7367521367521368,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.1922820456072592e-05,
+      "loss": 2.1227,
+      "step": 431
+    },
+    {
+      "epoch": 0.7384615384615385,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.1887710989582807e-05,
+      "loss": 2.0996,
+      "step": 432
+    },
+    {
+      "epoch": 0.7401709401709402,
+      "grad_norm": 0.6796875,
+      "learning_rate": 2.1852555718373388e-05,
+      "loss": 2.0341,
+      "step": 433
+    },
+    {
+      "epoch": 0.7418803418803419,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.1817354904802285e-05,
+      "loss": 2.0643,
+      "step": 434
+    },
+    {
+      "epoch": 0.7435897435897436,
+      "grad_norm": 0.7890625,
+      "learning_rate": 2.1782108811567347e-05,
+      "loss": 2.0514,
+      "step": 435
+    },
+    {
+      "epoch": 0.7452991452991453,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.1746817701704316e-05,
+      "loss": 2.0263,
+      "step": 436
+    },
+    {
+      "epoch": 0.747008547008547,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.1711481838584897e-05,
+      "loss": 2.1069,
+      "step": 437
+    },
+    {
+      "epoch": 0.7487179487179487,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.1676101485914772e-05,
+      "loss": 2.0526,
+      "step": 438
+    },
+    {
+      "epoch": 0.7504273504273504,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.1640676907731662e-05,
+      "loss": 1.9615,
+      "step": 439
+    },
+    {
+      "epoch": 0.7521367521367521,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.1605208368403316e-05,
+      "loss": 2.0418,
+      "step": 440
+    },
+    {
+      "epoch": 0.7538461538461538,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.1569696132625553e-05,
+      "loss": 2.0366,
+      "step": 441
+    },
+    {
+      "epoch": 0.7555555555555555,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.1534140465420312e-05,
+      "loss": 2.0726,
+      "step": 442
+    },
+    {
+      "epoch": 0.7572649572649572,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.1498541632133644e-05,
+      "loss": 2.0074,
+      "step": 443
+    },
+    {
+      "epoch": 0.7589743589743589,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.146289989843374e-05,
+      "loss": 2.113,
+      "step": 444
+    },
+    {
+      "epoch": 0.7606837606837606,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.1427215530308942e-05,
+      "loss": 1.9158,
+      "step": 445
+    },
+    {
+      "epoch": 0.7623931623931623,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.139148879406578e-05,
+      "loss": 2.2068,
+      "step": 446
+    },
+    {
+      "epoch": 0.764102564102564,
+      "grad_norm": 0.75,
+      "learning_rate": 2.1355719956326965e-05,
+      "loss": 2.1593,
+      "step": 447
+    },
+    {
+      "epoch": 0.7658119658119659,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.1319909284029402e-05,
+      "loss": 2.1209,
+      "step": 448
+    },
+    {
+      "epoch": 0.7675213675213676,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.1284057044422196e-05,
+      "loss": 2.0558,
+      "step": 449
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.124816350506468e-05,
+      "loss": 2.0581,
+      "step": 450
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "eval_loss": 2.170384645462036,
+      "eval_runtime": 153.6598,
+      "eval_samples_per_second": 1.666,
+      "eval_steps_per_second": 0.28,
+      "step": 450
+    },
+    {
+      "epoch": 0.770940170940171,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.1212228933824378e-05,
+      "loss": 2.2153,
+      "step": 451
+    },
+    {
+      "epoch": 0.7726495726495727,
+      "grad_norm": 0.8984375,
+      "learning_rate": 2.1176253598875042e-05,
+      "loss": 1.956,
+      "step": 452
+    },
+    {
+      "epoch": 0.7743589743589744,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.1140237768694633e-05,
+      "loss": 2.0069,
+      "step": 453
+    },
+    {
+      "epoch": 0.7760683760683761,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.1104181712063324e-05,
+      "loss": 2.003,
+      "step": 454
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.106808569806148e-05,
+      "loss": 2.0104,
+      "step": 455
+    },
+    {
+      "epoch": 0.7794871794871795,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.1031949996067674e-05,
+      "loss": 2.1266,
+      "step": 456
+    },
+    {
+      "epoch": 0.7811965811965812,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.0995774875756667e-05,
+      "loss": 2.0628,
+      "step": 457
+    },
+    {
+      "epoch": 0.7829059829059829,
+      "grad_norm": 0.76171875,
+      "learning_rate": 2.095956060709737e-05,
+      "loss": 1.9926,
+      "step": 458
+    },
+    {
+      "epoch": 0.7846153846153846,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.0923307460350876e-05,
+      "loss": 2.0338,
+      "step": 459
+    },
+    {
+      "epoch": 0.7863247863247863,
+      "grad_norm": 0.6953125,
+      "learning_rate": 2.088701570606841e-05,
+      "loss": 2.0658,
+      "step": 460
+    },
+    {
+      "epoch": 0.788034188034188,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.085068561508932e-05,
+      "loss": 2.0544,
+      "step": 461
+    },
+    {
+      "epoch": 0.7897435897435897,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.081431745853904e-05,
+      "loss": 2.0076,
+      "step": 462
+    },
+    {
+      "epoch": 0.7914529914529914,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.0777911507827107e-05,
+      "loss": 1.9997,
+      "step": 463
+    },
+    {
+      "epoch": 0.7931623931623931,
+      "grad_norm": 0.8359375,
+      "learning_rate": 2.074146803464509e-05,
+      "loss": 2.0927,
+      "step": 464
+    },
+    {
+      "epoch": 0.7948717948717948,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.0704987310964596e-05,
+      "loss": 1.9608,
+      "step": 465
+    },
+    {
+      "epoch": 0.7965811965811965,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.0668469609035197e-05,
+      "loss": 2.0182,
+      "step": 466
+    },
+    {
+      "epoch": 0.7982905982905983,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.0631915201382475e-05,
+      "loss": 2.048,
+      "step": 467
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.0595324360805896e-05,
+      "loss": 1.9037,
+      "step": 468
+    },
+    {
+      "epoch": 0.8017094017094017,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.0558697360376847e-05,
+      "loss": 2.2079,
+      "step": 469
+    },
+    {
+      "epoch": 0.8034188034188035,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.052203447343655e-05,
+      "loss": 2.0305,
+      "step": 470
+    },
+    {
+      "epoch": 0.8051282051282052,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.0485335973594067e-05,
+      "loss": 2.0819,
+      "step": 471
+    },
+    {
+      "epoch": 0.8068376068376069,
+      "grad_norm": 0.9140625,
+      "learning_rate": 2.044860213472421e-05,
+      "loss": 2.2356,
+      "step": 472
+    },
+    {
+      "epoch": 0.8085470085470086,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.041183323096553e-05,
+      "loss": 2.0693,
+      "step": 473
+    },
+    {
+      "epoch": 0.8102564102564103,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.0375029536718263e-05,
+      "loss": 1.9825,
+      "step": 474
+    },
+    {
+      "epoch": 0.811965811965812,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.033819132664227e-05,
+      "loss": 2.0774,
+      "step": 475
+    },
+    {
+      "epoch": 0.8136752136752137,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.0301318875655018e-05,
+      "loss": 2.2164,
+      "step": 476
+    },
+    {
+      "epoch": 0.8153846153846154,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.0264412458929483e-05,
+      "loss": 2.1419,
+      "step": 477
+    },
+    {
+      "epoch": 0.8170940170940171,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.0227472351892153e-05,
+      "loss": 2.1672,
+      "step": 478
+    },
+    {
+      "epoch": 0.8188034188034188,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.01904988302209e-05,
+      "loss": 1.917,
+      "step": 479
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.0153492169843002e-05,
+      "loss": 2.2503,
+      "step": 480
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "eval_loss": 2.170718193054199,
+      "eval_runtime": 153.5405,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 480
+    },
+    {
+      "epoch": 0.8222222222222222,
+      "grad_norm": 0.64453125,
+      "learning_rate": 2.011645264693303e-05,
+      "loss": 2.1902,
+      "step": 481
+    },
+    {
+      "epoch": 0.8239316239316239,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.00793805379108e-05,
+      "loss": 1.9729,
+      "step": 482
+    },
+    {
+      "epoch": 0.8256410256410256,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.0042276119439313e-05,
+      "loss": 2.1218,
+      "step": 483
+    },
+    {
+      "epoch": 0.8273504273504273,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.00051396684227e-05,
+      "loss": 2.0174,
+      "step": 484
+    },
+    {
+      "epoch": 0.8290598290598291,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.9967971462004136e-05,
+      "loss": 2.1274,
+      "step": 485
+    },
+    {
+      "epoch": 0.8307692307692308,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.993077177756378e-05,
+      "loss": 2.0213,
+      "step": 486
+    },
+    {
+      "epoch": 0.8324786324786325,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.989354089271672e-05,
+      "loss": 2.2674,
+      "step": 487
+    },
+    {
+      "epoch": 0.8341880341880342,
+      "grad_norm": 0.7421875,
+      "learning_rate": 1.9856279085310866e-05,
+      "loss": 2.054,
+      "step": 488
+    },
+    {
+      "epoch": 0.8358974358974359,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.9818986633424915e-05,
+      "loss": 2.2198,
+      "step": 489
+    },
+    {
+      "epoch": 0.8376068376068376,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.978166381536625e-05,
+      "loss": 2.1116,
+      "step": 490
+    },
+    {
+      "epoch": 0.8393162393162393,
+      "grad_norm": 0.71484375,
+      "learning_rate": 1.9744310909668883e-05,
+      "loss": 1.916,
+      "step": 491
+    },
+    {
+      "epoch": 0.841025641025641,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.970692819509136e-05,
+      "loss": 2.1808,
+      "step": 492
+    },
+    {
+      "epoch": 0.8427350427350427,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.966951595061466e-05,
+      "loss": 2.0524,
+      "step": 493
+    },
+    {
+      "epoch": 0.8444444444444444,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.9632074455440183e-05,
+      "loss": 2.2054,
+      "step": 494
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.9594603988987594e-05,
+      "loss": 2.2013,
+      "step": 495
+    },
+    {
+      "epoch": 0.8478632478632478,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.955710483089278e-05,
+      "loss": 2.0432,
+      "step": 496
+    },
+    {
+      "epoch": 0.8495726495726496,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.9519577261005724e-05,
+      "loss": 1.9487,
+      "step": 497
+    },
+    {
+      "epoch": 0.8512820512820513,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.948202155938847e-05,
+      "loss": 2.1093,
+      "step": 498
+    },
+    {
+      "epoch": 0.852991452991453,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.9444438006313e-05,
+      "loss": 2.0789,
+      "step": 499
+    },
+    {
+      "epoch": 0.8547008547008547,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.9406826882259126e-05,
+      "loss": 2.0101,
+      "step": 500
+    },
+    {
+      "epoch": 0.8564102564102564,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.9369188467912443e-05,
+      "loss": 2.1196,
+      "step": 501
+    },
+    {
+      "epoch": 0.8581196581196581,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.933152304416219e-05,
+      "loss": 2.0398,
+      "step": 502
+    },
+    {
+      "epoch": 0.8598290598290599,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.9293830892099183e-05,
+      "loss": 2.1621,
+      "step": 503
+    },
+    {
+      "epoch": 0.8615384615384616,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.9256112293013707e-05,
+      "loss": 1.8717,
+      "step": 504
+    },
+    {
+      "epoch": 0.8632478632478633,
+      "grad_norm": 0.7578125,
+      "learning_rate": 1.921836752839341e-05,
+      "loss": 1.9348,
+      "step": 505
+    },
+    {
+      "epoch": 0.864957264957265,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.9180596879921217e-05,
+      "loss": 2.0856,
+      "step": 506
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.9142800629473218e-05,
+      "loss": 1.8842,
+      "step": 507
+    },
+    {
+      "epoch": 0.8683760683760684,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.9104979059116557e-05,
+      "loss": 2.0198,
+      "step": 508
+    },
+    {
+      "epoch": 0.8700854700854701,
+      "grad_norm": 0.68359375,
+      "learning_rate": 1.9067132451107352e-05,
+      "loss": 2.0228,
+      "step": 509
+    },
+    {
+      "epoch": 0.8717948717948718,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.902926108788856e-05,
+      "loss": 2.1459,
+      "step": 510
+    },
+    {
+      "epoch": 0.8717948717948718,
+      "eval_loss": 2.1723549365997314,
+      "eval_runtime": 153.5352,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 510
+    },
+    {
+      "epoch": 0.8735042735042735,
+      "grad_norm": 0.65625,
+      "learning_rate": 1.899136525208788e-05,
+      "loss": 2.1702,
+      "step": 511
+    },
+    {
+      "epoch": 0.8752136752136752,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.8953445226515665e-05,
+      "loss": 2.0444,
+      "step": 512
+    },
+    {
+      "epoch": 0.8769230769230769,
+      "grad_norm": 0.7578125,
+      "learning_rate": 1.891550129416278e-05,
+      "loss": 2.0744,
+      "step": 513
+    },
+    {
+      "epoch": 0.8786324786324786,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.8877533738198487e-05,
+      "loss": 2.0105,
+      "step": 514
+    },
+    {
+      "epoch": 0.8803418803418803,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.8839542841968378e-05,
+      "loss": 2.0139,
+      "step": 515
+    },
+    {
+      "epoch": 0.882051282051282,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.88015288889922e-05,
+      "loss": 2.0756,
+      "step": 516
+    },
+    {
+      "epoch": 0.8837606837606837,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.8763492162961796e-05,
+      "loss": 1.9351,
+      "step": 517
+    },
+    {
+      "epoch": 0.8854700854700854,
+      "grad_norm": 0.7578125,
+      "learning_rate": 1.8725432947738937e-05,
+      "loss": 2.1328,
+      "step": 518
+    },
+    {
+      "epoch": 0.8871794871794871,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.868735152735324e-05,
+      "loss": 1.9942,
+      "step": 519
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.8649248186000035e-05,
+      "loss": 2.0258,
+      "step": 520
+    },
+    {
+      "epoch": 0.8905982905982905,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.8611123208038236e-05,
+      "loss": 2.112,
+      "step": 521
+    },
+    {
+      "epoch": 0.8923076923076924,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.8572976877988238e-05,
+      "loss": 1.9945,
+      "step": 522
+    },
+    {
+      "epoch": 0.8940170940170941,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.853480948052978e-05,
+      "loss": 2.0789,
+      "step": 523
+    },
+    {
+      "epoch": 0.8957264957264958,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.8496621300499818e-05,
+      "loss": 2.0239,
+      "step": 524
+    },
+    {
+      "epoch": 0.8974358974358975,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.8458412622890407e-05,
+      "loss": 2.0569,
+      "step": 525
+    },
+    {
+      "epoch": 0.8991452991452992,
+      "grad_norm": 0.76171875,
+      "learning_rate": 1.842018373284658e-05,
+      "loss": 1.9994,
+      "step": 526
+    },
+    {
+      "epoch": 0.9008547008547009,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.83819349156642e-05,
+      "loss": 2.0298,
+      "step": 527
+    },
+    {
+      "epoch": 0.9025641025641026,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.8343666456787856e-05,
+      "loss": 1.8457,
+      "step": 528
+    },
+    {
+      "epoch": 0.9042735042735043,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.8305378641808705e-05,
+      "loss": 1.981,
+      "step": 529
+    },
+    {
+      "epoch": 0.905982905982906,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.8267071756462375e-05,
+      "loss": 2.0879,
+      "step": 530
+    },
+    {
+      "epoch": 0.9076923076923077,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.8228746086626784e-05,
+      "loss": 1.9026,
+      "step": 531
+    },
+    {
+      "epoch": 0.9094017094017094,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.8190401918320064e-05,
+      "loss": 2.1244,
+      "step": 532
+    },
+    {
+      "epoch": 0.9111111111111111,
+      "grad_norm": 0.65625,
+      "learning_rate": 1.8152039537698384e-05,
+      "loss": 2.1991,
+      "step": 533
+    },
+    {
+      "epoch": 0.9128205128205128,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.811365923105384e-05,
+      "loss": 1.9811,
+      "step": 534
+    },
+    {
+      "epoch": 0.9145299145299145,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.8075261284812277e-05,
+      "loss": 1.9196,
+      "step": 535
+    },
+    {
+      "epoch": 0.9162393162393162,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.803684598553122e-05,
+      "loss": 2.0669,
+      "step": 536
+    },
+    {
+      "epoch": 0.9179487179487179,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.7998413619897677e-05,
+      "loss": 2.078,
+      "step": 537
+    },
+    {
+      "epoch": 0.9196581196581196,
+      "grad_norm": 0.875,
+      "learning_rate": 1.795996447472601e-05,
+      "loss": 1.9816,
+      "step": 538
+    },
+    {
+      "epoch": 0.9213675213675213,
+      "grad_norm": 0.7421875,
+      "learning_rate": 1.792149883695583e-05,
+      "loss": 1.9986,
+      "step": 539
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.7883016993649805e-05,
+      "loss": 2.0384,
+      "step": 540
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "eval_loss": 2.172126054763794,
+      "eval_runtime": 153.5246,
+      "eval_samples_per_second": 1.667,
+      "eval_steps_per_second": 0.28,
+      "step": 540
+    },
+    {
+      "epoch": 0.9247863247863248,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.7844519231991567e-05,
+      "loss": 1.8969,
+      "step": 541
+    },
+    {
+      "epoch": 0.9264957264957265,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.7806005839283506e-05,
+      "loss": 1.9619,
+      "step": 542
+    },
+    {
+      "epoch": 0.9282051282051282,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.7767477102944708e-05,
+      "loss": 1.8984,
+      "step": 543
+    },
+    {
+      "epoch": 0.9299145299145299,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.7728933310508732e-05,
+      "loss": 2.0389,
+      "step": 544
+    },
+    {
+      "epoch": 0.9316239316239316,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.7690374749621517e-05,
+      "loss": 1.8232,
+      "step": 545
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.7651801708039207e-05,
+      "loss": 2.156,
+      "step": 546
+    },
+    {
+      "epoch": 0.935042735042735,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.761321447362603e-05,
+      "loss": 2.2417,
+      "step": 547
+    },
+    {
+      "epoch": 0.9367521367521368,
+      "grad_norm": 0.76171875,
+      "learning_rate": 1.7574613334352102e-05,
+      "loss": 1.9024,
+      "step": 548
+    },
+    {
+      "epoch": 0.9384615384615385,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.7535998578291335e-05,
+      "loss": 2.025,
+      "step": 549
+    },
+    {
+      "epoch": 0.9401709401709402,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.7497370493619252e-05,
+      "loss": 2.1301,
+      "step": 550
+    },
+    {
+      "epoch": 0.9418803418803419,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.745872936861085e-05,
+      "loss": 2.2419,
+      "step": 551
+    },
+    {
+      "epoch": 0.9435897435897436,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.742007549163843e-05,
+      "loss": 2.0159,
+      "step": 552
+    },
+    {
+      "epoch": 0.9452991452991453,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.7381409151169473e-05,
+      "loss": 2.0406,
+      "step": 553
+    },
+    {
+      "epoch": 0.947008547008547,
+      "grad_norm": 0.75,
+      "learning_rate": 1.7342730635764463e-05,
+      "loss": 2.001,
+      "step": 554
+    },
+    {
+      "epoch": 0.9487179487179487,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.7304040234074755e-05,
+      "loss": 2.1059,
+      "step": 555
+    },
+    {
+      "epoch": 0.9504273504273504,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.72653382348404e-05,
+      "loss": 2.0781,
+      "step": 556
+    },
+    {
+      "epoch": 0.9521367521367521,
+      "grad_norm": 0.71484375,
+      "learning_rate": 1.7226624926888006e-05,
+      "loss": 2.0043,
+      "step": 557
+    },
+    {
+      "epoch": 0.9538461538461539,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.7187900599128564e-05,
+      "loss": 2.0611,
+      "step": 558
+    },
+    {
+      "epoch": 0.9555555555555556,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.714916554055532e-05,
+      "loss": 2.2539,
+      "step": 559
+    },
+    {
+      "epoch": 0.9572649572649573,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.7110420040241596e-05,
+      "loss": 2.1713,
+      "step": 560
+    },
+    {
+      "epoch": 0.958974358974359,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.7071664387338632e-05,
+      "loss": 1.9576,
+      "step": 561
+    },
+    {
+      "epoch": 0.9606837606837607,
+      "grad_norm": 0.75,
+      "learning_rate": 1.7032898871073445e-05,
+      "loss": 2.0423,
+      "step": 562
+    },
+    {
+      "epoch": 0.9623931623931624,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.699412378074665e-05,
+      "loss": 2.0474,
+      "step": 563
+    },
+    {
+      "epoch": 0.9641025641025641,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.6955339405730334e-05,
+      "loss": 2.0956,
+      "step": 564
+    },
+    {
+      "epoch": 0.9658119658119658,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.6916546035465838e-05,
+      "loss": 2.0223,
+      "step": 565
+    },
+    {
+      "epoch": 0.9675213675213675,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.6877743959461667e-05,
+      "loss": 2.052,
+      "step": 566
+    },
+    {
+      "epoch": 0.9692307692307692,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.6838933467291275e-05,
+      "loss": 2.1276,
+      "step": 567
+    },
+    {
+      "epoch": 0.9709401709401709,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.680011484859094e-05,
+      "loss": 1.9882,
+      "step": 568
+    },
+    {
+      "epoch": 0.9726495726495726,
+      "grad_norm": 0.7578125,
+      "learning_rate": 1.6761288393057567e-05,
+      "loss": 2.2599,
+      "step": 569
+    },
+    {
+      "epoch": 0.9743589743589743,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.6722454390446565e-05,
+      "loss": 2.0889,
+      "step": 570
+    },
+    {
+      "epoch": 0.9743589743589743,
+      "eval_loss": 2.1726019382476807,
+      "eval_runtime": 153.6502,
+      "eval_samples_per_second": 1.666,
+      "eval_steps_per_second": 0.28,
+      "step": 570
+    },
+    {
+      "epoch": 0.976068376068376,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.668361313056965e-05,
+      "loss": 2.3068,
+      "step": 571
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.6644764903292718e-05,
+      "loss": 1.9533,
+      "step": 572
+    },
+    {
+      "epoch": 0.9794871794871794,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.6605909998533637e-05,
+      "loss": 1.9942,
+      "step": 573
+    },
+    {
+      "epoch": 0.9811965811965812,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.656704870626013e-05,
+      "loss": 2.0547,
+      "step": 574
+    },
+    {
+      "epoch": 0.9829059829059829,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.652818131648757e-05,
+      "loss": 1.9273,
+      "step": 575
+    },
+    {
+      "epoch": 0.9846153846153847,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.648930811927685e-05,
+      "loss": 2.0566,
+      "step": 576
+    },
+    {
+      "epoch": 0.9863247863247864,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.645042940473219e-05,
+      "loss": 2.1004,
+      "step": 577
+    },
+    {
+      "epoch": 0.9880341880341881,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.6411545462999e-05,
+      "loss": 2.1112,
+      "step": 578
+    },
+    {
+      "epoch": 0.9897435897435898,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.6372656584261687e-05,
+      "loss": 1.9699,
+      "step": 579
+    },
+    {
+      "epoch": 0.9914529914529915,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.6333763058741503e-05,
+      "loss": 2.0101,
+      "step": 580
+    },
+    {
+      "epoch": 0.9931623931623932,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.6294865176694393e-05,
+      "loss": 2.0589,
+      "step": 581
+    },
+    {
+      "epoch": 0.9948717948717949,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.6255963228408784e-05,
+      "loss": 2.0307,
+      "step": 582
+    },
+    {
+      "epoch": 0.9965811965811966,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.6217057504203477e-05,
+      "loss": 1.9615,
+      "step": 583
+    },
+    {
+      "epoch": 0.9982905982905983,
+      "grad_norm": 0.765625,
+      "learning_rate": 1.617814829442545e-05,
+      "loss": 2.0134,
+      "step": 584
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.6139235889447677e-05,
+      "loss": 1.9138,
+      "step": 585
+    },
+    {
+      "epoch": 1.0017094017094017,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.610032057966698e-05,
+      "loss": 1.9708,
+      "step": 586
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1170,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 293,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.1009785193037824e+17,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}