diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5258 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9987398134923968,
+  "eval_steps": 248,
+  "global_step": 743,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0013441989414433337,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.0378,
+      "step": 1
+    },
+    {
+      "epoch": 0.0013441989414433337,
+      "eval_loss": 3.0436718463897705,
+      "eval_runtime": 5734.1832,
+      "eval_samples_per_second": 3.861,
+      "eval_steps_per_second": 0.483,
+      "step": 1
+    },
+    {
+      "epoch": 0.0026883978828866673,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.1143,
+      "step": 2
+    },
+    {
+      "epoch": 0.004032596824330001,
+      "grad_norm": 8.75,
+      "learning_rate": 3e-06,
+      "loss": 1.0933,
+      "step": 3
+    },
+    {
+      "epoch": 0.005376795765773335,
+      "grad_norm": 6.40625,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1014,
+      "step": 4
+    },
+    {
+      "epoch": 0.006720994707216668,
+      "grad_norm": 5.34375,
+      "learning_rate": 5e-06,
+      "loss": 1.0504,
+      "step": 5
+    },
+    {
+      "epoch": 0.008065193648660002,
+      "grad_norm": 3.453125,
+      "learning_rate": 6e-06,
+      "loss": 1.0031,
+      "step": 6
+    },
+    {
+      "epoch": 0.009409392590103335,
+      "grad_norm": 3.0,
+      "learning_rate": 7e-06,
+      "loss": 1.0001,
+      "step": 7
+    },
+    {
+      "epoch": 0.01075359153154667,
+      "grad_norm": 2.640625,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0196,
+      "step": 8
+    },
+    {
+      "epoch": 0.012097790472990002,
+      "grad_norm": 2.21875,
+      "learning_rate": 9e-06,
+      "loss": 0.9207,
+      "step": 9
+    },
+    {
+      "epoch": 0.013441989414433336,
+      "grad_norm": 1.859375,
+      "learning_rate": 1e-05,
+      "loss": 0.9502,
+      "step": 10
+    },
+    {
+      "epoch": 0.01478618835587667,
+      "grad_norm": 2.34375,
+      "learning_rate": 9.999954076906038e-06,
+      "loss": 0.9674,
+      "step": 11
+    },
+    {
+      "epoch": 0.016130387297320005,
+      "grad_norm": 1.6953125,
+      "learning_rate": 9.99981630846772e-06,
+      "loss": 0.8617,
+      "step": 12
+    },
+    {
+      "epoch": 0.017474586238763336,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.999586697215748e-06,
+      "loss": 0.9222,
+      "step": 13
+    },
+    {
+      "epoch": 0.01881878518020667,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.999265247367909e-06,
+      "loss": 0.884,
+      "step": 14
+    },
+    {
+      "epoch": 0.020162984121650004,
+      "grad_norm": 1.7578125,
+      "learning_rate": 9.998851964828987e-06,
+      "loss": 0.9086,
+      "step": 15
+    },
+    {
+      "epoch": 0.02150718306309334,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.99834685719067e-06,
+      "loss": 0.8723,
+      "step": 16
+    },
+    {
+      "epoch": 0.022851382004536673,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.997749933731397e-06,
+      "loss": 0.8678,
+      "step": 17
+    },
+    {
+      "epoch": 0.024195580945980004,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.997061205416203e-06,
+      "loss": 0.8497,
+      "step": 18
+    },
+    {
+      "epoch": 0.025539779887423338,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.996280684896496e-06,
+      "loss": 0.8475,
+      "step": 19
+    },
+    {
+      "epoch": 0.026883978828866673,
+      "grad_norm": 1.625,
+      "learning_rate": 9.995408386509846e-06,
+      "loss": 0.8673,
+      "step": 20
+    },
+    {
+      "epoch": 0.028228177770310007,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.99444432627971e-06,
+      "loss": 0.8704,
+      "step": 21
+    },
+    {
+      "epoch": 0.02957237671175334,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.993388521915134e-06,
+      "loss": 0.813,
+      "step": 22
+    },
+    {
+      "epoch": 0.030916575653196672,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.992240992810445e-06,
+      "loss": 0.8497,
+      "step": 23
+    },
+    {
+      "epoch": 0.03226077459464001,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.991001760044877e-06,
+      "loss": 0.8027,
+      "step": 24
+    },
+    {
+      "epoch": 0.03360497353608334,
+      "grad_norm": 1.375,
+      "learning_rate": 9.989670846382189e-06,
+      "loss": 0.8099,
+      "step": 25
+    },
+    {
+      "epoch": 0.03494917247752667,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.98824827627025e-06,
+      "loss": 0.7818,
+      "step": 26
+    },
+    {
+      "epoch": 0.03629337141897001,
+      "grad_norm": 1.5,
+      "learning_rate": 9.986734075840591e-06,
+      "loss": 0.8078,
+      "step": 27
+    },
+    {
+      "epoch": 0.03763757036041334,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.985128272907917e-06,
+      "loss": 0.819,
+      "step": 28
+    },
+    {
+      "epoch": 0.03898176930185668,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.983430896969606e-06,
+      "loss": 0.7949,
+      "step": 29
+    },
+    {
+      "epoch": 0.04032596824330001,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.981641979205158e-06,
+      "loss": 0.8044,
+      "step": 30
+    },
+    {
+      "epoch": 0.04167016718474334,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.97976155247563e-06,
+      "loss": 0.8025,
+      "step": 31
+    },
+    {
+      "epoch": 0.04301436612618668,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.977789651323025e-06,
+      "loss": 0.7461,
+      "step": 32
+    },
+    {
+      "epoch": 0.04435856506763001,
+      "grad_norm": 1.40625,
+      "learning_rate": 9.975726311969664e-06,
+      "loss": 0.787,
+      "step": 33
+    },
+    {
+      "epoch": 0.045702764009073346,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.973571572317519e-06,
+      "loss": 0.7837,
+      "step": 34
+    },
+    {
+      "epoch": 0.04704696295051668,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.971325471947518e-06,
+      "loss": 0.8101,
+      "step": 35
+    },
+    {
+      "epoch": 0.04839116189196001,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.968988052118804e-06,
+      "loss": 0.783,
+      "step": 36
+    },
+    {
+      "epoch": 0.049735360833403346,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.966559355768005e-06,
+      "loss": 0.8141,
+      "step": 37
+    },
+    {
+      "epoch": 0.051079559774846676,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.964039427508418e-06,
+      "loss": 0.7753,
+      "step": 38
+    },
+    {
+      "epoch": 0.052423758716290014,
+      "grad_norm": 1.375,
+      "learning_rate": 9.961428313629203e-06,
+      "loss": 0.773,
+      "step": 39
+    },
+    {
+      "epoch": 0.053767957657733345,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.958726062094533e-06,
+      "loss": 0.7557,
+      "step": 40
+    },
+    {
+      "epoch": 0.055112156599176676,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.955932722542709e-06,
+      "loss": 0.7241,
+      "step": 41
+    },
+    {
+      "epoch": 0.056456355540620014,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.953048346285245e-06,
+      "loss": 0.781,
+      "step": 42
+    },
+    {
+      "epoch": 0.057800554482063345,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.950072986305938e-06,
+      "loss": 0.7644,
+      "step": 43
+    },
+    {
+      "epoch": 0.05914475342350668,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.947006697259881e-06,
+      "loss": 0.7406,
+      "step": 44
+    },
+    {
+      "epoch": 0.06048895236495001,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.943849535472468e-06,
+      "loss": 0.762,
+      "step": 45
+    },
+    {
+      "epoch": 0.061833151306393344,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.940601558938348e-06,
+      "loss": 0.7448,
+      "step": 46
+    },
+    {
+      "epoch": 0.06317735024783667,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.93726282732038e-06,
+      "loss": 0.7658,
+      "step": 47
+    },
+    {
+      "epoch": 0.06452154918928002,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.933833401948514e-06,
+      "loss": 0.7876,
+      "step": 48
+    },
+    {
+      "epoch": 0.06586574813072335,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.930313345818683e-06,
+      "loss": 0.8166,
+      "step": 49
+    },
+    {
+      "epoch": 0.06720994707216668,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.92670272359163e-06,
+      "loss": 0.7927,
+      "step": 50
+    },
+    {
+      "epoch": 0.06855414601361001,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.923001601591738e-06,
+      "loss": 0.7906,
+      "step": 51
+    },
+    {
+      "epoch": 0.06989834495505334,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.919210047805791e-06,
+      "loss": 0.7576,
+      "step": 52
+    },
+    {
+      "epoch": 0.07124254389649669,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.915328131881745e-06,
+      "loss": 0.7847,
+      "step": 53
+    },
+    {
+      "epoch": 0.07258674283794002,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.911355925127433e-06,
+      "loss": 0.7728,
+      "step": 54
+    },
+    {
+      "epoch": 0.07393094177938335,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.907293500509268e-06,
+      "loss": 0.7947,
+      "step": 55
+    },
+    {
+      "epoch": 0.07527514072082668,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.903140932650891e-06,
+      "loss": 0.7503,
+      "step": 56
+    },
+    {
+      "epoch": 0.07661933966227001,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.898898297831808e-06,
+      "loss": 0.7519,
+      "step": 57
+    },
+    {
+      "epoch": 0.07796353860371336,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.894565673985986e-06,
+      "loss": 0.7438,
+      "step": 58
+    },
+    {
+      "epoch": 0.07930773754515669,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.890143140700419e-06,
+      "loss": 0.7635,
+      "step": 59
+    },
+    {
+      "epoch": 0.08065193648660002,
+      "grad_norm": 1.375,
+      "learning_rate": 9.885630779213678e-06,
+      "loss": 0.7176,
+      "step": 60
+    },
+    {
+      "epoch": 0.08199613542804335,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.881028672414397e-06,
+      "loss": 0.7536,
+      "step": 61
+    },
+    {
+      "epoch": 0.08334033436948668,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.876336904839772e-06,
+      "loss": 0.7572,
+      "step": 62
+    },
+    {
+      "epoch": 0.08468453331093002,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.871555562673996e-06,
+      "loss": 0.7767,
+      "step": 63
+    },
+    {
+      "epoch": 0.08602873225237335,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.86668473374668e-06,
+      "loss": 0.763,
+      "step": 64
+    },
+    {
+      "epoch": 0.08737293119381669,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.861724507531234e-06,
+      "loss": 0.7576,
+      "step": 65
+    },
+    {
+      "epoch": 0.08871713013526002,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.856674975143237e-06,
+      "loss": 0.7342,
+      "step": 66
+    },
+    {
+      "epoch": 0.09006132907670335,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.851536229338747e-06,
+      "loss": 0.7487,
+      "step": 67
+    },
+    {
+      "epoch": 0.09140552801814669,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.846308364512607e-06,
+      "loss": 0.7604,
+      "step": 68
+    },
+    {
+      "epoch": 0.09274972695959002,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.840991476696707e-06,
+      "loss": 0.7537,
+      "step": 69
+    },
+    {
+      "epoch": 0.09409392590103335,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.835585663558221e-06,
+      "loss": 0.7608,
+      "step": 70
+    },
+    {
+      "epoch": 0.09543812484247668,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.830091024397818e-06,
+      "loss": 0.7298,
+      "step": 71
+    },
+    {
+      "epoch": 0.09678232378392002,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.824507660147831e-06,
+      "loss": 0.7646,
+      "step": 72
+    },
+    {
+      "epoch": 0.09812652272536336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.8188356733704e-06,
+      "loss": 0.7679,
+      "step": 73
+    },
+    {
+      "epoch": 0.09947072166680669,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.813075168255601e-06,
+      "loss": 0.7263,
+      "step": 74
+    },
+    {
+      "epoch": 0.10081492060825002,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.807226250619522e-06,
+      "loss": 0.7589,
+      "step": 75
+    },
+    {
+      "epoch": 0.10215911954969335,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.801289027902316e-06,
+      "loss": 0.7216,
+      "step": 76
+    },
+    {
+      "epoch": 0.10350331849113668,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.795263609166243e-06,
+      "loss": 0.7779,
+      "step": 77
+    },
+    {
+      "epoch": 0.10484751743258003,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.789150105093647e-06,
+      "loss": 0.6941,
+      "step": 78
+    },
+    {
+      "epoch": 0.10619171637402336,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.78294862798494e-06,
+      "loss": 0.7362,
+      "step": 79
+    },
+    {
+      "epoch": 0.10753591531546669,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.776659291756528e-06,
+      "loss": 0.7103,
+      "step": 80
+    },
+    {
+      "epoch": 0.10888011425691002,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.77028221193872e-06,
+      "loss": 0.6866,
+      "step": 81
+    },
+    {
+      "epoch": 0.11022431319835335,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.763817505673614e-06,
+      "loss": 0.7517,
+      "step": 82
+    },
+    {
+      "epoch": 0.1115685121397967,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.75726529171293e-06,
+      "loss": 0.7508,
+      "step": 83
+    },
+    {
+      "epoch": 0.11291271108124003,
+      "grad_norm": 1.5,
+      "learning_rate": 9.750625690415848e-06,
+      "loss": 0.7275,
+      "step": 84
+    },
+    {
+      "epoch": 0.11425691002268336,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.74389882374678e-06,
+      "loss": 0.6994,
+      "step": 85
+    },
+    {
+      "epoch": 0.11560110896412669,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.737084815273137e-06,
+      "loss": 0.7365,
+      "step": 86
+    },
+    {
+      "epoch": 0.11694530790557002,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.730183790163061e-06,
+      "loss": 0.7294,
+      "step": 87
+    },
+    {
+      "epoch": 0.11828950684701336,
+      "grad_norm": 1.375,
+      "learning_rate": 9.72319587518312e-06,
+      "loss": 0.7226,
+      "step": 88
+    },
+    {
+      "epoch": 0.1196337057884567,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.716121198695987e-06,
+      "loss": 0.7126,
+      "step": 89
+    },
+    {
+      "epoch": 0.12097790472990003,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.708959890658074e-06,
+      "loss": 0.7118,
+      "step": 90
+    },
+    {
+      "epoch": 0.12232210367134336,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.70171208261715e-06,
+      "loss": 0.7164,
+      "step": 91
+    },
+    {
+      "epoch": 0.12366630261278669,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.69437790770992e-06,
+      "loss": 0.7015,
+      "step": 92
+    },
+    {
+      "epoch": 0.12501050155423002,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.68695750065959e-06,
+      "loss": 0.7212,
+      "step": 93
+    },
+    {
+      "epoch": 0.12635470049567335,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.679450997773378e-06,
+      "loss": 0.7301,
+      "step": 94
+    },
+    {
+      "epoch": 0.12769889943711668,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.67185853694002e-06,
+      "loss": 0.7434,
+      "step": 95
+    },
+    {
+      "epoch": 0.12904309837856004,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.664180257627231e-06,
+      "loss": 0.7503,
+      "step": 96
+    },
+    {
+      "epoch": 0.13038729732000337,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.656416300879147e-06,
+      "loss": 0.704,
+      "step": 97
+    },
+    {
+      "epoch": 0.1317314962614467,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.648566809313738e-06,
+      "loss": 0.7091,
+      "step": 98
+    },
+    {
+      "epoch": 0.13307569520289003,
+      "grad_norm": 1.25,
+      "learning_rate": 9.640631927120177e-06,
+      "loss": 0.6939,
+      "step": 99
+    },
+    {
+      "epoch": 0.13441989414433336,
+      "grad_norm": 1.25,
+      "learning_rate": 9.632611800056202e-06,
+      "loss": 0.6645,
+      "step": 100
+    },
+    {
+      "epoch": 0.1357640930857767,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.62450657544543e-06,
+      "loss": 0.7314,
+      "step": 101
+    },
+    {
+      "epoch": 0.13710829202722002,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.616316402174657e-06,
+      "loss": 0.7022,
+      "step": 102
+    },
+    {
+      "epoch": 0.13845249096866336,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.608041430691126e-06,
+      "loss": 0.7015,
+      "step": 103
+    },
+    {
+      "epoch": 0.1397966899101067,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.59968181299975e-06,
+      "loss": 0.6831,
+      "step": 104
+    },
+    {
+      "epoch": 0.14114088885155002,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.591237702660335e-06,
+      "loss": 0.6903,
+      "step": 105
+    },
+    {
+      "epoch": 0.14248508779299338,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.58270925478475e-06,
+      "loss": 0.7236,
+      "step": 106
+    },
+    {
+      "epoch": 0.1438292867344367,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.574096626034077e-06,
+      "loss": 0.7375,
+      "step": 107
+    },
+    {
+      "epoch": 0.14517348567588004,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.565399974615744e-06,
+      "loss": 0.7051,
+      "step": 108
+    },
+    {
+      "epoch": 0.14651768461732337,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.556619460280605e-06,
+      "loss": 0.6961,
+      "step": 109
+    },
+    {
+      "epoch": 0.1478618835587667,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.547755244320013e-06,
+      "loss": 0.731,
+      "step": 110
+    },
+    {
+      "epoch": 0.14920608250021003,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.53880748956286e-06,
+      "loss": 0.6992,
+      "step": 111
+    },
+    {
+      "epoch": 0.15055028144165336,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.529776360372576e-06,
+      "loss": 0.6954,
+      "step": 112
+    },
+    {
+      "epoch": 0.1518944803830967,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.52066202264412e-06,
+      "loss": 0.6932,
+      "step": 113
+    },
+    {
+      "epoch": 0.15323867932454002,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.511464643800926e-06,
+      "loss": 0.7117,
+      "step": 114
+    },
+    {
+      "epoch": 0.15458287826598335,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.502184392791834e-06,
+      "loss": 0.6992,
+      "step": 115
+    },
+    {
+      "epoch": 0.1559270772074267,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.492821440087978e-06,
+      "loss": 0.6744,
+      "step": 116
+    },
+    {
+      "epoch": 0.15727127614887004,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.48337595767966e-06,
+      "loss": 0.6553,
+      "step": 117
+    },
+    {
+      "epoch": 0.15861547509031337,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.473848119073188e-06,
+      "loss": 0.7206,
+      "step": 118
+    },
+    {
+      "epoch": 0.1599596740317567,
+      "grad_norm": 1.125,
+      "learning_rate": 9.4642380992877e-06,
+      "loss": 0.7014,
+      "step": 119
+    },
+    {
+      "epoch": 0.16130387297320004,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.454546074851927e-06,
+      "loss": 0.7249,
+      "step": 120
+    },
+    {
+      "epoch": 0.16264807191464337,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.444772223800972e-06,
+      "loss": 0.7142,
+      "step": 121
+    },
+    {
+      "epoch": 0.1639922708560867,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.434916725673023e-06,
+      "loss": 0.6778,
+      "step": 122
+    },
+    {
+      "epoch": 0.16533646979753003,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.42497976150607e-06,
+      "loss": 0.6831,
+      "step": 123
+    },
+    {
+      "epoch": 0.16668066873897336,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.414961513834569e-06,
+      "loss": 0.6744,
+      "step": 124
+    },
+    {
+      "epoch": 0.1680248676804167,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.404862166686089e-06,
+      "loss": 0.7059,
+      "step": 125
+    },
+    {
+      "epoch": 0.16936906662186005,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.394681905577938e-06,
+      "loss": 0.7245,
+      "step": 126
+    },
+    {
+      "epoch": 0.17071326556330338,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.384420917513752e-06,
+      "loss": 0.7064,
+      "step": 127
+    },
+    {
+      "epoch": 0.1720574645047467,
+      "grad_norm": 1.0,
+      "learning_rate": 9.374079390980058e-06,
+      "loss": 0.7177,
+      "step": 128
+    },
+    {
+      "epoch": 0.17340166344619004,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.363657515942814e-06,
+      "loss": 0.7031,
+      "step": 129
+    },
+    {
+      "epoch": 0.17474586238763337,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.35315548384392e-06,
+      "loss": 0.6752,
+      "step": 130
+    },
+    {
+      "epoch": 0.1760900613290767,
+      "grad_norm": 1.0,
+      "learning_rate": 9.342573487597696e-06,
+      "loss": 0.7189,
+      "step": 131
+    },
+    {
+      "epoch": 0.17743426027052003,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.331911721587345e-06,
+      "loss": 0.6873,
+      "step": 132
+    },
+    {
+      "epoch": 0.17877845921196336,
+      "grad_norm": 1.0,
+      "learning_rate": 9.321170381661383e-06,
+      "loss": 0.7056,
+      "step": 133
+    },
+    {
+      "epoch": 0.1801226581534067,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.310349665130035e-06,
+      "loss": 0.7024,
+      "step": 134
+    },
+    {
+      "epoch": 0.18146685709485003,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.299449770761612e-06,
+      "loss": 0.708,
+      "step": 135
+    },
+    {
+      "epoch": 0.18281105603629338,
+      "grad_norm": 0.8984375,
+      "learning_rate": 9.288470898778863e-06,
+      "loss": 0.6624,
+      "step": 136
+    },
+    {
+      "epoch": 0.18415525497773672,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.277413250855296e-06,
+      "loss": 0.6622,
+      "step": 137
+    },
+    {
+      "epoch": 0.18549945391918005,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.266277030111474e-06,
+      "loss": 0.7074,
+      "step": 138
+    },
+    {
+      "epoch": 0.18684365286062338,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.25506244111128e-06,
+      "loss": 0.6745,
+      "step": 139
+    },
+    {
+      "epoch": 0.1881878518020667,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9.243769689858167e-06,
+      "loss": 0.6627,
+      "step": 140
+    },
+    {
+      "epoch": 0.18953205074351004,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.232398983791363e-06,
+      "loss": 0.6563,
+      "step": 141
+    },
+    {
+      "epoch": 0.19087624968495337,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.220950531782069e-06,
+      "loss": 0.688,
+      "step": 142
+    },
+    {
+      "epoch": 0.1922204486263967,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.209424544129621e-06,
+      "loss": 0.7021,
+      "step": 143
+    },
+    {
+      "epoch": 0.19356464756784003,
+      "grad_norm": 0.87890625,
+      "learning_rate": 9.197821232557625e-06,
+      "loss": 0.6826,
+      "step": 144
+    },
+    {
+      "epoch": 0.19490884650928336,
+      "grad_norm": 0.9140625,
+      "learning_rate": 9.186140810210066e-06,
+      "loss": 0.6904,
+      "step": 145
+    },
+    {
+      "epoch": 0.19625304545072672,
+      "grad_norm": 1.0,
+      "learning_rate": 9.1743834916474e-06,
+      "loss": 0.6747,
+      "step": 146
+    },
+    {
+      "epoch": 0.19759724439217005,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.162549492842603e-06,
+      "loss": 0.7073,
+      "step": 147
+    },
+    {
+      "epoch": 0.19894144333361338,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.150639031177211e-06,
+      "loss": 0.6773,
+      "step": 148
+    },
+    {
+      "epoch": 0.2002856422750567,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.138652325437326e-06,
+      "loss": 0.689,
+      "step": 149
+    },
+    {
+      "epoch": 0.20162984121650004,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.12658959580959e-06,
+      "loss": 0.6874,
+      "step": 150
+    },
+    {
+      "epoch": 0.20297404015794338,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.114451063877152e-06,
+      "loss": 0.6617,
+      "step": 151
+    },
+    {
+      "epoch": 0.2043182390993867,
+      "grad_norm": 1.0,
+      "learning_rate": 9.102236952615588e-06,
+      "loss": 0.6882,
+      "step": 152
+    },
+    {
+      "epoch": 0.20566243804083004,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.08994748638881e-06,
+      "loss": 0.6919,
+      "step": 153
+    },
+    {
+      "epoch": 0.20700663698227337,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.077582890944945e-06,
+      "loss": 0.6713,
+      "step": 154
+    },
+    {
+      "epoch": 0.2083508359237167,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.065143393412179e-06,
+      "loss": 0.6504,
+      "step": 155
+    },
+    {
+      "epoch": 0.20969503486516006,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.052629222294605e-06,
+      "loss": 0.6674,
+      "step": 156
+    },
+    {
+      "epoch": 0.2110392338066034,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.040040607467999e-06,
+      "loss": 0.7111,
+      "step": 157
+    },
+    {
+      "epoch": 0.21238343274804672,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.02737778017562e-06,
+      "loss": 0.6596,
+      "step": 158
+    },
+    {
+      "epoch": 0.21372763168949005,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.014640973023951e-06,
+      "loss": 0.6846,
+      "step": 159
+    },
+    {
+      "epoch": 0.21507183063093338,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.00183041997843e-06,
+      "loss": 0.6788,
+      "step": 160
+    },
+    {
+      "epoch": 0.2164160295723767,
+      "grad_norm": 0.890625,
+      "learning_rate": 8.988946356359147e-06,
+      "loss": 0.6888,
+      "step": 161
+    },
+    {
+      "epoch": 0.21776022851382004,
+      "grad_norm": 0.92578125,
+      "learning_rate": 8.97598901883653e-06,
+      "loss": 0.6854,
+      "step": 162
+    },
+    {
+      "epoch": 0.21910442745526337,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.962958645426989e-06,
+      "loss": 0.7137,
+      "step": 163
+    },
+    {
+      "epoch": 0.2204486263967067,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.949855475488549e-06,
+      "loss": 0.7041,
+      "step": 164
+    },
+    {
+      "epoch": 0.22179282533815003,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.936679749716452e-06,
+      "loss": 0.6573,
+      "step": 165
+    },
+    {
+      "epoch": 0.2231370242795934,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.923431710138735e-06,
+      "loss": 0.6694,
+      "step": 166
+    },
+    {
+      "epoch": 0.22448122322103672,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.910111600111786e-06,
+      "loss": 0.6904,
+      "step": 167
+    },
+    {
+      "epoch": 0.22582542216248006,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.896719664315866e-06,
+      "loss": 0.679,
+      "step": 168
+    },
+    {
+      "epoch": 0.22716962110392339,
+      "grad_norm": 0.9140625,
+      "learning_rate": 8.883256148750634e-06,
+      "loss": 0.6631,
+      "step": 169
+    },
+    {
+      "epoch": 0.22851382004536672,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.869721300730596e-06,
+      "loss": 0.6576,
+      "step": 170
+    },
+    {
+      "epoch": 0.22985801898681005,
+      "grad_norm": 0.9296875,
+      "learning_rate": 8.856115368880598e-06,
+      "loss": 0.6714,
+      "step": 171
+    },
+    {
+      "epoch": 0.23120221792825338,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.842438603131232e-06,
+      "loss": 0.6572,
+      "step": 172
+    },
+    {
+      "epoch": 0.2325464168696967,
+      "grad_norm": 0.91796875,
+      "learning_rate": 8.828691254714259e-06,
+      "loss": 0.6798,
+      "step": 173
+    },
+    {
+      "epoch": 0.23389061581114004,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.814873576157988e-06,
+      "loss": 0.6934,
+      "step": 174
+    },
+    {
+      "epoch": 0.23523481475258337,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.800985821282637e-06,
+      "loss": 0.6648,
+      "step": 175
+    },
+    {
+      "epoch": 0.23657901369402673,
+      "grad_norm": 0.8203125,
+      "learning_rate": 8.787028245195676e-06,
+      "loss": 0.6705,
+      "step": 176
+    },
+    {
+      "epoch": 0.23792321263547006,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.773001104287137e-06,
+      "loss": 0.6661,
+      "step": 177
+    },
+    {
+      "epoch": 0.2392674115769134,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.758904656224904e-06,
+      "loss": 0.6968,
+      "step": 178
+    },
+    {
+      "epoch": 0.24061161051835672,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.744739159949981e-06,
+      "loss": 0.6817,
+      "step": 179
+    },
+    {
+      "epoch": 0.24195580945980005,
+      "grad_norm": 0.9296875,
+      "learning_rate": 8.730504875671732e-06,
+      "loss": 0.6704,
+      "step": 180
+    },
+    {
+      "epoch": 0.24330000840124338,
+      "grad_norm": 0.90625,
+      "learning_rate": 8.71620206486311e-06,
+      "loss": 0.6484,
+      "step": 181
+    },
+    {
+      "epoch": 0.24464420734268671,
+      "grad_norm": 0.8515625,
+      "learning_rate": 8.701830990255843e-06,
+      "loss": 0.6711,
+      "step": 182
+    },
+    {
+      "epoch": 0.24598840628413005,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.687391915835617e-06,
+      "loss": 0.6526,
+      "step": 183
+    },
+    {
+      "epoch": 0.24733260522557338,
+      "grad_norm": 0.87890625,
+      "learning_rate": 8.672885106837216e-06,
+      "loss": 0.6907,
+      "step": 184
+    },
+    {
+      "epoch": 0.2486768041670167,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.658310829739666e-06,
+      "loss": 0.7003,
+      "step": 185
+    },
+    {
+      "epoch": 0.25002100310846004,
+      "grad_norm": 0.85546875,
+      "learning_rate": 8.643669352261321e-06,
+      "loss": 0.6353,
+      "step": 186
+    },
+    {
+      "epoch": 0.25136520204990337,
+      "grad_norm": 0.921875,
+      "learning_rate": 8.628960943354965e-06,
+      "loss": 0.6566,
+      "step": 187
+    },
+    {
+      "epoch": 0.2527094009913467,
+      "grad_norm": 0.92578125,
+      "learning_rate": 8.614185873202852e-06,
+      "loss": 0.6676,
+      "step": 188
+    },
+    {
+      "epoch": 0.25405359993279003,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.599344413211755e-06,
+      "loss": 0.6647,
+      "step": 189
+    },
+    {
+      "epoch": 0.25539779887423336,
+      "grad_norm": 0.9140625,
+      "learning_rate": 8.58443683600798e-06,
+      "loss": 0.6771,
+      "step": 190
+    },
+    {
+      "epoch": 0.2567419978156767,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.569463415432356e-06,
+      "loss": 0.6629,
+      "step": 191
+    },
+    {
+      "epoch": 0.2580861967571201,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.554424426535202e-06,
+      "loss": 0.7327,
+      "step": 192
+    },
+    {
+      "epoch": 0.2594303956985634,
+      "grad_norm": 0.94921875,
+      "learning_rate": 8.539320145571277e-06,
+      "loss": 0.6809,
+      "step": 193
+    },
+    {
+      "epoch": 0.26077459464000674,
+      "grad_norm": 0.875,
+      "learning_rate": 8.524150849994708e-06,
+      "loss": 0.6895,
+      "step": 194
+    },
+    {
+      "epoch": 0.26211879358145007,
+      "grad_norm": 0.9140625,
+      "learning_rate": 8.50891681845389e-06,
+      "loss": 0.6739,
+      "step": 195
+    },
+    {
+      "epoch": 0.2634629925228934,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.493618330786365e-06,
+      "loss": 0.7063,
+      "step": 196
+    },
+    {
+      "epoch": 0.26480719146433673,
+      "grad_norm": 0.85546875,
+      "learning_rate": 8.47825566801369e-06,
+      "loss": 0.6676,
+      "step": 197
+    },
+    {
+      "epoch": 0.26615139040578006,
+      "grad_norm": 0.81640625,
+      "learning_rate": 8.462829112336266e-06,
+      "loss": 0.6842,
+      "step": 198
+    },
+    {
+      "epoch": 0.2674955893472234,
+      "grad_norm": 0.90625,
+      "learning_rate": 8.44733894712816e-06,
+      "loss": 0.7045,
+      "step": 199
+    },
+    {
+      "epoch": 0.2688397882886667,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.431785456931898e-06,
+      "loss": 0.6569,
+      "step": 200
+    },
+    {
+      "epoch": 0.27018398723011006,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.416168927453237e-06,
+      "loss": 0.679,
+      "step": 201
+    },
+    {
+      "epoch": 0.2715281861715534,
+      "grad_norm": 0.8671875,
+      "learning_rate": 8.400489645555914e-06,
+      "loss": 0.6452,
+      "step": 202
+    },
+    {
+      "epoch": 0.2728723851129967,
+      "grad_norm": 0.9296875,
+      "learning_rate": 8.384747899256386e-06,
+      "loss": 0.6744,
+      "step": 203
+    },
+    {
+      "epoch": 0.27421658405444005,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.368943977718528e-06,
+      "loss": 0.6501,
+      "step": 204
+    },
+    {
+      "epoch": 0.2755607829958834,
+      "grad_norm": 0.890625,
+      "learning_rate": 8.353078171248335e-06,
+      "loss": 0.666,
+      "step": 205
+    },
+    {
+      "epoch": 0.2769049819373267,
+      "grad_norm": 0.87109375,
+      "learning_rate": 8.337150771288571e-06,
+      "loss": 0.6663,
+      "step": 206
+    },
+    {
+      "epoch": 0.27824918087877004,
+      "grad_norm": 0.9296875,
+      "learning_rate": 8.32116207041343e-06,
+      "loss": 0.6481,
+      "step": 207
+    },
+    {
+      "epoch": 0.2795933798202134,
+      "grad_norm": 0.921875,
+      "learning_rate": 8.30511236232316e-06,
+      "loss": 0.6884,
+      "step": 208
+    },
+    {
+      "epoch": 0.2809375787616567,
+      "grad_norm": 0.890625,
+      "learning_rate": 8.289001941838659e-06,
+      "loss": 0.6562,
+      "step": 209
+    },
+    {
+      "epoch": 0.28228177770310003,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.27283110489607e-06,
+      "loss": 0.7106,
+      "step": 210
+    },
+    {
+      "epoch": 0.28362597664454336,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.256600148541339e-06,
+      "loss": 0.689,
+      "step": 211
+    },
+    {
+      "epoch": 0.28497017558598675,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.240309370924758e-06,
+      "loss": 0.6683,
+      "step": 212
+    },
+    {
+      "epoch": 0.2863143745274301,
+      "grad_norm": 0.921875,
+      "learning_rate": 8.223959071295492e-06,
+      "loss": 0.6866,
+      "step": 213
+    },
+    {
+      "epoch": 0.2876585734688734,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.207549549996083e-06,
+      "loss": 0.6688,
+      "step": 214
+    },
+    {
+      "epoch": 0.28900277241031674,
+      "grad_norm": 0.91796875,
+      "learning_rate": 8.191081108456922e-06,
+      "loss": 0.6721,
+      "step": 215
+    },
+    {
+      "epoch": 0.2903469713517601,
+      "grad_norm": 0.89453125,
+      "learning_rate": 8.174554049190726e-06,
+      "loss": 0.6579,
+      "step": 216
+    },
+    {
+      "epoch": 0.2916911702932034,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.157968675786971e-06,
+      "loss": 0.6841,
+      "step": 217
+    },
+    {
+      "epoch": 0.29303536923464674,
+      "grad_norm": 0.93359375,
+      "learning_rate": 8.141325292906325e-06,
+      "loss": 0.6739,
+      "step": 218
+    },
+    {
+      "epoch": 0.29437956817609007,
+      "grad_norm": 0.875,
+      "learning_rate": 8.124624206275041e-06,
+      "loss": 0.6544,
+      "step": 219
+    },
+    {
+      "epoch": 0.2957237671175334,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.107865722679347e-06,
+      "loss": 0.6525,
+      "step": 220
+    },
+    {
+      "epoch": 0.29706796605897673,
+      "grad_norm": 0.890625,
+      "learning_rate": 8.091050149959808e-06,
+      "loss": 0.6394,
+      "step": 221
+    },
+    {
+      "epoch": 0.29841216500042006,
+      "grad_norm": 0.87890625,
+      "learning_rate": 8.074177797005677e-06,
+      "loss": 0.6674,
+      "step": 222
+    },
+    {
+      "epoch": 0.2997563639418634,
+      "grad_norm": 0.984375,
+      "learning_rate": 8.057248973749216e-06,
+      "loss": 0.6749,
+      "step": 223
+    },
+    {
+      "epoch": 0.3011005628833067,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.040263991159996e-06,
+      "loss": 0.7223,
+      "step": 224
+    },
+    {
+      "epoch": 0.30244476182475005,
+      "grad_norm": 0.859375,
+      "learning_rate": 8.0232231612392e-06,
+      "loss": 0.6861,
+      "step": 225
+    },
+    {
+      "epoch": 0.3037889607661934,
+      "grad_norm": 0.9140625,
+      "learning_rate": 8.006126797013884e-06,
+      "loss": 0.6883,
+      "step": 226
+    },
+    {
+      "epoch": 0.3051331597076367,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.98897521253122e-06,
+      "loss": 0.655,
+      "step": 227
+    },
+    {
+      "epoch": 0.30647735864908004,
+      "grad_norm": 0.8984375,
+      "learning_rate": 7.971768722852741e-06,
+      "loss": 0.6696,
+      "step": 228
+    },
+    {
+      "epoch": 0.3078215575905234,
+      "grad_norm": 0.875,
+      "learning_rate": 7.954507644048544e-06,
+      "loss": 0.6676,
+      "step": 229
+    },
+    {
+      "epoch": 0.3091657565319667,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.937192293191485e-06,
+      "loss": 0.6508,
+      "step": 230
+    },
+    {
+      "epoch": 0.31050995547341004,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.919822988351359e-06,
+      "loss": 0.6676,
+      "step": 231
+    },
+    {
+      "epoch": 0.3118541544148534,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.902400048589051e-06,
+      "loss": 0.6512,
+      "step": 232
+    },
+    {
+      "epoch": 0.31319835335629675,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.884923793950684e-06,
+      "loss": 0.6846,
+      "step": 233
+    },
+    {
+      "epoch": 0.3145425522977401,
+      "grad_norm": 0.8515625,
+      "learning_rate": 7.86739454546173e-06,
+      "loss": 0.6644,
+      "step": 234
+    },
+    {
+      "epoch": 0.3158867512391834,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.849812625121122e-06,
+      "loss": 0.6701,
+      "step": 235
+    },
+    {
+      "epoch": 0.31723095018062675,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.832178355895327e-06,
+      "loss": 0.6749,
+      "step": 236
+    },
+    {
+      "epoch": 0.3185751491220701,
+      "grad_norm": 0.87890625,
+      "learning_rate": 7.81449206171243e-06,
+      "loss": 0.6584,
+      "step": 237
+    },
+    {
+      "epoch": 0.3199193480635134,
+      "grad_norm": 0.890625,
+      "learning_rate": 7.796754067456168e-06,
+      "loss": 0.6685,
+      "step": 238
+    },
+    {
+      "epoch": 0.32126354700495674,
+      "grad_norm": 0.8671875,
+      "learning_rate": 7.778964698959973e-06,
+      "loss": 0.6659,
+      "step": 239
+    },
+    {
+      "epoch": 0.32260774594640007,
+      "grad_norm": 0.8671875,
+      "learning_rate": 7.761124283000982e-06,
+      "loss": 0.6647,
+      "step": 240
+    },
+    {
+      "epoch": 0.3239519448878434,
+      "grad_norm": 0.9140625,
+      "learning_rate": 7.743233147294036e-06,
+      "loss": 0.6793,
+      "step": 241
+    },
+    {
+      "epoch": 0.32529614382928673,
+      "grad_norm": 0.87109375,
+      "learning_rate": 7.725291620485653e-06,
+      "loss": 0.6633,
+      "step": 242
+    },
+    {
+      "epoch": 0.32664034277073006,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.707300032148004e-06,
+      "loss": 0.6812,
+      "step": 243
+    },
+    {
+      "epoch": 0.3279845417121734,
+      "grad_norm": 0.88671875,
+      "learning_rate": 7.689258712772851e-06,
+      "loss": 0.6734,
+      "step": 244
+    },
+    {
+      "epoch": 0.3293287406536167,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.671167993765474e-06,
+      "loss": 0.6775,
+      "step": 245
+    },
+    {
+      "epoch": 0.33067293959506006,
+      "grad_norm": 0.84375,
+      "learning_rate": 7.653028207438588e-06,
+      "loss": 0.6723,
+      "step": 246
+    },
+    {
+      "epoch": 0.3320171385365034,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.634839687006242e-06,
+      "loss": 0.6814,
+      "step": 247
+    },
+    {
+      "epoch": 0.3333613374779467,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.616602766577683e-06,
+      "loss": 0.6816,
+      "step": 248
+    },
+    {
+      "epoch": 0.3333613374779467,
+      "eval_loss": 2.7341208457946777,
+      "eval_runtime": 5734.9058,
+      "eval_samples_per_second": 3.861,
+      "eval_steps_per_second": 0.483,
+      "step": 248
+    },
+    {
+      "epoch": 0.33470553641939005,
+      "grad_norm": 0.8359375,
+      "learning_rate": 7.59831778115124e-06,
+      "loss": 0.6496,
+      "step": 249
+    },
+    {
+      "epoch": 0.3360497353608334,
+      "grad_norm": 0.8515625,
+      "learning_rate": 7.579985066608153e-06,
+      "loss": 0.6608,
+      "step": 250
+    },
+    {
+      "epoch": 0.3373939343022767,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.56160495970641e-06,
+      "loss": 0.6673,
+      "step": 251
+    },
+    {
+      "epoch": 0.3387381332437201,
+      "grad_norm": 0.8984375,
+      "learning_rate": 7.543177798074564e-06,
+      "loss": 0.6474,
+      "step": 252
+    },
+    {
+      "epoch": 0.3400823321851634,
+      "grad_norm": 0.8671875,
+      "learning_rate": 7.524703920205521e-06,
+      "loss": 0.6679,
+      "step": 253
+    },
+    {
+      "epoch": 0.34142653112660676,
+      "grad_norm": 0.87890625,
+      "learning_rate": 7.506183665450336e-06,
+      "loss": 0.6638,
+      "step": 254
+    },
+    {
+      "epoch": 0.3427707300680501,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.487617374011968e-06,
+      "loss": 0.6451,
+      "step": 255
+    },
+    {
+      "epoch": 0.3441149290094934,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.469005386939036e-06,
+      "loss": 0.656,
+      "step": 256
+    },
+    {
+      "epoch": 0.34545912795093675,
+      "grad_norm": 0.92578125,
+      "learning_rate": 7.450348046119551e-06,
+      "loss": 0.6103,
+      "step": 257
+    },
+    {
+      "epoch": 0.3468033268923801,
+      "grad_norm": 0.87890625,
+      "learning_rate": 7.43164569427464e-06,
+      "loss": 0.6534,
+      "step": 258
+    },
+    {
+      "epoch": 0.3481475258338234,
+      "grad_norm": 0.84375,
+      "learning_rate": 7.4128986749522495e-06,
+      "loss": 0.647,
+      "step": 259
+    },
+    {
+      "epoch": 0.34949172477526674,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.394107332520828e-06,
+      "loss": 0.6478,
+      "step": 260
+    },
+    {
+      "epoch": 0.3508359237167101,
+      "grad_norm": 0.84765625,
+      "learning_rate": 7.375272012163011e-06,
+      "loss": 0.657,
+      "step": 261
+    },
+    {
+      "epoch": 0.3521801226581534,
+      "grad_norm": 0.81640625,
+      "learning_rate": 7.3563930598692725e-06,
+      "loss": 0.6087,
+      "step": 262
+    },
+    {
+      "epoch": 0.35352432159959674,
+      "grad_norm": 0.828125,
+      "learning_rate": 7.3374708224315725e-06,
+      "loss": 0.6572,
+      "step": 263
+    },
+    {
+      "epoch": 0.35486852054104007,
+      "grad_norm": 0.84375,
+      "learning_rate": 7.318505647436986e-06,
+      "loss": 0.636,
+      "step": 264
+    },
+    {
+      "epoch": 0.3562127194824834,
+      "grad_norm": 0.875,
+      "learning_rate": 7.299497883261319e-06,
+      "loss": 0.6426,
+      "step": 265
+    },
+    {
+      "epoch": 0.35755691842392673,
+      "grad_norm": 0.83984375,
+      "learning_rate": 7.280447879062711e-06,
+      "loss": 0.6429,
+      "step": 266
+    },
+    {
+      "epoch": 0.35890111736537006,
+      "grad_norm": 0.91796875,
+      "learning_rate": 7.261355984775208e-06,
+      "loss": 0.6658,
+      "step": 267
+    },
+    {
+      "epoch": 0.3602453163068134,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.2422225511023555e-06,
+      "loss": 0.6765,
+      "step": 268
+    },
+    {
+      "epoch": 0.3615895152482567,
+      "grad_norm": 0.84375,
+      "learning_rate": 7.223047929510744e-06,
+      "loss": 0.6313,
+      "step": 269
+    },
+    {
+      "epoch": 0.36293371418970005,
+      "grad_norm": 0.80859375,
+      "learning_rate": 7.203832472223551e-06,
+      "loss": 0.6479,
+      "step": 270
+    },
+    {
+      "epoch": 0.3642779131311434,
+      "grad_norm": 0.83984375,
+      "learning_rate": 7.184576532214077e-06,
+      "loss": 0.6491,
+      "step": 271
+    },
+    {
+      "epoch": 0.36562211207258677,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.16528046319926e-06,
+      "loss": 0.7071,
+      "step": 272
+    },
+    {
+      "epoch": 0.3669663110140301,
+      "grad_norm": 0.91015625,
+      "learning_rate": 7.145944619633176e-06,
+      "loss": 0.6588,
+      "step": 273
+    },
+    {
+      "epoch": 0.36831050995547343,
+      "grad_norm": 0.86328125,
+      "learning_rate": 7.126569356700529e-06,
+      "loss": 0.6662,
+      "step": 274
+    },
+    {
+      "epoch": 0.36965470889691676,
+      "grad_norm": 0.87109375,
+      "learning_rate": 7.107155030310127e-06,
+      "loss": 0.6488,
+      "step": 275
+    },
+    {
+      "epoch": 0.3709989078383601,
+      "grad_norm": 0.89453125,
+      "learning_rate": 7.087701997088344e-06,
+      "loss": 0.677,
+      "step": 276
+    },
+    {
+      "epoch": 0.3723431067798034,
+      "grad_norm": 0.875,
+      "learning_rate": 7.068210614372568e-06,
+      "loss": 0.6609,
+      "step": 277
+    },
+    {
+      "epoch": 0.37368730572124675,
+      "grad_norm": 0.83203125,
+      "learning_rate": 7.048681240204641e-06,
+      "loss": 0.6594,
+      "step": 278
+    },
+    {
+      "epoch": 0.3750315046626901,
+      "grad_norm": 0.875,
+      "learning_rate": 7.029114233324277e-06,
+      "loss": 0.6553,
+      "step": 279
+    },
+    {
+      "epoch": 0.3763757036041334,
+      "grad_norm": 0.921875,
+      "learning_rate": 7.0095099531624715e-06,
+      "loss": 0.6669,
+      "step": 280
+    },
+    {
+      "epoch": 0.37771990254557675,
+      "grad_norm": 0.890625,
+      "learning_rate": 6.989868759834908e-06,
+      "loss": 0.6445,
+      "step": 281
+    },
+    {
+      "epoch": 0.3790641014870201,
+      "grad_norm": 0.8359375,
+      "learning_rate": 6.970191014135331e-06,
+      "loss": 0.6635,
+      "step": 282
+    },
+    {
+      "epoch": 0.3804083004284634,
+      "grad_norm": 0.86328125,
+      "learning_rate": 6.950477077528927e-06,
+      "loss": 0.7111,
+      "step": 283
+    },
+    {
+      "epoch": 0.38175249936990674,
+      "grad_norm": 0.84375,
+      "learning_rate": 6.93072731214568e-06,
+      "loss": 0.6149,
+      "step": 284
+    },
+    {
+      "epoch": 0.38309669831135007,
+      "grad_norm": 0.88671875,
+      "learning_rate": 6.910942080773725e-06,
+      "loss": 0.6709,
+      "step": 285
+    },
+    {
+      "epoch": 0.3844408972527934,
+      "grad_norm": 0.91015625,
+      "learning_rate": 6.891121746852675e-06,
+      "loss": 0.6614,
+      "step": 286
+    },
+    {
+      "epoch": 0.38578509619423673,
+      "grad_norm": 0.8515625,
+      "learning_rate": 6.8712666744669555e-06,
+      "loss": 0.6694,
+      "step": 287
+    },
+    {
+      "epoch": 0.38712929513568006,
+      "grad_norm": 0.83203125,
+      "learning_rate": 6.851377228339106e-06,
+      "loss": 0.6441,
+      "step": 288
+    },
+    {
+      "epoch": 0.3884734940771234,
+      "grad_norm": 0.8203125,
+      "learning_rate": 6.831453773823091e-06,
+      "loss": 0.6131,
+      "step": 289
+    },
+    {
+      "epoch": 0.3898176930185667,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.811496676897577e-06,
+      "loss": 0.6441,
+      "step": 290
+    },
+    {
+      "epoch": 0.39116189196001006,
+      "grad_norm": 0.859375,
+      "learning_rate": 6.791506304159221e-06,
+      "loss": 0.6395,
+      "step": 291
+    },
+    {
+      "epoch": 0.39250609090145344,
+      "grad_norm": 0.88671875,
+      "learning_rate": 6.771483022815926e-06,
+      "loss": 0.6647,
+      "step": 292
+    },
+    {
+      "epoch": 0.3938502898428968,
+      "grad_norm": 0.86328125,
+      "learning_rate": 6.751427200680109e-06,
+      "loss": 0.6705,
+      "step": 293
+    },
+    {
+      "epoch": 0.3951944887843401,
+      "grad_norm": 0.8203125,
+      "learning_rate": 6.731339206161928e-06,
+      "loss": 0.6415,
+      "step": 294
+    },
+    {
+      "epoch": 0.39653868772578343,
+      "grad_norm": 0.90625,
+      "learning_rate": 6.711219408262527e-06,
+      "loss": 0.6557,
+      "step": 295
+    },
+    {
+      "epoch": 0.39788288666722677,
+      "grad_norm": 0.83203125,
+      "learning_rate": 6.691068176567257e-06,
+      "loss": 0.6598,
+      "step": 296
+    },
+    {
+      "epoch": 0.3992270856086701,
+      "grad_norm": 0.87890625,
+      "learning_rate": 6.6708858812388776e-06,
+      "loss": 0.6459,
+      "step": 297
+    },
+    {
+      "epoch": 0.4005712845501134,
+      "grad_norm": 0.8203125,
+      "learning_rate": 6.650672893010769e-06,
+      "loss": 0.6435,
+      "step": 298
+    },
+    {
+      "epoch": 0.40191548349155676,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.630429583180113e-06,
+      "loss": 0.6317,
+      "step": 299
+    },
+    {
+      "epoch": 0.4032596824330001,
+      "grad_norm": 0.828125,
+      "learning_rate": 6.610156323601076e-06,
+      "loss": 0.6647,
+      "step": 300
+    },
+    {
+      "epoch": 0.4046038813744434,
+      "grad_norm": 0.91796875,
+      "learning_rate": 6.5898534866779815e-06,
+      "loss": 0.6754,
+      "step": 301
+    },
+    {
+      "epoch": 0.40594808031588675,
+      "grad_norm": 0.84765625,
+      "learning_rate": 6.5695214453584645e-06,
+      "loss": 0.6588,
+      "step": 302
+    },
+    {
+      "epoch": 0.4072922792573301,
+      "grad_norm": 0.86328125,
+      "learning_rate": 6.549160573126623e-06,
+      "loss": 0.6554,
+      "step": 303
+    },
+    {
+      "epoch": 0.4086364781987734,
+      "grad_norm": 0.81640625,
+      "learning_rate": 6.528771243996157e-06,
+      "loss": 0.6297,
+      "step": 304
+    },
+    {
+      "epoch": 0.40998067714021674,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.508353832503495e-06,
+      "loss": 0.6734,
+      "step": 305
+    },
+    {
+      "epoch": 0.4113248760816601,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.48790871370092e-06,
+      "loss": 0.6298,
+      "step": 306
+    },
+    {
+      "epoch": 0.4126690750231034,
+      "grad_norm": 0.8828125,
+      "learning_rate": 6.467436263149679e-06,
+      "loss": 0.6294,
+      "step": 307
+    },
+    {
+      "epoch": 0.41401327396454674,
+      "grad_norm": 0.8515625,
+      "learning_rate": 6.4469368569130786e-06,
+      "loss": 0.6473,
+      "step": 308
+    },
+    {
+      "epoch": 0.41535747290599007,
+      "grad_norm": 0.8515625,
+      "learning_rate": 6.426410871549582e-06,
+      "loss": 0.651,
+      "step": 309
+    },
+    {
+      "epoch": 0.4167016718474334,
+      "grad_norm": 0.87890625,
+      "learning_rate": 6.405858684105892e-06,
+      "loss": 0.6939,
+      "step": 310
+    },
+    {
+      "epoch": 0.41804587078887673,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.3852806721100235e-06,
+      "loss": 0.6567,
+      "step": 311
+    },
+    {
+      "epoch": 0.4193900697303201,
+      "grad_norm": 0.84375,
+      "learning_rate": 6.364677213564365e-06,
+      "loss": 0.6303,
+      "step": 312
+    },
+    {
+      "epoch": 0.42073426867176345,
+      "grad_norm": 0.859375,
+      "learning_rate": 6.344048686938745e-06,
+      "loss": 0.6575,
+      "step": 313
+    },
+    {
+      "epoch": 0.4220784676132068,
+      "grad_norm": 0.79296875,
+      "learning_rate": 6.323395471163467e-06,
+      "loss": 0.6529,
+      "step": 314
+    },
+    {
+      "epoch": 0.4234226665546501,
+      "grad_norm": 0.80078125,
+      "learning_rate": 6.30271794562236e-06,
+      "loss": 0.6593,
+      "step": 315
+    },
+    {
+      "epoch": 0.42476686549609344,
+      "grad_norm": 0.8515625,
+      "learning_rate": 6.282016490145804e-06,
+      "loss": 0.647,
+      "step": 316
+    },
+    {
+      "epoch": 0.42611106443753677,
+      "grad_norm": 0.8125,
+      "learning_rate": 6.261291485003751e-06,
+      "loss": 0.6456,
+      "step": 317
+    },
+    {
+      "epoch": 0.4274552633789801,
+      "grad_norm": 0.828125,
+      "learning_rate": 6.2405433108987456e-06,
+      "loss": 0.6761,
+      "step": 318
+    },
+    {
+      "epoch": 0.42879946232042343,
+      "grad_norm": 0.80859375,
+      "learning_rate": 6.219772348958928e-06,
+      "loss": 0.6518,
+      "step": 319
+    },
+    {
+      "epoch": 0.43014366126186676,
+      "grad_norm": 0.8203125,
+      "learning_rate": 6.198978980731034e-06,
+      "loss": 0.6249,
+      "step": 320
+    },
+    {
+      "epoch": 0.4314878602033101,
+      "grad_norm": 0.84375,
+      "learning_rate": 6.17816358817338e-06,
+      "loss": 0.6582,
+      "step": 321
+    },
+    {
+      "epoch": 0.4328320591447534,
+      "grad_norm": 0.8671875,
+      "learning_rate": 6.157326553648862e-06,
+      "loss": 0.6616,
+      "step": 322
+    },
+    {
+      "epoch": 0.43417625808619675,
+      "grad_norm": 0.8359375,
+      "learning_rate": 6.136468259917917e-06,
+      "loss": 0.6463,
+      "step": 323
+    },
+    {
+      "epoch": 0.4355204570276401,
+      "grad_norm": 0.8359375,
+      "learning_rate": 6.115589090131496e-06,
+      "loss": 0.6469,
+      "step": 324
+    },
+    {
+      "epoch": 0.4368646559690834,
+      "grad_norm": 0.8671875,
+      "learning_rate": 6.094689427824031e-06,
+      "loss": 0.6575,
+      "step": 325
+    },
+    {
+      "epoch": 0.43820885491052675,
+      "grad_norm": 0.86328125,
+      "learning_rate": 6.073769656906386e-06,
+      "loss": 0.6369,
+      "step": 326
+    },
+    {
+      "epoch": 0.4395530538519701,
+      "grad_norm": 0.8828125,
+      "learning_rate": 6.052830161658799e-06,
+      "loss": 0.6787,
+      "step": 327
+    },
+    {
+      "epoch": 0.4408972527934134,
+      "grad_norm": 0.84375,
+      "learning_rate": 6.031871326723837e-06,
+      "loss": 0.6556,
+      "step": 328
+    },
+    {
+      "epoch": 0.44224145173485674,
+      "grad_norm": 0.86328125,
+      "learning_rate": 6.010893537099316e-06,
+      "loss": 0.6669,
+      "step": 329
+    },
+    {
+      "epoch": 0.44358565067630007,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5.989897178131239e-06,
+      "loss": 0.6495,
+      "step": 330
+    },
+    {
+      "epoch": 0.4449298496177434,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.968882635506711e-06,
+      "loss": 0.6547,
+      "step": 331
+    },
+    {
+      "epoch": 0.4462740485591868,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5.947850295246859e-06,
+      "loss": 0.6327,
+      "step": 332
+    },
+    {
+      "epoch": 0.4476182475006301,
+      "grad_norm": 0.83203125,
+      "learning_rate": 5.92680054369974e-06,
+      "loss": 0.6238,
+      "step": 333
+    },
+    {
+      "epoch": 0.44896244644207345,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.905733767533238e-06,
+      "loss": 0.6675,
+      "step": 334
+    },
+    {
+      "epoch": 0.4503066453835168,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.884650353727972e-06,
+      "loss": 0.676,
+      "step": 335
+    },
+    {
+      "epoch": 0.4516508443249601,
+      "grad_norm": 0.859375,
+      "learning_rate": 5.863550689570179e-06,
+      "loss": 0.6651,
+      "step": 336
+    },
+    {
+      "epoch": 0.45299504326640344,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.842435162644602e-06,
+      "loss": 0.6574,
+      "step": 337
+    },
+    {
+      "epoch": 0.45433924220784677,
+      "grad_norm": 0.80859375,
+      "learning_rate": 5.821304160827371e-06,
+      "loss": 0.6397,
+      "step": 338
+    },
+    {
+      "epoch": 0.4556834411492901,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5.80015807227888e-06,
+      "loss": 0.6435,
+      "step": 339
+    },
+    {
+      "epoch": 0.45702764009073343,
+      "grad_norm": 0.80859375,
+      "learning_rate": 5.778997285436654e-06,
+      "loss": 0.6343,
+      "step": 340
+    },
+    {
+      "epoch": 0.45837183903217676,
+      "grad_norm": 0.84375,
+      "learning_rate": 5.757822189008215e-06,
+      "loss": 0.6599,
+      "step": 341
+    },
+    {
+      "epoch": 0.4597160379736201,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5.736633171963937e-06,
+      "loss": 0.6413,
+      "step": 342
+    },
+    {
+      "epoch": 0.4610602369150634,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.71543062352991e-06,
+      "loss": 0.6395,
+      "step": 343
+    },
+    {
+      "epoch": 0.46240443585650676,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5.694214933180784e-06,
+      "loss": 0.6258,
+      "step": 344
+    },
+    {
+      "epoch": 0.4637486347979501,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.672986490632614e-06,
+      "loss": 0.6446,
+      "step": 345
+    },
+    {
+      "epoch": 0.4650928337393934,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.6517456858357065e-06,
+      "loss": 0.6337,
+      "step": 346
+    },
+    {
+      "epoch": 0.46643703268083675,
+      "grad_norm": 0.80859375,
+      "learning_rate": 5.630492908967451e-06,
+      "loss": 0.6534,
+      "step": 347
+    },
+    {
+      "epoch": 0.4677812316222801,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5.609228550425154e-06,
+      "loss": 0.6646,
+      "step": 348
+    },
+    {
+      "epoch": 0.4691254305637234,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.587953000818872e-06,
+      "loss": 0.6372,
+      "step": 349
+    },
+    {
+      "epoch": 0.47046962950516674,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.566666650964229e-06,
+      "loss": 0.6562,
+      "step": 350
+    },
+    {
+      "epoch": 0.4718138284466101,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.5453698918752404e-06,
+      "loss": 0.6673,
+      "step": 351
+    },
+    {
+      "epoch": 0.47315802738805346,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.524063114757139e-06,
+      "loss": 0.6604,
+      "step": 352
+    },
+    {
+      "epoch": 0.4745022263294968,
+      "grad_norm": 0.828125,
+      "learning_rate": 5.50274671099917e-06,
+      "loss": 0.6768,
+      "step": 353
+    },
+    {
+      "epoch": 0.4758464252709401,
+      "grad_norm": 0.85546875,
+      "learning_rate": 5.481421072167424e-06,
+      "loss": 0.6106,
+      "step": 354
+    },
+    {
+      "epoch": 0.47719062421238345,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.460086589997623e-06,
+      "loss": 0.6522,
+      "step": 355
+    },
+    {
+      "epoch": 0.4785348231538268,
+      "grad_norm": 0.84375,
+      "learning_rate": 5.43874365638794e-06,
+      "loss": 0.6482,
+      "step": 356
+    },
+    {
+      "epoch": 0.4798790220952701,
+      "grad_norm": 0.84375,
+      "learning_rate": 5.417392663391796e-06,
+      "loss": 0.6525,
+      "step": 357
+    },
+    {
+      "epoch": 0.48122322103671344,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.3960340032106515e-06,
+      "loss": 0.6455,
+      "step": 358
+    },
+    {
+      "epoch": 0.4825674199781568,
+      "grad_norm": 0.80078125,
+      "learning_rate": 5.374668068186809e-06,
+      "loss": 0.6325,
+      "step": 359
+    },
+    {
+      "epoch": 0.4839116189196001,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.353295250796206e-06,
+      "loss": 0.6651,
+      "step": 360
+    },
+    {
+      "epoch": 0.48525581786104344,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5.331915943641205e-06,
+      "loss": 0.6724,
+      "step": 361
+    },
+    {
+      "epoch": 0.48660001680248677,
+      "grad_norm": 0.828125,
+      "learning_rate": 5.3105305394433745e-06,
+      "loss": 0.6722,
+      "step": 362
+    },
+    {
+      "epoch": 0.4879442157439301,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.28913943103629e-06,
+      "loss": 0.6086,
+      "step": 363
+    },
+    {
+      "epoch": 0.48928841468537343,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.267743011358301e-06,
+      "loss": 0.636,
+      "step": 364
+    },
+    {
+      "epoch": 0.49063261362681676,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5.246341673445323e-06,
+      "loss": 0.6824,
+      "step": 365
+    },
+    {
+      "epoch": 0.4919768125682601,
+      "grad_norm": 0.84375,
+      "learning_rate": 5.2249358104236205e-06,
+      "loss": 0.6846,
+      "step": 366
+    },
+    {
+      "epoch": 0.4933210115097034,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5.203525815502574e-06,
+      "loss": 0.6837,
+      "step": 367
+    },
+    {
+      "epoch": 0.49466521045114675,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.182112081967467e-06,
+      "loss": 0.6567,
+      "step": 368
+    },
+    {
+      "epoch": 0.4960094093925901,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5.160695003172259e-06,
+      "loss": 0.6655,
+      "step": 369
+    },
+    {
+      "epoch": 0.4973536083340334,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5.13927497253236e-06,
+      "loss": 0.6424,
+      "step": 370
+    },
+    {
+      "epoch": 0.49869780727547675,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.1178523835174e-06,
+      "loss": 0.6536,
+      "step": 371
+    },
+    {
+      "epoch": 0.5000420062169201,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.096427629644007e-06,
+      "loss": 0.6441,
+      "step": 372
+    },
+    {
+      "epoch": 0.5013862051583634,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.075001104468576e-06,
+      "loss": 0.6562,
+      "step": 373
+    },
+    {
+      "epoch": 0.5027304040998067,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.053573201580039e-06,
+      "loss": 0.6624,
+      "step": 374
+    },
+    {
+      "epoch": 0.5040746030412501,
+      "grad_norm": 0.828125,
+      "learning_rate": 5.032144314592633e-06,
+      "loss": 0.6581,
+      "step": 375
+    },
+    {
+      "epoch": 0.5054188019826934,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.010714837138675e-06,
+      "loss": 0.6804,
+      "step": 376
+    },
+    {
+      "epoch": 0.5067630009241367,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.989285162861327e-06,
+      "loss": 0.6657,
+      "step": 377
+    },
+    {
+      "epoch": 0.5081071998655801,
+      "grad_norm": 0.83203125,
+      "learning_rate": 4.967855685407368e-06,
+      "loss": 0.6203,
+      "step": 378
+    },
+    {
+      "epoch": 0.5094513988070234,
+      "grad_norm": 0.81640625,
+      "learning_rate": 4.946426798419963e-06,
+      "loss": 0.6388,
+      "step": 379
+    },
+    {
+      "epoch": 0.5107955977484667,
+      "grad_norm": 0.8046875,
+      "learning_rate": 4.924998895531425e-06,
+      "loss": 0.6311,
+      "step": 380
+    },
+    {
+      "epoch": 0.51213979668991,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.903572370355993e-06,
+      "loss": 0.6218,
+      "step": 381
+    },
+    {
+      "epoch": 0.5134839956313534,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.882147616482602e-06,
+      "loss": 0.6695,
+      "step": 382
+    },
+    {
+      "epoch": 0.5148281945727968,
+      "grad_norm": 0.80859375,
+      "learning_rate": 4.860725027467641e-06,
+      "loss": 0.6359,
+      "step": 383
+    },
+    {
+      "epoch": 0.5161723935142402,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.839304996827741e-06,
+      "loss": 0.6733,
+      "step": 384
+    },
+    {
+      "epoch": 0.5175165924556835,
+      "grad_norm": 0.94921875,
+      "learning_rate": 4.817887918032535e-06,
+      "loss": 0.6709,
+      "step": 385
+    },
+    {
+      "epoch": 0.5188607913971268,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.796474184497428e-06,
+      "loss": 0.6456,
+      "step": 386
+    },
+    {
+      "epoch": 0.5202049903385702,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.775064189576381e-06,
+      "loss": 0.6788,
+      "step": 387
+    },
+    {
+      "epoch": 0.5215491892800135,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.7536583265546775e-06,
+      "loss": 0.6346,
+      "step": 388
+    },
+    {
+      "epoch": 0.5228933882214568,
+      "grad_norm": 0.8203125,
+      "learning_rate": 4.732256988641701e-06,
+      "loss": 0.6375,
+      "step": 389
+    },
+    {
+      "epoch": 0.5242375871629001,
+      "grad_norm": 0.859375,
+      "learning_rate": 4.71086056896371e-06,
+      "loss": 0.6887,
+      "step": 390
+    },
+    {
+      "epoch": 0.5255817861043435,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.689469460556626e-06,
+      "loss": 0.6727,
+      "step": 391
+    },
+    {
+      "epoch": 0.5269259850457868,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.668084056358796e-06,
+      "loss": 0.6355,
+      "step": 392
+    },
+    {
+      "epoch": 0.5282701839872301,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.646704749203794e-06,
+      "loss": 0.644,
+      "step": 393
+    },
+    {
+      "epoch": 0.5296143829286735,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.625331931813193e-06,
+      "loss": 0.6344,
+      "step": 394
+    },
+    {
+      "epoch": 0.5309585818701168,
+      "grad_norm": 0.8125,
+      "learning_rate": 4.603965996789349e-06,
+      "loss": 0.6421,
+      "step": 395
+    },
+    {
+      "epoch": 0.5323027808115601,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.582607336608205e-06,
+      "loss": 0.662,
+      "step": 396
+    },
+    {
+      "epoch": 0.5336469797530035,
+      "grad_norm": 0.83203125,
+      "learning_rate": 4.561256343612061e-06,
+      "loss": 0.6751,
+      "step": 397
+    },
+    {
+      "epoch": 0.5349911786944468,
+      "grad_norm": 0.8046875,
+      "learning_rate": 4.539913410002378e-06,
+      "loss": 0.6576,
+      "step": 398
+    },
+    {
+      "epoch": 0.5363353776358901,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.518578927832577e-06,
+      "loss": 0.5937,
+      "step": 399
+    },
+    {
+      "epoch": 0.5376795765773335,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.497253289000831e-06,
+      "loss": 0.6482,
+      "step": 400
+    },
+    {
+      "epoch": 0.5390237755187768,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.475936885242863e-06,
+      "loss": 0.646,
+      "step": 401
+    },
+    {
+      "epoch": 0.5403679744602201,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.45463010812476e-06,
+      "loss": 0.6614,
+      "step": 402
+    },
+    {
+      "epoch": 0.5417121734016634,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.433333349035773e-06,
+      "loss": 0.629,
+      "step": 403
+    },
+    {
+      "epoch": 0.5430563723431068,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.41204699918113e-06,
+      "loss": 0.6608,
+      "step": 404
+    },
+    {
+      "epoch": 0.5444005712845501,
+      "grad_norm": 0.83984375,
+      "learning_rate": 4.390771449574846e-06,
+      "loss": 0.6352,
+      "step": 405
+    },
+    {
+      "epoch": 0.5457447702259934,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.369507091032551e-06,
+      "loss": 0.6474,
+      "step": 406
+    },
+    {
+      "epoch": 0.5470889691674368,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.348254314164294e-06,
+      "loss": 0.6162,
+      "step": 407
+    },
+    {
+      "epoch": 0.5484331681088801,
+      "grad_norm": 0.8046875,
+      "learning_rate": 4.327013509367386e-06,
+      "loss": 0.6688,
+      "step": 408
+    },
+    {
+      "epoch": 0.5497773670503234,
+      "grad_norm": 0.8203125,
+      "learning_rate": 4.305785066819218e-06,
+      "loss": 0.6492,
+      "step": 409
+    },
+    {
+      "epoch": 0.5511215659917668,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.284569376470091e-06,
+      "loss": 0.6739,
+      "step": 410
+    },
+    {
+      "epoch": 0.5524657649332101,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.263366828036065e-06,
+      "loss": 0.6367,
+      "step": 411
+    },
+    {
+      "epoch": 0.5538099638746534,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.242177810991789e-06,
+      "loss": 0.6451,
+      "step": 412
+    },
+    {
+      "epoch": 0.5551541628160968,
+      "grad_norm": 0.859375,
+      "learning_rate": 4.221002714563347e-06,
+      "loss": 0.6406,
+      "step": 413
+    },
+    {
+      "epoch": 0.5564983617575401,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.199841927721121e-06,
+      "loss": 0.6732,
+      "step": 414
+    },
+    {
+      "epoch": 0.5578425606989834,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.178695839172631e-06,
+      "loss": 0.6307,
+      "step": 415
+    },
+    {
+      "epoch": 0.5591867596404267,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.1575648373554e-06,
+      "loss": 0.6583,
+      "step": 416
+    },
+    {
+      "epoch": 0.5605309585818701,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.136449310429822e-06,
+      "loss": 0.6443,
+      "step": 417
+    },
+    {
+      "epoch": 0.5618751575233134,
+      "grad_norm": 0.8203125,
+      "learning_rate": 4.115349646272029e-06,
+      "loss": 0.6657,
+      "step": 418
+    },
+    {
+      "epoch": 0.5632193564647567,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.094266232466763e-06,
+      "loss": 0.6467,
+      "step": 419
+    },
+    {
+      "epoch": 0.5645635554062001,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.0731994563002604e-06,
+      "loss": 0.6445,
+      "step": 420
+    },
+    {
+      "epoch": 0.5659077543476434,
+      "grad_norm": 0.80859375,
+      "learning_rate": 4.052149704753142e-06,
+      "loss": 0.677,
+      "step": 421
+    },
+    {
+      "epoch": 0.5672519532890867,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.03111736449329e-06,
+      "loss": 0.6323,
+      "step": 422
+    },
+    {
+      "epoch": 0.5685961522305302,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.010102821868762e-06,
+      "loss": 0.6582,
+      "step": 423
+    },
+    {
+      "epoch": 0.5699403511719735,
+      "grad_norm": 0.80859375,
+      "learning_rate": 3.989106462900686e-06,
+      "loss": 0.6554,
+      "step": 424
+    },
+    {
+      "epoch": 0.5712845501134168,
+      "grad_norm": 0.80859375,
+      "learning_rate": 3.968128673276164e-06,
+      "loss": 0.6369,
+      "step": 425
+    },
+    {
+      "epoch": 0.5726287490548602,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.947169838341202e-06,
+      "loss": 0.6658,
+      "step": 426
+    },
+    {
+      "epoch": 0.5739729479963035,
+      "grad_norm": 0.85546875,
+      "learning_rate": 3.926230343093617e-06,
+      "loss": 0.6925,
+      "step": 427
+    },
+    {
+      "epoch": 0.5753171469377468,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.90531057217597e-06,
+      "loss": 0.7031,
+      "step": 428
+    },
+    {
+      "epoch": 0.5766613458791902,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.8844109098685045e-06,
+      "loss": 0.6587,
+      "step": 429
+    },
+    {
+      "epoch": 0.5780055448206335,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.863531740082086e-06,
+      "loss": 0.66,
+      "step": 430
+    },
+    {
+      "epoch": 0.5793497437620768,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.8426734463511385e-06,
+      "loss": 0.6507,
+      "step": 431
+    },
+    {
+      "epoch": 0.5806939427035201,
+      "grad_norm": 0.83203125,
+      "learning_rate": 3.82183641182662e-06,
+      "loss": 0.6456,
+      "step": 432
+    },
+    {
+      "epoch": 0.5820381416449635,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.8010210192689688e-06,
+      "loss": 0.6328,
+      "step": 433
+    },
+    {
+      "epoch": 0.5833823405864068,
+      "grad_norm": 0.81640625,
+      "learning_rate": 3.780227651041073e-06,
+      "loss": 0.6614,
+      "step": 434
+    },
+    {
+      "epoch": 0.5847265395278501,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.7594566891012544e-06,
+      "loss": 0.662,
+      "step": 435
+    },
+    {
+      "epoch": 0.5860707384692935,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.738708514996251e-06,
+      "loss": 0.6694,
+      "step": 436
+    },
+    {
+      "epoch": 0.5874149374107368,
+      "grad_norm": 0.80859375,
+      "learning_rate": 3.7179835098541984e-06,
+      "loss": 0.6382,
+      "step": 437
+    },
+    {
+      "epoch": 0.5887591363521801,
+      "grad_norm": 0.8125,
+      "learning_rate": 3.6972820543776404e-06,
+      "loss": 0.6363,
+      "step": 438
+    },
+    {
+      "epoch": 0.5901033352936235,
+      "grad_norm": 0.828125,
+      "learning_rate": 3.676604528836535e-06,
+      "loss": 0.6392,
+      "step": 439
+    },
+    {
+      "epoch": 0.5914475342350668,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.6559513130612567e-06,
+      "loss": 0.6573,
+      "step": 440
+    },
+    {
+      "epoch": 0.5927917331765101,
+      "grad_norm": 0.76953125,
+      "learning_rate": 3.6353227864356354e-06,
+      "loss": 0.6257,
+      "step": 441
+    },
+    {
+      "epoch": 0.5941359321179535,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.614719327889978e-06,
+      "loss": 0.6674,
+      "step": 442
+    },
+    {
+      "epoch": 0.5954801310593968,
+      "grad_norm": 0.859375,
+      "learning_rate": 3.5941413158941086e-06,
+      "loss": 0.6569,
+      "step": 443
+    },
+    {
+      "epoch": 0.5968243300008401,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.573589128450418e-06,
+      "loss": 0.6259,
+      "step": 444
+    },
+    {
+      "epoch": 0.5981685289422835,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.5530631430869235e-06,
+      "loss": 0.6497,
+      "step": 445
+    },
+    {
+      "epoch": 0.5995127278837268,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.5325637368503224e-06,
+      "loss": 0.6382,
+      "step": 446
+    },
+    {
+      "epoch": 0.6008569268251701,
+      "grad_norm": 0.80859375,
+      "learning_rate": 3.5120912862990807e-06,
+      "loss": 0.6879,
+      "step": 447
+    },
+    {
+      "epoch": 0.6022011257666134,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.4916461674965074e-06,
+      "loss": 0.6606,
+      "step": 448
+    },
+    {
+      "epoch": 0.6035453247080568,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.4712287560038447e-06,
+      "loss": 0.637,
+      "step": 449
+    },
+    {
+      "epoch": 0.6048895236495001,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.450839426873378e-06,
+      "loss": 0.6468,
+      "step": 450
+    },
+    {
+      "epoch": 0.6062337225909434,
+      "grad_norm": 0.8203125,
+      "learning_rate": 3.4304785546415376e-06,
+      "loss": 0.671,
+      "step": 451
+    },
+    {
+      "epoch": 0.6075779215323868,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.41014651332202e-06,
+      "loss": 0.658,
+      "step": 452
+    },
+    {
+      "epoch": 0.6089221204738301,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.389843676398925e-06,
+      "loss": 0.6747,
+      "step": 453
+    },
+    {
+      "epoch": 0.6102663194152734,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.3695704168198895e-06,
+      "loss": 0.6783,
+      "step": 454
+    },
+    {
+      "epoch": 0.6116105183567168,
+      "grad_norm": 0.859375,
+      "learning_rate": 3.349327106989232e-06,
+      "loss": 0.6662,
+      "step": 455
+    },
+    {
+      "epoch": 0.6129547172981601,
+      "grad_norm": 0.8046875,
+      "learning_rate": 3.3291141187611233e-06,
+      "loss": 0.642,
+      "step": 456
+    },
+    {
+      "epoch": 0.6142989162396034,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.3089318234327446e-06,
+      "loss": 0.6241,
+      "step": 457
+    },
+    {
+      "epoch": 0.6156431151810468,
+      "grad_norm": 0.75,
+      "learning_rate": 3.2887805917374736e-06,
+      "loss": 0.6397,
+      "step": 458
+    },
+    {
+      "epoch": 0.6169873141224901,
+      "grad_norm": 0.8125,
+      "learning_rate": 3.2686607938380736e-06,
+      "loss": 0.6485,
+      "step": 459
+    },
+    {
+      "epoch": 0.6183315130639334,
+      "grad_norm": 0.9375,
+      "learning_rate": 3.2485727993198945e-06,
+      "loss": 0.6542,
+      "step": 460
+    },
+    {
+      "epoch": 0.6196757120053767,
+      "grad_norm": 0.82421875,
+      "learning_rate": 3.2285169771840752e-06,
+      "loss": 0.6461,
+      "step": 461
+    },
+    {
+      "epoch": 0.6210199109468201,
+      "grad_norm": 0.81640625,
+      "learning_rate": 3.208493695840781e-06,
+      "loss": 0.6646,
+      "step": 462
+    },
+    {
+      "epoch": 0.6223641098882635,
+      "grad_norm": 0.77734375,
+      "learning_rate": 3.1885033231024253e-06,
+      "loss": 0.6265,
+      "step": 463
+    },
+    {
+      "epoch": 0.6237083088297068,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.1685462261769105e-06,
+      "loss": 0.6475,
+      "step": 464
+    },
+    {
+      "epoch": 0.6250525077711502,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.148622771660895e-06,
+      "loss": 0.6892,
+      "step": 465
+    },
+    {
+      "epoch": 0.6263967067125935,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.128733325533047e-06,
+      "loss": 0.6632,
+      "step": 466
+    },
+    {
+      "epoch": 0.6277409056540368,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.1088782531473268e-06,
+      "loss": 0.6329,
+      "step": 467
+    },
+    {
+      "epoch": 0.6290851045954802,
+      "grad_norm": 0.84375,
+      "learning_rate": 3.089057919226277e-06,
+      "loss": 0.6611,
+      "step": 468
+    },
+    {
+      "epoch": 0.6304293035369235,
+      "grad_norm": 0.828125,
+      "learning_rate": 3.069272687854322e-06,
+      "loss": 0.6494,
+      "step": 469
+    },
+    {
+      "epoch": 0.6317735024783668,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.0495229224710754e-06,
+      "loss": 0.6206,
+      "step": 470
+    },
+    {
+      "epoch": 0.6331177014198102,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.02980898586467e-06,
+      "loss": 0.6415,
+      "step": 471
+    },
+    {
+      "epoch": 0.6344619003612535,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.010131240165094e-06,
+      "loss": 0.6442,
+      "step": 472
+    },
+    {
+      "epoch": 0.6358060993026968,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.9904900468375298e-06,
+      "loss": 0.6449,
+      "step": 473
+    },
+    {
+      "epoch": 0.6371502982441402,
+      "grad_norm": 0.79296875,
+      "learning_rate": 2.9708857666757244e-06,
+      "loss": 0.6277,
+      "step": 474
+    },
+    {
+      "epoch": 0.6384944971855835,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.951318759795361e-06,
+      "loss": 0.635,
+      "step": 475
+    },
+    {
+      "epoch": 0.6398386961270268,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.9317893856274333e-06,
+      "loss": 0.6666,
+      "step": 476
+    },
+    {
+      "epoch": 0.6411828950684701,
+      "grad_norm": 0.82421875,
+      "learning_rate": 2.912298002911659e-06,
+      "loss": 0.656,
+      "step": 477
+    },
+    {
+      "epoch": 0.6425270940099135,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.892844969689876e-06,
+      "loss": 0.6807,
+      "step": 478
+    },
+    {
+      "epoch": 0.6438712929513568,
+      "grad_norm": 0.8046875,
+      "learning_rate": 2.8734306432994733e-06,
+      "loss": 0.6521,
+      "step": 479
+    },
+    {
+      "epoch": 0.6452154918928001,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.8540553803668252e-06,
+      "loss": 0.6573,
+      "step": 480
+    },
+    {
+      "epoch": 0.6465596908342435,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.834719536800742e-06,
+      "loss": 0.6376,
+      "step": 481
+    },
+    {
+      "epoch": 0.6479038897756868,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.815423467785925e-06,
+      "loss": 0.6733,
+      "step": 482
+    },
+    {
+      "epoch": 0.6492480887171301,
+      "grad_norm": 0.8359375,
+      "learning_rate": 2.7961675277764498e-06,
+      "loss": 0.65,
+      "step": 483
+    },
+    {
+      "epoch": 0.6505922876585735,
+      "grad_norm": 0.82421875,
+      "learning_rate": 2.7769520704892566e-06,
+      "loss": 0.6501,
+      "step": 484
+    },
+    {
+      "epoch": 0.6519364866000168,
+      "grad_norm": 0.84765625,
+      "learning_rate": 2.757777448897646e-06,
+      "loss": 0.655,
+      "step": 485
+    },
+    {
+      "epoch": 0.6532806855414601,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.7386440152247933e-06,
+      "loss": 0.6156,
+      "step": 486
+    },
+    {
+      "epoch": 0.6546248844829035,
+      "grad_norm": 0.8125,
+      "learning_rate": 2.71955212093729e-06,
+      "loss": 0.645,
+      "step": 487
+    },
+    {
+      "epoch": 0.6559690834243468,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.7005021167386804e-06,
+      "loss": 0.642,
+      "step": 488
+    },
+    {
+      "epoch": 0.6573132823657901,
+      "grad_norm": 0.82421875,
+      "learning_rate": 2.6814943525630134e-06,
+      "loss": 0.6601,
+      "step": 489
+    },
+    {
+      "epoch": 0.6586574813072334,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.662529177568429e-06,
+      "loss": 0.6241,
+      "step": 490
+    },
+    {
+      "epoch": 0.6600016802486768,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.6436069401307287e-06,
+      "loss": 0.6447,
+      "step": 491
+    },
+    {
+      "epoch": 0.6613458791901201,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.6247279878369914e-06,
+      "loss": 0.6519,
+      "step": 492
+    },
+    {
+      "epoch": 0.6626900781315634,
+      "grad_norm": 0.8359375,
+      "learning_rate": 2.605892667479173e-06,
+      "loss": 0.658,
+      "step": 493
+    },
+    {
+      "epoch": 0.6640342770730068,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.587101325047753e-06,
+      "loss": 0.638,
+      "step": 494
+    },
+    {
+      "epoch": 0.6653784760144501,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.56835430572536e-06,
+      "loss": 0.6535,
+      "step": 495
+    },
+    {
+      "epoch": 0.6667226749558934,
+      "grad_norm": 0.7890625,
+      "learning_rate": 2.549651953880449e-06,
+      "loss": 0.6543,
+      "step": 496
+    },
+    {
+      "epoch": 0.6667226749558934,
+      "eval_loss": 2.7381093502044678,
+      "eval_runtime": 5731.7858,
+      "eval_samples_per_second": 3.863,
+      "eval_steps_per_second": 0.483,
+      "step": 496
+    },
+    {
+      "epoch": 0.6680668738973368,
+      "grad_norm": 0.8125,
+      "learning_rate": 2.5309946130609654e-06,
+      "loss": 0.6406,
+      "step": 497
+    },
+    {
+      "epoch": 0.6694110728387801,
+      "grad_norm": 0.79296875,
+      "learning_rate": 2.5123826259880324e-06,
+      "loss": 0.6338,
+      "step": 498
+    },
+    {
+      "epoch": 0.6707552717802234,
+      "grad_norm": 0.7578125,
+      "learning_rate": 2.493816334549664e-06,
+      "loss": 0.616,
+      "step": 499
+    },
+    {
+      "epoch": 0.6720994707216668,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.47529607979448e-06,
+      "loss": 0.6364,
+      "step": 500
+    },
+    {
+      "epoch": 0.6734436696631101,
+      "grad_norm": 0.7890625,
+      "learning_rate": 2.456822201925438e-06,
+      "loss": 0.6448,
+      "step": 501
+    },
+    {
+      "epoch": 0.6747878686045534,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.43839504029359e-06,
+      "loss": 0.6149,
+      "step": 502
+    },
+    {
+      "epoch": 0.6761320675459969,
+      "grad_norm": 0.8046875,
+      "learning_rate": 2.420014933391849e-06,
+      "loss": 0.6309,
+      "step": 503
+    },
+    {
+      "epoch": 0.6774762664874402,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.4016822188487604e-06,
+      "loss": 0.6276,
+      "step": 504
+    },
+    {
+      "epoch": 0.6788204654288835,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.3833972334223183e-06,
+      "loss": 0.6373,
+      "step": 505
+    },
+    {
+      "epoch": 0.6801646643703269,
+      "grad_norm": 0.79296875,
+      "learning_rate": 2.3651603129937597e-06,
+      "loss": 0.6238,
+      "step": 506
+    },
+    {
+      "epoch": 0.6815088633117702,
+      "grad_norm": 0.8046875,
+      "learning_rate": 2.3469717925614133e-06,
+      "loss": 0.6265,
+      "step": 507
+    },
+    {
+      "epoch": 0.6828530622532135,
+      "grad_norm": 0.83984375,
+      "learning_rate": 2.3288320062345276e-06,
+      "loss": 0.6804,
+      "step": 508
+    },
+    {
+      "epoch": 0.6841972611946568,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.310741287227152e-06,
+      "loss": 0.6436,
+      "step": 509
+    },
+    {
+      "epoch": 0.6855414601361002,
+      "grad_norm": 0.82421875,
+      "learning_rate": 2.2926999678519975e-06,
+      "loss": 0.6501,
+      "step": 510
+    },
+    {
+      "epoch": 0.6868856590775435,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.274708379514348e-06,
+      "loss": 0.6544,
+      "step": 511
+    },
+    {
+      "epoch": 0.6882298580189868,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.256766852705967e-06,
+      "loss": 0.6254,
+      "step": 512
+    },
+    {
+      "epoch": 0.6895740569604302,
+      "grad_norm": 0.82421875,
+      "learning_rate": 2.238875716999019e-06,
+      "loss": 0.6157,
+      "step": 513
+    },
+    {
+      "epoch": 0.6909182559018735,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.221035301040027e-06,
+      "loss": 0.637,
+      "step": 514
+    },
+    {
+      "epoch": 0.6922624548433168,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.203245932543834e-06,
+      "loss": 0.646,
+      "step": 515
+    },
+    {
+      "epoch": 0.6936066537847602,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.185507938287572e-06,
+      "loss": 0.6492,
+      "step": 516
+    },
+    {
+      "epoch": 0.6949508527262035,
+      "grad_norm": 0.90625,
+      "learning_rate": 2.1678216441046733e-06,
+      "loss": 0.6379,
+      "step": 517
+    },
+    {
+      "epoch": 0.6962950516676468,
+      "grad_norm": 0.859375,
+      "learning_rate": 2.1501873748788804e-06,
+      "loss": 0.6383,
+      "step": 518
+    },
+    {
+      "epoch": 0.6976392506090902,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.1326054545382698e-06,
+      "loss": 0.6593,
+      "step": 519
+    },
+    {
+      "epoch": 0.6989834495505335,
+      "grad_norm": 0.8046875,
+      "learning_rate": 2.1150762060493157e-06,
+      "loss": 0.6555,
+      "step": 520
+    },
+    {
+      "epoch": 0.7003276484919768,
+      "grad_norm": 0.8125,
+      "learning_rate": 2.0975999514109503e-06,
+      "loss": 0.6482,
+      "step": 521
+    },
+    {
+      "epoch": 0.7016718474334201,
+      "grad_norm": 0.83203125,
+      "learning_rate": 2.080177011648645e-06,
+      "loss": 0.6926,
+      "step": 522
+    },
+    {
+      "epoch": 0.7030160463748635,
+      "grad_norm": 0.83203125,
+      "learning_rate": 2.0628077068085173e-06,
+      "loss": 0.6603,
+      "step": 523
+    },
+    {
+      "epoch": 0.7043602453163068,
+      "grad_norm": 0.83984375,
+      "learning_rate": 2.0454923559514595e-06,
+      "loss": 0.6679,
+      "step": 524
+    },
+    {
+      "epoch": 0.7057044442577501,
+      "grad_norm": 0.79296875,
+      "learning_rate": 2.028231277147261e-06,
+      "loss": 0.6322,
+      "step": 525
+    },
+    {
+      "epoch": 0.7070486431991935,
+      "grad_norm": 0.78515625,
+      "learning_rate": 2.0110247874687816e-06,
+      "loss": 0.6446,
+      "step": 526
+    },
+    {
+      "epoch": 0.7083928421406368,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.9938732029861192e-06,
+      "loss": 0.629,
+      "step": 527
+    },
+    {
+      "epoch": 0.7097370410820801,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.976776838760801e-06,
+      "loss": 0.6586,
+      "step": 528
+    },
+    {
+      "epoch": 0.7110812400235235,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.9597360088400053e-06,
+      "loss": 0.6929,
+      "step": 529
+    },
+    {
+      "epoch": 0.7124254389649668,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.9427510262507864e-06,
+      "loss": 0.635,
+      "step": 530
+    },
+    {
+      "epoch": 0.7137696379064101,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.9258222029943228e-06,
+      "loss": 0.653,
+      "step": 531
+    },
+    {
+      "epoch": 0.7151138368478535,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.9089498500401915e-06,
+      "loss": 0.6295,
+      "step": 532
+    },
+    {
+      "epoch": 0.7164580357892968,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.8921342773206553e-06,
+      "loss": 0.6375,
+      "step": 533
+    },
+    {
+      "epoch": 0.7178022347307401,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.87537579372496e-06,
+      "loss": 0.6327,
+      "step": 534
+    },
+    {
+      "epoch": 0.7191464336721834,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.858674707093675e-06,
+      "loss": 0.6597,
+      "step": 535
+    },
+    {
+      "epoch": 0.7204906326136268,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.8420313242130294e-06,
+      "loss": 0.6585,
+      "step": 536
+    },
+    {
+      "epoch": 0.7218348315550701,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.825445950809277e-06,
+      "loss": 0.6267,
+      "step": 537
+    },
+    {
+      "epoch": 0.7231790304965134,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.8089188915430794e-06,
+      "loss": 0.6727,
+      "step": 538
+    },
+    {
+      "epoch": 0.7245232294379568,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.7924504500039193e-06,
+      "loss": 0.653,
+      "step": 539
+    },
+    {
+      "epoch": 0.7258674283794001,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.7760409287045076e-06,
+      "loss": 0.642,
+      "step": 540
+    },
+    {
+      "epoch": 0.7272116273208434,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.7596906290752424e-06,
+      "loss": 0.6479,
+      "step": 541
+    },
+    {
+      "epoch": 0.7285558262622868,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.7433998514586631e-06,
+      "loss": 0.6649,
+      "step": 542
+    },
+    {
+      "epoch": 0.7299000252037302,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.7271688951039312e-06,
+      "loss": 0.6754,
+      "step": 543
+    },
+    {
+      "epoch": 0.7312442241451735,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.7109980581613417e-06,
+      "loss": 0.632,
+      "step": 544
+    },
+    {
+      "epoch": 0.7325884230866169,
+      "grad_norm": 0.76171875,
+      "learning_rate": 1.694887637676842e-06,
+      "loss": 0.5958,
+      "step": 545
+    },
+    {
+      "epoch": 0.7339326220280602,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.6788379295865703e-06,
+      "loss": 0.6313,
+      "step": 546
+    },
+    {
+      "epoch": 0.7352768209695035,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.6628492287114296e-06,
+      "loss": 0.6562,
+      "step": 547
+    },
+    {
+      "epoch": 0.7366210199109469,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.6469218287516664e-06,
+      "loss": 0.6277,
+      "step": 548
+    },
+    {
+      "epoch": 0.7379652188523902,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.6310560222814714e-06,
+      "loss": 0.6473,
+      "step": 549
+    },
+    {
+      "epoch": 0.7393094177938335,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.6152521007436146e-06,
+      "loss": 0.6446,
+      "step": 550
+    },
+    {
+      "epoch": 0.7406536167352769,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.599510354444087e-06,
+      "loss": 0.6552,
+      "step": 551
+    },
+    {
+      "epoch": 0.7419978156767202,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.5838310725467643e-06,
+      "loss": 0.6373,
+      "step": 552
+    },
+    {
+      "epoch": 0.7433420146181635,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.568214543068103e-06,
+      "loss": 0.6516,
+      "step": 553
+    },
+    {
+      "epoch": 0.7446862135596068,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.5526610528718416e-06,
+      "loss": 0.6185,
+      "step": 554
+    },
+    {
+      "epoch": 0.7460304125010502,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.5371708876637353e-06,
+      "loss": 0.691,
+      "step": 555
+    },
+    {
+      "epoch": 0.7473746114424935,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.5217443319863112e-06,
+      "loss": 0.6475,
+      "step": 556
+    },
+    {
+      "epoch": 0.7487188103839368,
+      "grad_norm": 0.7578125,
+      "learning_rate": 1.5063816692136374e-06,
+      "loss": 0.6037,
+      "step": 557
+    },
+    {
+      "epoch": 0.7500630093253802,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.4910831815461125e-06,
+      "loss": 0.6247,
+      "step": 558
+    },
+    {
+      "epoch": 0.7514072082668235,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.4758491500052924e-06,
+      "loss": 0.6306,
+      "step": 559
+    },
+    {
+      "epoch": 0.7527514072082668,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.4606798544287243e-06,
+      "loss": 0.6729,
+      "step": 560
+    },
+    {
+      "epoch": 0.7540956061497102,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.4455755734647992e-06,
+      "loss": 0.6456,
+      "step": 561
+    },
+    {
+      "epoch": 0.7554398050911535,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.4305365845676438e-06,
+      "loss": 0.6482,
+      "step": 562
+    },
+    {
+      "epoch": 0.7567840040325968,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.4155631639920208e-06,
+      "loss": 0.6465,
+      "step": 563
+    },
+    {
+      "epoch": 0.7581282029740402,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.4006555867882466e-06,
+      "loss": 0.6638,
+      "step": 564
+    },
+    {
+      "epoch": 0.7594724019154835,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.3858141267971492e-06,
+      "loss": 0.6282,
+      "step": 565
+    },
+    {
+      "epoch": 0.7608166008569268,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.3710390566450366e-06,
+      "loss": 0.6749,
+      "step": 566
+    },
+    {
+      "epoch": 0.7621607997983701,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.3563306477386783e-06,
+      "loss": 0.655,
+      "step": 567
+    },
+    {
+      "epoch": 0.7635049987398135,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.3416891702603357e-06,
+      "loss": 0.6596,
+      "step": 568
+    },
+    {
+      "epoch": 0.7648491976812568,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.3271148931627859e-06,
+      "loss": 0.6813,
+      "step": 569
+    },
+    {
+      "epoch": 0.7661933966227001,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.3126080841643856e-06,
+      "loss": 0.6312,
+      "step": 570
+    },
+    {
+      "epoch": 0.7675375955641435,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.2981690097441574e-06,
+      "loss": 0.6442,
+      "step": 571
+    },
+    {
+      "epoch": 0.7688817945055868,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.2837979351368912e-06,
+      "loss": 0.6484,
+      "step": 572
+    },
+    {
+      "epoch": 0.7702259934470301,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.2694951243282682e-06,
+      "loss": 0.6373,
+      "step": 573
+    },
+    {
+      "epoch": 0.7715701923884735,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.25526084005002e-06,
+      "loss": 0.6318,
+      "step": 574
+    },
+    {
+      "epoch": 0.7729143913299168,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.2410953437750966e-06,
+      "loss": 0.6471,
+      "step": 575
+    },
+    {
+      "epoch": 0.7742585902713601,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.2269988957128638e-06,
+      "loss": 0.6353,
+      "step": 576
+    },
+    {
+      "epoch": 0.7756027892128035,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.212971754804324e-06,
+      "loss": 0.6639,
+      "step": 577
+    },
+    {
+      "epoch": 0.7769469881542468,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.1990141787173648e-06,
+      "loss": 0.6428,
+      "step": 578
+    },
+    {
+      "epoch": 0.7782911870956901,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1851264238420136e-06,
+      "loss": 0.6572,
+      "step": 579
+    },
+    {
+      "epoch": 0.7796353860371334,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.1713087452857408e-06,
+      "loss": 0.6291,
+      "step": 580
+    },
+    {
+      "epoch": 0.7809795849785768,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.1575613968687683e-06,
+      "loss": 0.6529,
+      "step": 581
+    },
+    {
+      "epoch": 0.7823237839200201,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.1438846311194024e-06,
+      "loss": 0.6679,
+      "step": 582
+    },
+    {
+      "epoch": 0.7836679828614634,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1302786992694049e-06,
+      "loss": 0.6377,
+      "step": 583
+    },
+    {
+      "epoch": 0.7850121818029069,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.1167438512493684e-06,
+      "loss": 0.637,
+      "step": 584
+    },
+    {
+      "epoch": 0.7863563807443502,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1032803356841343e-06,
+      "loss": 0.6484,
+      "step": 585
+    },
+    {
+      "epoch": 0.7877005796857935,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.089888399888216e-06,
+      "loss": 0.6971,
+      "step": 586
+    },
+    {
+      "epoch": 0.7890447786272369,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.0765682898612657e-06,
+      "loss": 0.6715,
+      "step": 587
+    },
+    {
+      "epoch": 0.7903889775686802,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.0633202502835494e-06,
+      "loss": 0.6915,
+      "step": 588
+    },
+    {
+      "epoch": 0.7917331765101235,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.0501445245114523e-06,
+      "loss": 0.6216,
+      "step": 589
+    },
+    {
+      "epoch": 0.7930773754515669,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.0370413545730117e-06,
+      "loss": 0.6246,
+      "step": 590
+    },
+    {
+      "epoch": 0.7944215743930102,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.0240109811634714e-06,
+      "loss": 0.6232,
+      "step": 591
+    },
+    {
+      "epoch": 0.7957657733344535,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.0110536436408535e-06,
+      "loss": 0.6392,
+      "step": 592
+    },
+    {
+      "epoch": 0.7971099722758969,
+      "grad_norm": 0.8125,
+      "learning_rate": 9.981695800215701e-07,
+      "loss": 0.6428,
+      "step": 593
+    },
+    {
+      "epoch": 0.7984541712173402,
+      "grad_norm": 0.8046875,
+      "learning_rate": 9.853590269760493e-07,
+      "loss": 0.6712,
+      "step": 594
+    },
+    {
+      "epoch": 0.7997983701587835,
+      "grad_norm": 0.78125,
+      "learning_rate": 9.726222198243806e-07,
+      "loss": 0.6173,
+      "step": 595
+    },
+    {
+      "epoch": 0.8011425691002269,
+      "grad_norm": 0.78515625,
+      "learning_rate": 9.599593925320017e-07,
+      "loss": 0.6397,
+      "step": 596
+    },
+    {
+      "epoch": 0.8024867680416702,
+      "grad_norm": 0.84765625,
+      "learning_rate": 9.47370777705397e-07,
+      "loss": 0.6875,
+      "step": 597
+    },
+    {
+      "epoch": 0.8038309669831135,
+      "grad_norm": 0.78125,
+      "learning_rate": 9.348566065878218e-07,
+      "loss": 0.6455,
+      "step": 598
+    },
+    {
+      "epoch": 0.8051751659245568,
+      "grad_norm": 0.76953125,
+      "learning_rate": 9.224171090550571e-07,
+      "loss": 0.6325,
+      "step": 599
+    },
+    {
+      "epoch": 0.8065193648660002,
+      "grad_norm": 0.8046875,
+      "learning_rate": 9.100525136111915e-07,
+      "loss": 0.6663,
+      "step": 600
+    },
+    {
+      "epoch": 0.8078635638074435,
+      "grad_norm": 0.7578125,
+      "learning_rate": 8.97763047384414e-07,
+      "loss": 0.6358,
+      "step": 601
+    },
+    {
+      "epoch": 0.8092077627488868,
+      "grad_norm": 0.79296875,
+      "learning_rate": 8.855489361228497e-07,
+      "loss": 0.6556,
+      "step": 602
+    },
+    {
+      "epoch": 0.8105519616903302,
+      "grad_norm": 0.796875,
+      "learning_rate": 8.734104041904129e-07,
+      "loss": 0.6613,
+      "step": 603
+    },
+    {
+      "epoch": 0.8118961606317735,
+      "grad_norm": 0.81640625,
+      "learning_rate": 8.613476745626769e-07,
+      "loss": 0.6688,
+      "step": 604
+    },
+    {
+      "epoch": 0.8132403595732168,
+      "grad_norm": 0.78515625,
+      "learning_rate": 8.4936096882279e-07,
+      "loss": 0.637,
+      "step": 605
+    },
+    {
+      "epoch": 0.8145845585146602,
+      "grad_norm": 0.7578125,
+      "learning_rate": 8.374505071573991e-07,
+      "loss": 0.6512,
+      "step": 606
+    },
+    {
+      "epoch": 0.8159287574561035,
+      "grad_norm": 0.8046875,
+      "learning_rate": 8.256165083526019e-07,
+      "loss": 0.6522,
+      "step": 607
+    },
+    {
+      "epoch": 0.8172729563975468,
+      "grad_norm": 0.765625,
+      "learning_rate": 8.138591897899345e-07,
+      "loss": 0.621,
+      "step": 608
+    },
+    {
+      "epoch": 0.8186171553389902,
+      "grad_norm": 0.8203125,
+      "learning_rate": 8.021787674423775e-07,
+      "loss": 0.6703,
+      "step": 609
+    },
+    {
+      "epoch": 0.8199613542804335,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.905754558703805e-07,
+      "loss": 0.6519,
+      "step": 610
+    },
+    {
+      "epoch": 0.8213055532218768,
+      "grad_norm": 0.8203125,
+      "learning_rate": 7.790494682179317e-07,
+      "loss": 0.6565,
+      "step": 611
+    },
+    {
+      "epoch": 0.8226497521633201,
+      "grad_norm": 0.81640625,
+      "learning_rate": 7.676010162086389e-07,
+      "loss": 0.6753,
+      "step": 612
+    },
+    {
+      "epoch": 0.8239939511047635,
+      "grad_norm": 0.8046875,
+      "learning_rate": 7.56230310141835e-07,
+      "loss": 0.6604,
+      "step": 613
+    },
+    {
+      "epoch": 0.8253381500462068,
+      "grad_norm": 0.78515625,
+      "learning_rate": 7.449375588887203e-07,
+      "loss": 0.6374,
+      "step": 614
+    },
+    {
+      "epoch": 0.8266823489876501,
+      "grad_norm": 0.796875,
+      "learning_rate": 7.337229698885279e-07,
+      "loss": 0.6595,
+      "step": 615
+    },
+    {
+      "epoch": 0.8280265479290935,
+      "grad_norm": 0.78515625,
+      "learning_rate": 7.225867491447053e-07,
+      "loss": 0.6368,
+      "step": 616
+    },
+    {
+      "epoch": 0.8293707468705368,
+      "grad_norm": 0.8125,
+      "learning_rate": 7.115291012211384e-07,
+      "loss": 0.63,
+      "step": 617
+    },
+    {
+      "epoch": 0.8307149458119801,
+      "grad_norm": 0.81640625,
+      "learning_rate": 7.005502292383898e-07,
+      "loss": 0.6361,
+      "step": 618
+    },
+    {
+      "epoch": 0.8320591447534235,
+      "grad_norm": 0.78125,
+      "learning_rate": 6.896503348699657e-07,
+      "loss": 0.6454,
+      "step": 619
+    },
+    {
+      "epoch": 0.8334033436948668,
+      "grad_norm": 0.8125,
+      "learning_rate": 6.788296183386162e-07,
+      "loss": 0.6877,
+      "step": 620
+    },
+    {
+      "epoch": 0.8347475426363101,
+      "grad_norm": 0.76171875,
+      "learning_rate": 6.680882784126552e-07,
+      "loss": 0.6173,
+      "step": 621
+    },
+    {
+      "epoch": 0.8360917415777535,
+      "grad_norm": 0.796875,
+      "learning_rate": 6.574265124023055e-07,
+      "loss": 0.6452,
+      "step": 622
+    },
+    {
+      "epoch": 0.8374359405191968,
+      "grad_norm": 0.82421875,
+      "learning_rate": 6.46844516156081e-07,
+      "loss": 0.6147,
+      "step": 623
+    },
+    {
+      "epoch": 0.8387801394606402,
+      "grad_norm": 0.765625,
+      "learning_rate": 6.363424840571869e-07,
+      "loss": 0.6646,
+      "step": 624
+    },
+    {
+      "epoch": 0.8401243384020836,
+      "grad_norm": 0.83203125,
+      "learning_rate": 6.259206090199427e-07,
+      "loss": 0.6547,
+      "step": 625
+    },
+    {
+      "epoch": 0.8414685373435269,
+      "grad_norm": 0.80078125,
+      "learning_rate": 6.155790824862484e-07,
+      "loss": 0.6231,
+      "step": 626
+    },
+    {
+      "epoch": 0.8428127362849702,
+      "grad_norm": 0.80078125,
+      "learning_rate": 6.053180944220627e-07,
+      "loss": 0.6684,
+      "step": 627
+    },
+    {
+      "epoch": 0.8441569352264136,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.951378333139118e-07,
+      "loss": 0.6465,
+      "step": 628
+    },
+    {
+      "epoch": 0.8455011341678569,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5.850384861654329e-07,
+      "loss": 0.6451,
+      "step": 629
+    },
+    {
+      "epoch": 0.8468453331093002,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.750202384939313e-07,
+      "loss": 0.6367,
+      "step": 630
+    },
+    {
+      "epoch": 0.8481895320507435,
+      "grad_norm": 0.7734375,
+      "learning_rate": 5.65083274326978e-07,
+      "loss": 0.6602,
+      "step": 631
+    },
+    {
+      "epoch": 0.8495337309921869,
+      "grad_norm": 0.80078125,
+      "learning_rate": 5.552277761990293e-07,
+      "loss": 0.6587,
+      "step": 632
+    },
+    {
+      "epoch": 0.8508779299336302,
+      "grad_norm": 0.78125,
+      "learning_rate": 5.454539251480739e-07,
+      "loss": 0.6605,
+      "step": 633
+    },
+    {
+      "epoch": 0.8522221288750735,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5.35761900712301e-07,
+      "loss": 0.6672,
+      "step": 634
+    },
+    {
+      "epoch": 0.8535663278165169,
+      "grad_norm": 0.78125,
+      "learning_rate": 5.261518809268118e-07,
+      "loss": 0.6428,
+      "step": 635
+    },
+    {
+      "epoch": 0.8549105267579602,
+      "grad_norm": 0.7890625,
+      "learning_rate": 5.166240423203428e-07,
+      "loss": 0.6359,
+      "step": 636
+    },
+    {
+      "epoch": 0.8562547256994035,
+      "grad_norm": 0.80078125,
+      "learning_rate": 5.071785599120243e-07,
+      "loss": 0.6562,
+      "step": 637
+    },
+    {
+      "epoch": 0.8575989246408469,
+      "grad_norm": 0.91015625,
+      "learning_rate": 4.978156072081669e-07,
+      "loss": 0.6627,
+      "step": 638
+    },
+    {
+      "epoch": 0.8589431235822902,
+      "grad_norm": 0.8046875,
+      "learning_rate": 4.885353561990752e-07,
+      "loss": 0.6588,
+      "step": 639
+    },
+    {
+      "epoch": 0.8602873225237335,
+      "grad_norm": 0.8203125,
+      "learning_rate": 4.793379773558816e-07,
+      "loss": 0.6279,
+      "step": 640
+    },
+    {
+      "epoch": 0.8616315214651769,
+      "grad_norm": 0.86328125,
+      "learning_rate": 4.702236396274251e-07,
+      "loss": 0.662,
+      "step": 641
+    },
+    {
+      "epoch": 0.8629757204066202,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.611925104371423e-07,
+      "loss": 0.6556,
+      "step": 642
+    },
+    {
+      "epoch": 0.8643199193480635,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.522447556799875e-07,
+      "loss": 0.6417,
+      "step": 643
+    },
+    {
+      "epoch": 0.8656641182895068,
+      "grad_norm": 0.81640625,
+      "learning_rate": 4.433805397193969e-07,
+      "loss": 0.6636,
+      "step": 644
+    },
+    {
+      "epoch": 0.8670083172309502,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.3460002538425806e-07,
+      "loss": 0.6509,
+      "step": 645
+    },
+    {
+      "epoch": 0.8683525161723935,
+      "grad_norm": 0.8046875,
+      "learning_rate": 4.2590337396592406e-07,
+      "loss": 0.6481,
+      "step": 646
+    },
+    {
+      "epoch": 0.8696967151138368,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.1729074521525193e-07,
+      "loss": 0.6307,
+      "step": 647
+    },
+    {
+      "epoch": 0.8710409140552802,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.0876229733966655e-07,
+      "loss": 0.626,
+      "step": 648
+    },
+    {
+      "epoch": 0.8723851129967235,
+      "grad_norm": 0.8046875,
+      "learning_rate": 4.0031818700025095e-07,
+      "loss": 0.6414,
+      "step": 649
+    },
+    {
+      "epoch": 0.8737293119381668,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.919585693088751e-07,
+      "loss": 0.6427,
+      "step": 650
+    },
+    {
+      "epoch": 0.8750735108796102,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.836835978253434e-07,
+      "loss": 0.6505,
+      "step": 651
+    },
+    {
+      "epoch": 0.8764177098210535,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.754934245545722e-07,
+      "loss": 0.6691,
+      "step": 652
+    },
+    {
+      "epoch": 0.8777619087624968,
+      "grad_norm": 0.7890625,
+      "learning_rate": 3.6738819994379946e-07,
+      "loss": 0.6116,
+      "step": 653
+    },
+    {
+      "epoch": 0.8791061077039402,
+      "grad_norm": 0.82421875,
+      "learning_rate": 3.593680728798238e-07,
+      "loss": 0.6865,
+      "step": 654
+    },
+    {
+      "epoch": 0.8804503066453835,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.514331906862622e-07,
+      "loss": 0.6743,
+      "step": 655
+    },
+    {
+      "epoch": 0.8817945055868268,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.435836991208524e-07,
+      "loss": 0.6521,
+      "step": 656
+    },
+    {
+      "epoch": 0.8831387045282701,
+      "grad_norm": 0.81640625,
+      "learning_rate": 3.3581974237277106e-07,
+      "loss": 0.6471,
+      "step": 657
+    },
+    {
+      "epoch": 0.8844829034697135,
+      "grad_norm": 0.81640625,
+      "learning_rate": 3.281414630599811e-07,
+      "loss": 0.6595,
+      "step": 658
+    },
+    {
+      "epoch": 0.8858271024111568,
+      "grad_norm": 0.8046875,
+      "learning_rate": 3.205490022266228e-07,
+      "loss": 0.6279,
+      "step": 659
+    },
+    {
+      "epoch": 0.8871713013526001,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.130424993404102e-07,
+      "loss": 0.6854,
+      "step": 660
+    },
+    {
+      "epoch": 0.8885155002940435,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.0562209229008043e-07,
+      "loss": 0.6259,
+      "step": 661
+    },
+    {
+      "epoch": 0.8898596992354868,
+      "grad_norm": 0.7890625,
+      "learning_rate": 2.9828791738285235e-07,
+      "loss": 0.6296,
+      "step": 662
+    },
+    {
+      "epoch": 0.8912038981769301,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.9104010934192795e-07,
+      "loss": 0.649,
+      "step": 663
+    },
+    {
+      "epoch": 0.8925480971183736,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.838788013040139e-07,
+      "loss": 0.669,
+      "step": 664
+    },
+    {
+      "epoch": 0.8938922960598169,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.7680412481688013e-07,
+      "loss": 0.693,
+      "step": 665
+    },
+    {
+      "epoch": 0.8952364950012602,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.698162098369406e-07,
+      "loss": 0.649,
+      "step": 666
+    },
+    {
+      "epoch": 0.8965806939427036,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.6291518472686404e-07,
+      "loss": 0.6503,
+      "step": 667
+    },
+    {
+      "epoch": 0.8979248928841469,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.561011762532212e-07,
+      "loss": 0.6714,
+      "step": 668
+    },
+    {
+      "epoch": 0.8992690918255902,
+      "grad_norm": 0.78515625,
+      "learning_rate": 2.4937430958415275e-07,
+      "loss": 0.646,
+      "step": 669
+    },
+    {
+      "epoch": 0.9006132907670336,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.4273470828707014e-07,
+      "loss": 0.6876,
+      "step": 670
+    },
+    {
+      "epoch": 0.9019574897084769,
+      "grad_norm": 0.75390625,
+      "learning_rate": 2.361824943263874e-07,
+      "loss": 0.6294,
+      "step": 671
+    },
+    {
+      "epoch": 0.9033016886499202,
+      "grad_norm": 0.8046875,
+      "learning_rate": 2.2971778806127997e-07,
+      "loss": 0.6249,
+      "step": 672
+    },
+    {
+      "epoch": 0.9046458875913636,
+      "grad_norm": 0.78515625,
+      "learning_rate": 2.2334070824347243e-07,
+      "loss": 0.6247,
+      "step": 673
+    },
+    {
+      "epoch": 0.9059900865328069,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.1705137201505967e-07,
+      "loss": 0.6516,
+      "step": 674
+    },
+    {
+      "epoch": 0.9073342854742502,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.1084989490635254e-07,
+      "loss": 0.6497,
+      "step": 675
+    },
+    {
+      "epoch": 0.9086784844156935,
+      "grad_norm": 0.83203125,
+      "learning_rate": 2.0473639083375796e-07,
+      "loss": 0.6489,
+      "step": 676
+    },
+    {
+      "epoch": 0.9100226833571369,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.9871097209768375e-07,
+      "loss": 0.6255,
+      "step": 677
+    },
+    {
+      "epoch": 0.9113668822985802,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.9277374938047989e-07,
+      "loss": 0.6596,
+      "step": 678
+    },
+    {
+      "epoch": 0.9127110812400235,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.8692483174439946e-07,
+      "loss": 0.6328,
+      "step": 679
+    },
+    {
+      "epoch": 0.9140552801814669,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.8116432662960037e-07,
+      "loss": 0.6387,
+      "step": 680
+    },
+    {
+      "epoch": 0.9153994791229102,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.7549233985217074e-07,
+      "loss": 0.6653,
+      "step": 681
+    },
+    {
+      "epoch": 0.9167436780643535,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.6990897560218212e-07,
+      "loss": 0.6423,
+      "step": 682
+    },
+    {
+      "epoch": 0.9180878770057969,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.644143364417794e-07,
+      "loss": 0.6218,
+      "step": 683
+    },
+    {
+      "epoch": 0.9194320759472402,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.5900852330329564e-07,
+      "loss": 0.6484,
+      "step": 684
+    },
+    {
+      "epoch": 0.9207762748886835,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.5369163548739464e-07,
+      "loss": 0.6568,
+      "step": 685
+    },
+    {
+      "epoch": 0.9221204738301269,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.4846377066125351e-07,
+      "loss": 0.6457,
+      "step": 686
+    },
+    {
+      "epoch": 0.9234646727715702,
+      "grad_norm": 0.765625,
+      "learning_rate": 1.4332502485676358e-07,
+      "loss": 0.6276,
+      "step": 687
+    },
+    {
+      "epoch": 0.9248088717130135,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.3827549246876627e-07,
+      "loss": 0.6767,
+      "step": 688
+    },
+    {
+      "epoch": 0.9261530706544568,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.3331526625332268e-07,
+      "loss": 0.6218,
+      "step": 689
+    },
+    {
+      "epoch": 0.9274972695959002,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.2844443732600576e-07,
+      "loss": 0.6505,
+      "step": 690
+    },
+    {
+      "epoch": 0.9288414685373435,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.2366309516022966e-07,
+      "loss": 0.6412,
+      "step": 691
+    },
+    {
+      "epoch": 0.9301856674787868,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1897132758560471e-07,
+      "loss": 0.6617,
+      "step": 692
+    },
+    {
+      "epoch": 0.9315298664202302,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1436922078632395e-07,
+      "loss": 0.64,
+      "step": 693
+    },
+    {
+      "epoch": 0.9328740653616735,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.0985685929958135e-07,
+      "loss": 0.6292,
+      "step": 694
+    },
+    {
+      "epoch": 0.9342182643031168,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.0543432601401615e-07,
+      "loss": 0.6764,
+      "step": 695
+    },
+    {
+      "epoch": 0.9355624632445602,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.0110170216819315e-07,
+      "loss": 0.6528,
+      "step": 696
+    },
+    {
+      "epoch": 0.9369066621860035,
+      "grad_norm": 0.79296875,
+      "learning_rate": 9.685906734910988e-08,
+      "loss": 0.6624,
+      "step": 697
+    },
+    {
+      "epoch": 0.9382508611274468,
+      "grad_norm": 0.79296875,
+      "learning_rate": 9.270649949073229e-08,
+      "loss": 0.6362,
+      "step": 698
+    },
+    {
+      "epoch": 0.9395950600688902,
+      "grad_norm": 0.79296875,
+      "learning_rate": 8.8644074872567e-08,
+      "loss": 0.6367,
+      "step": 699
+    },
+    {
+      "epoch": 0.9409392590103335,
+      "grad_norm": 0.796875,
+      "learning_rate": 8.467186811825623e-08,
+      "loss": 0.6606,
+      "step": 700
+    },
+    {
+      "epoch": 0.9422834579517768,
+      "grad_norm": 0.80078125,
+      "learning_rate": 8.07899521942096e-08,
+      "loss": 0.6489,
+      "step": 701
+    },
+    {
+      "epoch": 0.9436276568932201,
+      "grad_norm": 0.765625,
+      "learning_rate": 7.69983984082634e-08,
+      "loss": 0.6556,
+      "step": 702
+    },
+    {
+      "epoch": 0.9449718558346635,
+      "grad_norm": 0.81640625,
+      "learning_rate": 7.329727640837059e-08,
+      "loss": 0.6676,
+      "step": 703
+    },
+    {
+      "epoch": 0.9463160547761069,
+      "grad_norm": 0.75390625,
+      "learning_rate": 6.968665418131848e-08,
+      "loss": 0.6127,
+      "step": 704
+    },
+    {
+      "epoch": 0.9476602537175502,
+      "grad_norm": 0.78515625,
+      "learning_rate": 6.616659805148695e-08,
+      "loss": 0.6773,
+      "step": 705
+    },
+    {
+      "epoch": 0.9490044526589936,
+      "grad_norm": 0.8046875,
+      "learning_rate": 6.273717267962165e-08,
+      "loss": 0.6753,
+      "step": 706
+    },
+    {
+      "epoch": 0.9503486516004369,
+      "grad_norm": 0.78125,
+      "learning_rate": 5.9398441061652714e-08,
+      "loss": 0.6306,
+      "step": 707
+    },
+    {
+      "epoch": 0.9516928505418802,
+      "grad_norm": 0.76953125,
+      "learning_rate": 5.615046452753403e-08,
+      "loss": 0.6623,
+      "step": 708
+    },
+    {
+      "epoch": 0.9530370494833236,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.299330274011916e-08,
+      "loss": 0.6346,
+      "step": 709
+    },
+    {
+      "epoch": 0.9543812484247669,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.992701369406161e-08,
+      "loss": 0.6213,
+      "step": 710
+    },
+    {
+      "epoch": 0.9557254473662102,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.695165371475463e-08,
+      "loss": 0.6431,
+      "step": 711
+    },
+    {
+      "epoch": 0.9570696463076536,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4067277457292553e-08,
+      "loss": 0.6355,
+      "step": 712
+    },
+    {
+      "epoch": 0.9584138452490969,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.127393790546719e-08,
+      "loss": 0.6526,
+      "step": 713
+    },
+    {
+      "epoch": 0.9597580441905402,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.857168637079744e-08,
+      "loss": 0.6663,
+      "step": 714
+    },
+    {
+      "epoch": 0.9611022431319836,
+      "grad_norm": 0.7890625,
+      "learning_rate": 3.596057249158347e-08,
+      "loss": 0.6246,
+      "step": 715
+    },
+    {
+      "epoch": 0.9624464420734269,
+      "grad_norm": 0.8125,
+      "learning_rate": 3.344064423199567e-08,
+      "loss": 0.6345,
+      "step": 716
+    },
+    {
+      "epoch": 0.9637906410148702,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.101194788119599e-08,
+      "loss": 0.6369,
+      "step": 717
+    },
+    {
+      "epoch": 0.9651348399563136,
+      "grad_norm": 0.8125,
+      "learning_rate": 2.8674528052484162e-08,
+      "loss": 0.6639,
+      "step": 718
+    },
+    {
+      "epoch": 0.9664790388977569,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.6428427682480552e-08,
+      "loss": 0.6701,
+      "step": 719
+    },
+    {
+      "epoch": 0.9678232378392002,
+      "grad_norm": 0.796875,
+      "learning_rate": 2.4273688030336805e-08,
+      "loss": 0.6696,
+      "step": 720
+    },
+    {
+      "epoch": 0.9691674367806435,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.221034867697702e-08,
+      "loss": 0.6525,
+      "step": 721
+    },
+    {
+      "epoch": 0.9705116357220869,
+      "grad_norm": 0.76953125,
+      "learning_rate": 2.0238447524372206e-08,
+      "loss": 0.6532,
+      "step": 722
+    },
+    {
+      "epoch": 0.9718558346635302,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.8358020794843056e-08,
+      "loss": 0.6504,
+      "step": 723
+    },
+    {
+      "epoch": 0.9732000336049735,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.656910303039494e-08,
+      "loss": 0.6338,
+      "step": 724
+    },
+    {
+      "epoch": 0.9745442325464169,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.4871727092083398e-08,
+      "loss": 0.6581,
+      "step": 725
+    },
+    {
+      "epoch": 0.9758884314878602,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.3265924159410193e-08,
+      "loss": 0.6551,
+      "step": 726
+    },
+    {
+      "epoch": 0.9772326304293035,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1751723729750975e-08,
+      "loss": 0.6729,
+      "step": 727
+    },
+    {
+      "epoch": 0.9785768293707469,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.0329153617812948e-08,
+      "loss": 0.6887,
+      "step": 728
+    },
+    {
+      "epoch": 0.9799210283121902,
+      "grad_norm": 0.78515625,
+      "learning_rate": 8.998239955124722e-09,
+      "loss": 0.6568,
+      "step": 729
+    },
+    {
+      "epoch": 0.9812652272536335,
+      "grad_norm": 0.80078125,
+      "learning_rate": 7.759007189555578e-09,
+      "loss": 0.6433,
+      "step": 730
+    },
+    {
+      "epoch": 0.9826094261950769,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.611478084866951e-09,
+      "loss": 0.6676,
+      "step": 731
+    },
+    {
+      "epoch": 0.9839536251365202,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.5556737202927534e-09,
+      "loss": 0.6519,
+      "step": 732
+    },
+    {
+      "epoch": 0.9852978240779635,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.5916134901552445e-09,
+      "loss": 0.6383,
+      "step": 733
+    },
+    {
+      "epoch": 0.9866420230194068,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.7193151035047614e-09,
+      "loss": 0.6309,
+      "step": 734
+    },
+    {
+      "epoch": 0.9879862219608502,
+      "grad_norm": 0.76953125,
+      "learning_rate": 2.93879458379831e-09,
+      "loss": 0.6682,
+      "step": 735
+    },
+    {
+      "epoch": 0.9893304209022935,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.2500662686025797e-09,
+      "loss": 0.6513,
+      "step": 736
+    },
+    {
+      "epoch": 0.9906746198437368,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.6531428093313761e-09,
+      "loss": 0.627,
+      "step": 737
+    },
+    {
+      "epoch": 0.9920188187851802,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.148035171014139e-09,
+      "loss": 0.6565,
+      "step": 738
+    },
+    {
+      "epoch": 0.9933630177266235,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.347526320927722e-10,
+      "loss": 0.6631,
+      "step": 739
+    },
+    {
+      "epoch": 0.9947072166680668,
+      "grad_norm": 0.81640625,
+      "learning_rate": 4.1330278425177896e-10,
+      "loss": 0.6566,
+      "step": 740
+    },
+    {
+      "epoch": 0.9960514156095102,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.8369153228114946e-10,
+      "loss": 0.6564,
+      "step": 741
+    },
+    {
+      "epoch": 0.9973956145509535,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.5923093963118334e-11,
+      "loss": 0.6299,
+      "step": 742
+    },
+    {
+      "epoch": 0.9987398134923968,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0,
+      "loss": 0.643,
+      "step": 743
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 743,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 248,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.7541079597200703e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}