diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,7049 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.88888888888889,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008888888888888889,
+      "grad_norm": 0.5841383934020996,
+      "learning_rate": 8.928571428571428e-07,
+      "loss": 0.4865,
+      "step": 1
+    },
+    {
+      "epoch": 0.017777777777777778,
+      "grad_norm": 0.37317243218421936,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 0.4116,
+      "step": 2
+    },
+    {
+      "epoch": 0.02666666666666667,
+      "grad_norm": 0.16240684688091278,
+      "learning_rate": 2.6785714285714285e-06,
+      "loss": 0.3085,
+      "step": 3
+    },
+    {
+      "epoch": 0.035555555555555556,
+      "grad_norm": 0.22618648409843445,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.3213,
+      "step": 4
+    },
+    {
+      "epoch": 0.044444444444444446,
+      "grad_norm": 0.33735740184783936,
+      "learning_rate": 4.464285714285715e-06,
+      "loss": 0.2869,
+      "step": 5
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 0.20259538292884827,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.3825,
+      "step": 6
+    },
+    {
+      "epoch": 0.06222222222222222,
+      "grad_norm": 0.3431294560432434,
+      "learning_rate": 6.25e-06,
+      "loss": 0.3896,
+      "step": 7
+    },
+    {
+      "epoch": 0.07111111111111111,
+      "grad_norm": 0.5504677891731262,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.3175,
+      "step": 8
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.1282545030117035,
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 0.2901,
+      "step": 9
+    },
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 0.18529950082302094,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.3735,
+      "step": 10
+    },
+    {
+      "epoch": 0.09777777777777778,
+      "grad_norm": 0.2145024836063385,
+      "learning_rate": 9.821428571428573e-06,
+      "loss": 0.3525,
+      "step": 11
+    },
+    {
+      "epoch": 0.10666666666666667,
+      "grad_norm": 0.9437930583953857,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.444,
+      "step": 12
+    },
+    {
+      "epoch": 0.11555555555555555,
+      "grad_norm": 0.24417199194431305,
+      "learning_rate": 1.1607142857142857e-05,
+      "loss": 0.2967,
+      "step": 13
+    },
+    {
+      "epoch": 0.12444444444444444,
+      "grad_norm": 0.2694835662841797,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2967,
+      "step": 14
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 0.20757225155830383,
+      "learning_rate": 1.3392857142857144e-05,
+      "loss": 0.3487,
+      "step": 15
+    },
+    {
+      "epoch": 0.14222222222222222,
+      "grad_norm": 0.2921523153781891,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.4247,
+      "step": 16
+    },
+    {
+      "epoch": 0.1511111111111111,
+      "grad_norm": 0.20908255875110626,
+      "learning_rate": 1.5178571428571429e-05,
+      "loss": 0.3009,
+      "step": 17
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.15087872743606567,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.3145,
+      "step": 18
+    },
+    {
+      "epoch": 0.1688888888888889,
+      "grad_norm": 0.8311613202095032,
+      "learning_rate": 1.6964285714285715e-05,
+      "loss": 0.3868,
+      "step": 19
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.27965763211250305,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.351,
+      "step": 20
+    },
+    {
+      "epoch": 0.18666666666666668,
+      "grad_norm": 0.23372279107570648,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.3025,
+      "step": 21
+    },
+    {
+      "epoch": 0.19555555555555557,
+      "grad_norm": 0.3673437237739563,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.388,
+      "step": 22
+    },
+    {
+      "epoch": 0.20444444444444446,
+      "grad_norm": 0.39919161796569824,
+      "learning_rate": 2.0535714285714285e-05,
+      "loss": 0.3611,
+      "step": 23
+    },
+    {
+      "epoch": 0.21333333333333335,
+      "grad_norm": 0.20096033811569214,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.3244,
+      "step": 24
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.27874326705932617,
+      "learning_rate": 2.2321428571428575e-05,
+      "loss": 0.359,
+      "step": 25
+    },
+    {
+      "epoch": 0.2311111111111111,
+      "grad_norm": 0.31668320298194885,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.3062,
+      "step": 26
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4131496548652649,
+      "learning_rate": 2.4107142857142858e-05,
+      "loss": 0.4351,
+      "step": 27
+    },
+    {
+      "epoch": 0.24888888888888888,
+      "grad_norm": 0.4470804035663605,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4332,
+      "step": 28
+    },
+    {
+      "epoch": 0.2577777777777778,
+      "grad_norm": 0.3746657967567444,
+      "learning_rate": 2.5892857142857148e-05,
+      "loss": 0.3811,
+      "step": 29
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.8335658311843872,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.4618,
+      "step": 30
+    },
+    {
+      "epoch": 0.27555555555555555,
+      "grad_norm": 0.8326655626296997,
+      "learning_rate": 2.767857142857143e-05,
+      "loss": 0.3872,
+      "step": 31
+    },
+    {
+      "epoch": 0.28444444444444444,
+      "grad_norm": 0.7701701521873474,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.3255,
+      "step": 32
+    },
+    {
+      "epoch": 0.29333333333333333,
+      "grad_norm": 0.40850526094436646,
+      "learning_rate": 2.9464285714285718e-05,
+      "loss": 0.3167,
+      "step": 33
+    },
+    {
+      "epoch": 0.3022222222222222,
+      "grad_norm": 0.35397693514823914,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.2987,
+      "step": 34
+    },
+    {
+      "epoch": 0.3111111111111111,
+      "grad_norm": 0.7541699409484863,
+      "learning_rate": 3.125e-05,
+      "loss": 0.3413,
+      "step": 35
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3483797609806061,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.2858,
+      "step": 36
+    },
+    {
+      "epoch": 0.3288888888888889,
+      "grad_norm": 0.33964604139328003,
+      "learning_rate": 3.303571428571429e-05,
+      "loss": 0.3036,
+      "step": 37
+    },
+    {
+      "epoch": 0.3377777777777778,
+      "grad_norm": 0.45213809609413147,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.3437,
+      "step": 38
+    },
+    {
+      "epoch": 0.3466666666666667,
+      "grad_norm": 0.4597315192222595,
+      "learning_rate": 3.4821428571428574e-05,
+      "loss": 0.3318,
+      "step": 39
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 0.44681742787361145,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.3107,
+      "step": 40
+    },
+    {
+      "epoch": 0.36444444444444446,
+      "grad_norm": 0.9226369857788086,
+      "learning_rate": 3.6607142857142853e-05,
+      "loss": 0.3231,
+      "step": 41
+    },
+    {
+      "epoch": 0.37333333333333335,
+      "grad_norm": 0.27465879917144775,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.3301,
+      "step": 42
+    },
+    {
+      "epoch": 0.38222222222222224,
+      "grad_norm": 0.4791021943092346,
+      "learning_rate": 3.839285714285715e-05,
+      "loss": 0.3527,
+      "step": 43
+    },
+    {
+      "epoch": 0.39111111111111113,
+      "grad_norm": 0.586669385433197,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.3359,
+      "step": 44
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34871557354927063,
+      "learning_rate": 4.017857142857143e-05,
+      "loss": 0.3268,
+      "step": 45
+    },
+    {
+      "epoch": 0.4088888888888889,
+      "grad_norm": 0.4256209135055542,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.2978,
+      "step": 46
+    },
+    {
+      "epoch": 0.4177777777777778,
+      "grad_norm": 0.30336323380470276,
+      "learning_rate": 4.196428571428572e-05,
+      "loss": 0.2898,
+      "step": 47
+    },
+    {
+      "epoch": 0.4266666666666667,
+      "grad_norm": 0.5795422792434692,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.3481,
+      "step": 48
+    },
+    {
+      "epoch": 0.43555555555555553,
+      "grad_norm": 0.38410332798957825,
+      "learning_rate": 4.375e-05,
+      "loss": 0.3414,
+      "step": 49
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.3348947763442993,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.336,
+      "step": 50
+    },
+    {
+      "epoch": 0.4533333333333333,
+      "grad_norm": 0.42829787731170654,
+      "learning_rate": 4.5535714285714286e-05,
+      "loss": 0.2913,
+      "step": 51
+    },
+    {
+      "epoch": 0.4622222222222222,
+      "grad_norm": 4.1896653175354,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.8937,
+      "step": 52
+    },
+    {
+      "epoch": 0.4711111111111111,
+      "grad_norm": 0.7059090733528137,
+      "learning_rate": 4.732142857142857e-05,
+      "loss": 0.4011,
+      "step": 53
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41015371680259705,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.3351,
+      "step": 54
+    },
+    {
+      "epoch": 0.4888888888888889,
+      "grad_norm": 0.43993642926216125,
+      "learning_rate": 4.910714285714286e-05,
+      "loss": 0.3313,
+      "step": 55
+    },
+    {
+      "epoch": 0.49777777777777776,
+      "grad_norm": 1.0636693239212036,
+      "learning_rate": 5e-05,
+      "loss": 0.4636,
+      "step": 56
+    },
+    {
+      "epoch": 0.5066666666666667,
+      "grad_norm": 1.5080366134643555,
+      "learning_rate": 5.089285714285714e-05,
+      "loss": 0.5489,
+      "step": 57
+    },
+    {
+      "epoch": 0.5155555555555555,
+      "grad_norm": 0.5345658659934998,
+      "learning_rate": 5.1785714285714296e-05,
+      "loss": 0.3057,
+      "step": 58
+    },
+    {
+      "epoch": 0.5244444444444445,
+      "grad_norm": 0.7976881861686707,
+      "learning_rate": 5.267857142857143e-05,
+      "loss": 0.3581,
+      "step": 59
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.726458728313446,
+      "learning_rate": 5.3571428571428575e-05,
+      "loss": 0.3279,
+      "step": 60
+    },
+    {
+      "epoch": 0.5422222222222223,
+      "grad_norm": 0.5178459286689758,
+      "learning_rate": 5.446428571428571e-05,
+      "loss": 0.3217,
+      "step": 61
+    },
+    {
+      "epoch": 0.5511111111111111,
+      "grad_norm": 0.6377764940261841,
+      "learning_rate": 5.535714285714286e-05,
+      "loss": 0.321,
+      "step": 62
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3174062967300415,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 0.3154,
+      "step": 63
+    },
+    {
+      "epoch": 0.5688888888888889,
+      "grad_norm": 0.31623443961143494,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 0.2735,
+      "step": 64
+    },
+    {
+      "epoch": 0.5777777777777777,
+      "grad_norm": 0.3521466851234436,
+      "learning_rate": 5.803571428571429e-05,
+      "loss": 0.2878,
+      "step": 65
+    },
+    {
+      "epoch": 0.5866666666666667,
+      "grad_norm": 0.8923875689506531,
+      "learning_rate": 5.8928571428571435e-05,
+      "loss": 0.3906,
+      "step": 66
+    },
+    {
+      "epoch": 0.5955555555555555,
+      "grad_norm": 0.6803852915763855,
+      "learning_rate": 5.982142857142857e-05,
+      "loss": 0.3573,
+      "step": 67
+    },
+    {
+      "epoch": 0.6044444444444445,
+      "grad_norm": 0.5237946510314941,
+      "learning_rate": 6.0714285714285715e-05,
+      "loss": 0.3068,
+      "step": 68
+    },
+    {
+      "epoch": 0.6133333333333333,
+      "grad_norm": 0.6161757111549377,
+      "learning_rate": 6.160714285714286e-05,
+      "loss": 0.2944,
+      "step": 69
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 0.3687132000923157,
+      "learning_rate": 6.25e-05,
+      "loss": 0.2901,
+      "step": 70
+    },
+    {
+      "epoch": 0.6311111111111111,
+      "grad_norm": 0.39233818650245667,
+      "learning_rate": 6.339285714285714e-05,
+      "loss": 0.2821,
+      "step": 71
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7180721759796143,
+      "learning_rate": 6.428571428571429e-05,
+      "loss": 0.3509,
+      "step": 72
+    },
+    {
+      "epoch": 0.6488888888888888,
+      "grad_norm": 0.9132435917854309,
+      "learning_rate": 6.517857142857143e-05,
+      "loss": 0.4027,
+      "step": 73
+    },
+    {
+      "epoch": 0.6577777777777778,
+      "grad_norm": 0.2931051552295685,
+      "learning_rate": 6.607142857142857e-05,
+      "loss": 0.3303,
+      "step": 74
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.4739736318588257,
+      "learning_rate": 6.696428571428572e-05,
+      "loss": 0.286,
+      "step": 75
+    },
+    {
+      "epoch": 0.6755555555555556,
+      "grad_norm": 0.645233154296875,
+      "learning_rate": 6.785714285714286e-05,
+      "loss": 0.3673,
+      "step": 76
+    },
+    {
+      "epoch": 0.6844444444444444,
+      "grad_norm": 0.6568748354911804,
+      "learning_rate": 6.875e-05,
+      "loss": 0.3451,
+      "step": 77
+    },
+    {
+      "epoch": 0.6933333333333334,
+      "grad_norm": 0.330121785402298,
+      "learning_rate": 6.964285714285715e-05,
+      "loss": 0.2916,
+      "step": 78
+    },
+    {
+      "epoch": 0.7022222222222222,
+      "grad_norm": 0.6969891786575317,
+      "learning_rate": 7.053571428571429e-05,
+      "loss": 0.3799,
+      "step": 79
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.3836056590080261,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 0.2785,
+      "step": 80
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6308532357215881,
+      "learning_rate": 7.232142857142858e-05,
+      "loss": 0.3365,
+      "step": 81
+    },
+    {
+      "epoch": 0.7288888888888889,
+      "grad_norm": 1.3300080299377441,
+      "learning_rate": 7.321428571428571e-05,
+      "loss": 0.4452,
+      "step": 82
+    },
+    {
+      "epoch": 0.7377777777777778,
+      "grad_norm": 0.4857744872570038,
+      "learning_rate": 7.410714285714286e-05,
+      "loss": 0.3408,
+      "step": 83
+    },
+    {
+      "epoch": 0.7466666666666667,
+      "grad_norm": 0.2752129137516022,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.2481,
+      "step": 84
+    },
+    {
+      "epoch": 0.7555555555555555,
+      "grad_norm": 0.37411218881607056,
+      "learning_rate": 7.589285714285714e-05,
+      "loss": 0.2714,
+      "step": 85
+    },
+    {
+      "epoch": 0.7644444444444445,
+      "grad_norm": 0.34179869294166565,
+      "learning_rate": 7.67857142857143e-05,
+      "loss": 0.322,
+      "step": 86
+    },
+    {
+      "epoch": 0.7733333333333333,
+      "grad_norm": 0.5072541236877441,
+      "learning_rate": 7.767857142857144e-05,
+      "loss": 0.3442,
+      "step": 87
+    },
+    {
+      "epoch": 0.7822222222222223,
+      "grad_norm": 0.3834559917449951,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 0.2904,
+      "step": 88
+    },
+    {
+      "epoch": 0.7911111111111111,
+      "grad_norm": 0.37922006845474243,
+      "learning_rate": 7.946428571428571e-05,
+      "loss": 0.3116,
+      "step": 89
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41435790061950684,
+      "learning_rate": 8.035714285714287e-05,
+      "loss": 0.2949,
+      "step": 90
+    },
+    {
+      "epoch": 0.8088888888888889,
+      "grad_norm": 0.5537578463554382,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 0.3337,
+      "step": 91
+    },
+    {
+      "epoch": 0.8177777777777778,
+      "grad_norm": 0.24957779049873352,
+      "learning_rate": 8.214285714285714e-05,
+      "loss": 0.2709,
+      "step": 92
+    },
+    {
+      "epoch": 0.8266666666666667,
+      "grad_norm": 0.43642184138298035,
+      "learning_rate": 8.30357142857143e-05,
+      "loss": 0.3033,
+      "step": 93
+    },
+    {
+      "epoch": 0.8355555555555556,
+      "grad_norm": 0.835472583770752,
+      "learning_rate": 8.392857142857144e-05,
+      "loss": 0.3828,
+      "step": 94
+    },
+    {
+      "epoch": 0.8444444444444444,
+      "grad_norm": 0.6168670654296875,
+      "learning_rate": 8.482142857142857e-05,
+      "loss": 0.3455,
+      "step": 95
+    },
+    {
+      "epoch": 0.8533333333333334,
+      "grad_norm": 0.46005958318710327,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 0.3489,
+      "step": 96
+    },
+    {
+      "epoch": 0.8622222222222222,
+      "grad_norm": 0.34383633732795715,
+      "learning_rate": 8.660714285714287e-05,
+      "loss": 0.2866,
+      "step": 97
+    },
+    {
+      "epoch": 0.8711111111111111,
+      "grad_norm": 0.4366074204444885,
+      "learning_rate": 8.75e-05,
+      "loss": 0.3239,
+      "step": 98
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.33174213767051697,
+      "learning_rate": 8.839285714285714e-05,
+      "loss": 0.2783,
+      "step": 99
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.26910072565078735,
+      "learning_rate": 8.92857142857143e-05,
+      "loss": 0.2807,
+      "step": 100
+    },
+    {
+      "epoch": 0.8977777777777778,
+      "grad_norm": 0.9884425401687622,
+      "learning_rate": 9.017857142857143e-05,
+      "loss": 0.4379,
+      "step": 101
+    },
+    {
+      "epoch": 0.9066666666666666,
+      "grad_norm": 0.24256423115730286,
+      "learning_rate": 9.107142857142857e-05,
+      "loss": 0.231,
+      "step": 102
+    },
+    {
+      "epoch": 0.9155555555555556,
+      "grad_norm": 0.36811211705207825,
+      "learning_rate": 9.196428571428572e-05,
+      "loss": 0.2668,
+      "step": 103
+    },
+    {
+      "epoch": 0.9244444444444444,
+      "grad_norm": 0.45584559440612793,
+      "learning_rate": 9.285714285714286e-05,
+      "loss": 0.3285,
+      "step": 104
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 0.4034405052661896,
+      "learning_rate": 9.375e-05,
+      "loss": 0.2823,
+      "step": 105
+    },
+    {
+      "epoch": 0.9422222222222222,
+      "grad_norm": 0.48256492614746094,
+      "learning_rate": 9.464285714285715e-05,
+      "loss": 0.3389,
+      "step": 106
+    },
+    {
+      "epoch": 0.9511111111111111,
+      "grad_norm": 0.4109070897102356,
+      "learning_rate": 9.553571428571429e-05,
+      "loss": 0.3515,
+      "step": 107
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.8162788152694702,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 0.5591,
+      "step": 108
+    },
+    {
+      "epoch": 0.9688888888888889,
+      "grad_norm": 0.33671724796295166,
+      "learning_rate": 9.732142857142858e-05,
+      "loss": 0.3512,
+      "step": 109
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 0.35684165358543396,
+      "learning_rate": 9.821428571428572e-05,
+      "loss": 0.3087,
+      "step": 110
+    },
+    {
+      "epoch": 0.9866666666666667,
+      "grad_norm": 0.34859076142311096,
+      "learning_rate": 9.910714285714286e-05,
+      "loss": 0.2894,
+      "step": 111
+    },
+    {
+      "epoch": 0.9955555555555555,
+      "grad_norm": 0.3526354432106018,
+      "learning_rate": 0.0001,
+      "loss": 0.3052,
+      "step": 112
+    },
+    {
+      "epoch": 1.0044444444444445,
+      "grad_norm": 0.2897067368030548,
+      "learning_rate": 9.999975716105452e-05,
+      "loss": 0.3058,
+      "step": 113
+    },
+    {
+      "epoch": 1.0133333333333334,
+      "grad_norm": 0.4105049967765808,
+      "learning_rate": 9.999902864657691e-05,
+      "loss": 0.2635,
+      "step": 114
+    },
+    {
+      "epoch": 1.0222222222222221,
+      "grad_norm": 0.36226218938827515,
+      "learning_rate": 9.999781446364365e-05,
+      "loss": 0.3257,
+      "step": 115
+    },
+    {
+      "epoch": 1.031111111111111,
+      "grad_norm": 0.31135883927345276,
+      "learning_rate": 9.999611462404875e-05,
+      "loss": 0.2646,
+      "step": 116
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.0171165466308594,
+      "learning_rate": 9.999392914430371e-05,
+      "loss": 0.5321,
+      "step": 117
+    },
+    {
+      "epoch": 1.048888888888889,
+      "grad_norm": 0.3250499665737152,
+      "learning_rate": 9.999125804563732e-05,
+      "loss": 0.3208,
+      "step": 118
+    },
+    {
+      "epoch": 1.0577777777777777,
+      "grad_norm": 0.2886015474796295,
+      "learning_rate": 9.998810135399546e-05,
+      "loss": 0.2932,
+      "step": 119
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.3275207281112671,
+      "learning_rate": 9.998445910004082e-05,
+      "loss": 0.2971,
+      "step": 120
+    },
+    {
+      "epoch": 1.0755555555555556,
+      "grad_norm": 0.5577352643013,
+      "learning_rate": 9.998033131915266e-05,
+      "loss": 0.3561,
+      "step": 121
+    },
+    {
+      "epoch": 1.0844444444444445,
+      "grad_norm": 1.0877000093460083,
+      "learning_rate": 9.997571805142639e-05,
+      "loss": 0.4113,
+      "step": 122
+    },
+    {
+      "epoch": 1.0933333333333333,
+      "grad_norm": 0.5185580253601074,
+      "learning_rate": 9.997061934167328e-05,
+      "loss": 0.3085,
+      "step": 123
+    },
+    {
+      "epoch": 1.1022222222222222,
+      "grad_norm": 0.4035893678665161,
+      "learning_rate": 9.996503523941994e-05,
+      "loss": 0.271,
+      "step": 124
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.3941110074520111,
+      "learning_rate": 9.995896579890784e-05,
+      "loss": 0.2875,
+      "step": 125
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3415586054325104,
+      "learning_rate": 9.99524110790929e-05,
+      "loss": 0.2711,
+      "step": 126
+    },
+    {
+      "epoch": 1.1288888888888888,
+      "grad_norm": 0.3006160259246826,
+      "learning_rate": 9.99453711436447e-05,
+      "loss": 0.2718,
+      "step": 127
+    },
+    {
+      "epoch": 1.1377777777777778,
+      "grad_norm": 0.522278368473053,
+      "learning_rate": 9.993784606094612e-05,
+      "loss": 0.3306,
+      "step": 128
+    },
+    {
+      "epoch": 1.1466666666666667,
+      "grad_norm": 0.5858255624771118,
+      "learning_rate": 9.992983590409246e-05,
+      "loss": 0.3475,
+      "step": 129
+    },
+    {
+      "epoch": 1.1555555555555554,
+      "grad_norm": 0.5486164689064026,
+      "learning_rate": 9.992134075089084e-05,
+      "loss": 0.3259,
+      "step": 130
+    },
+    {
+      "epoch": 1.1644444444444444,
+      "grad_norm": 0.2975933253765106,
+      "learning_rate": 9.991236068385941e-05,
+      "loss": 0.2588,
+      "step": 131
+    },
+    {
+      "epoch": 1.1733333333333333,
+      "grad_norm": 0.6659825444221497,
+      "learning_rate": 9.99028957902266e-05,
+      "loss": 0.3266,
+      "step": 132
+    },
+    {
+      "epoch": 1.1822222222222223,
+      "grad_norm": 0.2541256248950958,
+      "learning_rate": 9.989294616193017e-05,
+      "loss": 0.2649,
+      "step": 133
+    },
+    {
+      "epoch": 1.1911111111111112,
+      "grad_norm": 0.8277371525764465,
+      "learning_rate": 9.988251189561645e-05,
+      "loss": 0.4076,
+      "step": 134
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3177882432937622,
+      "learning_rate": 9.987159309263924e-05,
+      "loss": 0.304,
+      "step": 135
+    },
+    {
+      "epoch": 1.208888888888889,
+      "grad_norm": 0.36816540360450745,
+      "learning_rate": 9.986018985905901e-05,
+      "loss": 0.3187,
+      "step": 136
+    },
+    {
+      "epoch": 1.2177777777777778,
+      "grad_norm": 0.4456408619880676,
+      "learning_rate": 9.984830230564171e-05,
+      "loss": 0.2769,
+      "step": 137
+    },
+    {
+      "epoch": 1.2266666666666666,
+      "grad_norm": 0.7157383561134338,
+      "learning_rate": 9.983593054785776e-05,
+      "loss": 0.382,
+      "step": 138
+    },
+    {
+      "epoch": 1.2355555555555555,
+      "grad_norm": 0.5327372550964355,
+      "learning_rate": 9.982307470588098e-05,
+      "loss": 0.2732,
+      "step": 139
+    },
+    {
+      "epoch": 1.2444444444444445,
+      "grad_norm": 0.581408679485321,
+      "learning_rate": 9.980973490458728e-05,
+      "loss": 0.3956,
+      "step": 140
+    },
+    {
+      "epoch": 1.2533333333333334,
+      "grad_norm": 0.32390910387039185,
+      "learning_rate": 9.979591127355365e-05,
+      "loss": 0.2905,
+      "step": 141
+    },
+    {
+      "epoch": 1.2622222222222224,
+      "grad_norm": 0.4703962802886963,
+      "learning_rate": 9.978160394705668e-05,
+      "loss": 0.2897,
+      "step": 142
+    },
+    {
+      "epoch": 1.271111111111111,
+      "grad_norm": 0.5023928284645081,
+      "learning_rate": 9.976681306407148e-05,
+      "loss": 0.328,
+      "step": 143
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.3573106825351715,
+      "learning_rate": 9.975153876827008e-05,
+      "loss": 0.2687,
+      "step": 144
+    },
+    {
+      "epoch": 1.2888888888888888,
+      "grad_norm": 0.6152392625808716,
+      "learning_rate": 9.973578120802025e-05,
+      "loss": 0.3375,
+      "step": 145
+    },
+    {
+      "epoch": 1.2977777777777777,
+      "grad_norm": 0.40894556045532227,
+      "learning_rate": 9.971954053638399e-05,
+      "loss": 0.2888,
+      "step": 146
+    },
+    {
+      "epoch": 1.3066666666666666,
+      "grad_norm": 0.8339890241622925,
+      "learning_rate": 9.970281691111598e-05,
+      "loss": 0.4384,
+      "step": 147
+    },
+    {
+      "epoch": 1.3155555555555556,
+      "grad_norm": 0.41529226303100586,
+      "learning_rate": 9.968561049466214e-05,
+      "loss": 0.2831,
+      "step": 148
+    },
+    {
+      "epoch": 1.3244444444444445,
+      "grad_norm": 0.28021934628486633,
+      "learning_rate": 9.966792145415795e-05,
+      "loss": 0.2671,
+      "step": 149
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.22352413833141327,
+      "learning_rate": 9.964974996142698e-05,
+      "loss": 0.2839,
+      "step": 150
+    },
+    {
+      "epoch": 1.3422222222222222,
+      "grad_norm": 0.2703256905078888,
+      "learning_rate": 9.963109619297905e-05,
+      "loss": 0.2675,
+      "step": 151
+    },
+    {
+      "epoch": 1.3511111111111112,
+      "grad_norm": 0.2645833194255829,
+      "learning_rate": 9.961196033000861e-05,
+      "loss": 0.2708,
+      "step": 152
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 0.40951216220855713,
+      "learning_rate": 9.959234255839298e-05,
+      "loss": 0.3015,
+      "step": 153
+    },
+    {
+      "epoch": 1.3688888888888888,
+      "grad_norm": 0.3288329839706421,
+      "learning_rate": 9.957224306869053e-05,
+      "loss": 0.325,
+      "step": 154
+    },
+    {
+      "epoch": 1.3777777777777778,
+      "grad_norm": 0.3932753801345825,
+      "learning_rate": 9.955166205613879e-05,
+      "loss": 0.2716,
+      "step": 155
+    },
+    {
+      "epoch": 1.3866666666666667,
+      "grad_norm": 0.46717700362205505,
+      "learning_rate": 9.953059972065265e-05,
+      "loss": 0.3444,
+      "step": 156
+    },
+    {
+      "epoch": 1.3955555555555557,
+      "grad_norm": 0.20313459634780884,
+      "learning_rate": 9.950905626682228e-05,
+      "loss": 0.2584,
+      "step": 157
+    },
+    {
+      "epoch": 1.4044444444444444,
+      "grad_norm": 0.47062796354293823,
+      "learning_rate": 9.948703190391131e-05,
+      "loss": 0.372,
+      "step": 158
+    },
+    {
+      "epoch": 1.4133333333333333,
+      "grad_norm": 0.3500126898288727,
+      "learning_rate": 9.946452684585463e-05,
+      "loss": 0.2737,
+      "step": 159
+    },
+    {
+      "epoch": 1.4222222222222223,
+      "grad_norm": 0.5783170461654663,
+      "learning_rate": 9.944154131125642e-05,
+      "loss": 0.2938,
+      "step": 160
+    },
+    {
+      "epoch": 1.431111111111111,
+      "grad_norm": 0.27551746368408203,
+      "learning_rate": 9.941807552338804e-05,
+      "loss": 0.3139,
+      "step": 161
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2641688287258148,
+      "learning_rate": 9.939412971018574e-05,
+      "loss": 0.3009,
+      "step": 162
+    },
+    {
+      "epoch": 1.448888888888889,
+      "grad_norm": 0.4430491030216217,
+      "learning_rate": 9.936970410424857e-05,
+      "loss": 0.2779,
+      "step": 163
+    },
+    {
+      "epoch": 1.4577777777777778,
+      "grad_norm": 0.44353199005126953,
+      "learning_rate": 9.934479894283606e-05,
+      "loss": 0.2694,
+      "step": 164
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 0.3297507166862488,
+      "learning_rate": 9.931941446786594e-05,
+      "loss": 0.2638,
+      "step": 165
+    },
+    {
+      "epoch": 1.4755555555555555,
+      "grad_norm": 0.5766128897666931,
+      "learning_rate": 9.92935509259118e-05,
+      "loss": 0.3052,
+      "step": 166
+    },
+    {
+      "epoch": 1.4844444444444445,
+      "grad_norm": 0.3493499755859375,
+      "learning_rate": 9.92672085682006e-05,
+      "loss": 0.2728,
+      "step": 167
+    },
+    {
+      "epoch": 1.4933333333333334,
+      "grad_norm": 0.34138888120651245,
+      "learning_rate": 9.924038765061042e-05,
+      "loss": 0.2679,
+      "step": 168
+    },
+    {
+      "epoch": 1.5022222222222221,
+      "grad_norm": 0.40943869948387146,
+      "learning_rate": 9.921308843366772e-05,
+      "loss": 0.2556,
+      "step": 169
+    },
+    {
+      "epoch": 1.511111111111111,
+      "grad_norm": 0.4275529384613037,
+      "learning_rate": 9.918531118254507e-05,
+      "loss": 0.3012,
+      "step": 170
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3822483718395233,
+      "learning_rate": 9.915705616705839e-05,
+      "loss": 0.2984,
+      "step": 171
+    },
+    {
+      "epoch": 1.528888888888889,
+      "grad_norm": 0.3507990837097168,
+      "learning_rate": 9.912832366166442e-05,
+      "loss": 0.2839,
+      "step": 172
+    },
+    {
+      "epoch": 1.537777777777778,
+      "grad_norm": 0.3176634907722473,
+      "learning_rate": 9.909911394545799e-05,
+      "loss": 0.2715,
+      "step": 173
+    },
+    {
+      "epoch": 1.5466666666666666,
+      "grad_norm": 0.7413046956062317,
+      "learning_rate": 9.906942730216939e-05,
+      "loss": 0.2995,
+      "step": 174
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 0.5602743625640869,
+      "learning_rate": 9.903926402016153e-05,
+      "loss": 0.303,
+      "step": 175
+    },
+    {
+      "epoch": 1.5644444444444443,
+      "grad_norm": 0.3049962818622589,
+      "learning_rate": 9.900862439242719e-05,
+      "loss": 0.2866,
+      "step": 176
+    },
+    {
+      "epoch": 1.5733333333333333,
+      "grad_norm": 0.20894083380699158,
+      "learning_rate": 9.89775087165862e-05,
+      "loss": 0.2801,
+      "step": 177
+    },
+    {
+      "epoch": 1.5822222222222222,
+      "grad_norm": 0.4999159574508667,
+      "learning_rate": 9.894591729488242e-05,
+      "loss": 0.3153,
+      "step": 178
+    },
+    {
+      "epoch": 1.5911111111111111,
+      "grad_norm": 0.6849189400672913,
+      "learning_rate": 9.8913850434181e-05,
+      "loss": 0.2794,
+      "step": 179
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.44084635376930237,
+      "learning_rate": 9.888130844596524e-05,
+      "loss": 0.2953,
+      "step": 180
+    },
+    {
+      "epoch": 1.608888888888889,
+      "grad_norm": 0.2744970917701721,
+      "learning_rate": 9.884829164633359e-05,
+      "loss": 0.2654,
+      "step": 181
+    },
+    {
+      "epoch": 1.6177777777777778,
+      "grad_norm": 0.7441728711128235,
+      "learning_rate": 9.881480035599667e-05,
+      "loss": 0.4128,
+      "step": 182
+    },
+    {
+      "epoch": 1.6266666666666667,
+      "grad_norm": 0.2983834147453308,
+      "learning_rate": 9.878083490027406e-05,
+      "loss": 0.3103,
+      "step": 183
+    },
+    {
+      "epoch": 1.6355555555555554,
+      "grad_norm": 0.2417658269405365,
+      "learning_rate": 9.874639560909117e-05,
+      "loss": 0.248,
+      "step": 184
+    },
+    {
+      "epoch": 1.6444444444444444,
+      "grad_norm": 0.4583745002746582,
+      "learning_rate": 9.871148281697608e-05,
+      "loss": 0.2747,
+      "step": 185
+    },
+    {
+      "epoch": 1.6533333333333333,
+      "grad_norm": 0.42793506383895874,
+      "learning_rate": 9.867609686305617e-05,
+      "loss": 0.282,
+      "step": 186
+    },
+    {
+      "epoch": 1.6622222222222223,
+      "grad_norm": 0.32263195514678955,
+      "learning_rate": 9.864023809105497e-05,
+      "loss": 0.2709,
+      "step": 187
+    },
+    {
+      "epoch": 1.6711111111111112,
+      "grad_norm": 0.35320043563842773,
+      "learning_rate": 9.860390684928873e-05,
+      "loss": 0.3429,
+      "step": 188
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 0.6125680804252625,
+      "learning_rate": 9.856710349066307e-05,
+      "loss": 0.2844,
+      "step": 189
+    },
+    {
+      "epoch": 1.6888888888888889,
+      "grad_norm": 0.36522263288497925,
+      "learning_rate": 9.852982837266955e-05,
+      "loss": 0.2413,
+      "step": 190
+    },
+    {
+      "epoch": 1.6977777777777778,
+      "grad_norm": 0.3167021870613098,
+      "learning_rate": 9.849208185738217e-05,
+      "loss": 0.2682,
+      "step": 191
+    },
+    {
+      "epoch": 1.7066666666666666,
+      "grad_norm": 0.46384674310684204,
+      "learning_rate": 9.84538643114539e-05,
+      "loss": 0.2671,
+      "step": 192
+    },
+    {
+      "epoch": 1.7155555555555555,
+      "grad_norm": 0.27667102217674255,
+      "learning_rate": 9.841517610611309e-05,
+      "loss": 0.2929,
+      "step": 193
+    },
+    {
+      "epoch": 1.7244444444444444,
+      "grad_norm": 0.34263694286346436,
+      "learning_rate": 9.837601761715983e-05,
+      "loss": 0.2837,
+      "step": 194
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 0.5394869446754456,
+      "learning_rate": 9.833638922496238e-05,
+      "loss": 0.2535,
+      "step": 195
+    },
+    {
+      "epoch": 1.7422222222222223,
+      "grad_norm": 0.30996885895729065,
+      "learning_rate": 9.829629131445342e-05,
+      "loss": 0.2845,
+      "step": 196
+    },
+    {
+      "epoch": 1.751111111111111,
+      "grad_norm": 0.3415825664997101,
+      "learning_rate": 9.825572427512632e-05,
+      "loss": 0.2525,
+      "step": 197
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.23367656767368317,
+      "learning_rate": 9.82146885010314e-05,
+      "loss": 0.295,
+      "step": 198
+    },
+    {
+      "epoch": 1.7688888888888887,
+      "grad_norm": 0.32408076524734497,
+      "learning_rate": 9.817318439077195e-05,
+      "loss": 0.3182,
+      "step": 199
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.4190412759780884,
+      "learning_rate": 9.81312123475006e-05,
+      "loss": 0.2723,
+      "step": 200
+    },
+    {
+      "epoch": 1.7866666666666666,
+      "grad_norm": 0.28320616483688354,
+      "learning_rate": 9.808877277891521e-05,
+      "loss": 0.2618,
+      "step": 201
+    },
+    {
+      "epoch": 1.7955555555555556,
+      "grad_norm": 0.3757404386997223,
+      "learning_rate": 9.804586609725499e-05,
+      "loss": 0.3327,
+      "step": 202
+    },
+    {
+      "epoch": 1.8044444444444445,
+      "grad_norm": 0.2790829539299011,
+      "learning_rate": 9.800249271929645e-05,
+      "loss": 0.2746,
+      "step": 203
+    },
+    {
+      "epoch": 1.8133333333333335,
+      "grad_norm": 0.5675872564315796,
+      "learning_rate": 9.79586530663494e-05,
+      "loss": 0.3178,
+      "step": 204
+    },
+    {
+      "epoch": 1.8222222222222222,
+      "grad_norm": 0.7035036683082581,
+      "learning_rate": 9.791434756425288e-05,
+      "loss": 0.283,
+      "step": 205
+    },
+    {
+      "epoch": 1.8311111111111111,
+      "grad_norm": 0.22361284494400024,
+      "learning_rate": 9.78695766433709e-05,
+      "loss": 0.2674,
+      "step": 206
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 0.39389535784721375,
+      "learning_rate": 9.782434073858844e-05,
+      "loss": 0.3173,
+      "step": 207
+    },
+    {
+      "epoch": 1.8488888888888888,
+      "grad_norm": 0.38015905022621155,
+      "learning_rate": 9.777864028930705e-05,
+      "loss": 0.2277,
+      "step": 208
+    },
+    {
+      "epoch": 1.8577777777777778,
+      "grad_norm": 0.44615599513053894,
+      "learning_rate": 9.773247573944066e-05,
+      "loss": 0.2951,
+      "step": 209
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 0.19754816591739655,
+      "learning_rate": 9.768584753741134e-05,
+      "loss": 0.2519,
+      "step": 210
+    },
+    {
+      "epoch": 1.8755555555555556,
+      "grad_norm": 0.5915124416351318,
+      "learning_rate": 9.763875613614482e-05,
+      "loss": 0.3013,
+      "step": 211
+    },
+    {
+      "epoch": 1.8844444444444446,
+      "grad_norm": 0.4793195426464081,
+      "learning_rate": 9.759120199306613e-05,
+      "loss": 0.2667,
+      "step": 212
+    },
+    {
+      "epoch": 1.8933333333333333,
+      "grad_norm": 0.3155132830142975,
+      "learning_rate": 9.754318557009519e-05,
+      "loss": 0.2911,
+      "step": 213
+    },
+    {
+      "epoch": 1.9022222222222223,
+      "grad_norm": 0.3596380949020386,
+      "learning_rate": 9.74947073336423e-05,
+      "loss": 0.3004,
+      "step": 214
+    },
+    {
+      "epoch": 1.911111111111111,
+      "grad_norm": 0.30630800127983093,
+      "learning_rate": 9.744576775460364e-05,
+      "loss": 0.2379,
+      "step": 215
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3115442097187042,
+      "learning_rate": 9.73963673083566e-05,
+      "loss": 0.2837,
+      "step": 216
+    },
+    {
+      "epoch": 1.9288888888888889,
+      "grad_norm": 0.36604344844818115,
+      "learning_rate": 9.73465064747553e-05,
+      "loss": 0.3264,
+      "step": 217
+    },
+    {
+      "epoch": 1.9377777777777778,
+      "grad_norm": 0.29043278098106384,
+      "learning_rate": 9.72961857381258e-05,
+      "loss": 0.3202,
+      "step": 218
+    },
+    {
+      "epoch": 1.9466666666666668,
+      "grad_norm": 0.31544229388237,
+      "learning_rate": 9.724540558726151e-05,
+      "loss": 0.2623,
+      "step": 219
+    },
+    {
+      "epoch": 1.9555555555555557,
+      "grad_norm": 0.30620795488357544,
+      "learning_rate": 9.719416651541839e-05,
+      "loss": 0.2421,
+      "step": 220
+    },
+    {
+      "epoch": 1.9644444444444444,
+      "grad_norm": 0.38000237941741943,
+      "learning_rate": 9.714246902031006e-05,
+      "loss": 0.237,
+      "step": 221
+    },
+    {
+      "epoch": 1.9733333333333334,
+      "grad_norm": 0.4513542950153351,
+      "learning_rate": 9.709031360410318e-05,
+      "loss": 0.2942,
+      "step": 222
+    },
+    {
+      "epoch": 1.982222222222222,
+      "grad_norm": 0.28410232067108154,
+      "learning_rate": 9.703770077341236e-05,
+      "loss": 0.2918,
+      "step": 223
+    },
+    {
+      "epoch": 1.991111111111111,
+      "grad_norm": 0.29834455251693726,
+      "learning_rate": 9.698463103929542e-05,
+      "loss": 0.2444,
+      "step": 224
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.304579496383667,
+      "learning_rate": 9.693110491724827e-05,
+      "loss": 0.3,
+      "step": 225
+    },
+    {
+      "epoch": 2.008888888888889,
+      "grad_norm": 0.49522674083709717,
+      "learning_rate": 9.687712292719997e-05,
+      "loss": 0.317,
+      "step": 226
+    },
+    {
+      "epoch": 2.017777777777778,
+      "grad_norm": 0.2522561252117157,
+      "learning_rate": 9.682268559350771e-05,
+      "loss": 0.2269,
+      "step": 227
+    },
+    {
+      "epoch": 2.026666666666667,
+      "grad_norm": 0.48555663228034973,
+      "learning_rate": 9.67677934449517e-05,
+      "loss": 0.2827,
+      "step": 228
+    },
+    {
+      "epoch": 2.0355555555555553,
+      "grad_norm": 0.3875712752342224,
+      "learning_rate": 9.671244701472999e-05,
+      "loss": 0.2813,
+      "step": 229
+    },
+    {
+      "epoch": 2.0444444444444443,
+      "grad_norm": 0.2848590910434723,
+      "learning_rate": 9.665664684045333e-05,
+      "loss": 0.2451,
+      "step": 230
+    },
+    {
+      "epoch": 2.0533333333333332,
+      "grad_norm": 0.30722254514694214,
+      "learning_rate": 9.660039346413994e-05,
+      "loss": 0.1899,
+      "step": 231
+    },
+    {
+      "epoch": 2.062222222222222,
+      "grad_norm": 0.22592511773109436,
+      "learning_rate": 9.654368743221022e-05,
+      "loss": 0.2078,
+      "step": 232
+    },
+    {
+      "epoch": 2.071111111111111,
+      "grad_norm": 0.5311526656150818,
+      "learning_rate": 9.648652929548152e-05,
+      "loss": 0.2834,
+      "step": 233
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.28005415201187134,
+      "learning_rate": 9.642891960916268e-05,
+      "loss": 0.2628,
+      "step": 234
+    },
+    {
+      "epoch": 2.088888888888889,
+      "grad_norm": 0.29537031054496765,
+      "learning_rate": 9.637085893284876e-05,
+      "loss": 0.2429,
+      "step": 235
+    },
+    {
+      "epoch": 2.097777777777778,
+      "grad_norm": 0.5397939682006836,
+      "learning_rate": 9.631234783051544e-05,
+      "loss": 0.3032,
+      "step": 236
+    },
+    {
+      "epoch": 2.1066666666666665,
+      "grad_norm": 0.36687228083610535,
+      "learning_rate": 9.625338687051375e-05,
+      "loss": 0.2908,
+      "step": 237
+    },
+    {
+      "epoch": 2.1155555555555554,
+      "grad_norm": 0.33907392621040344,
+      "learning_rate": 9.619397662556435e-05,
+      "loss": 0.2333,
+      "step": 238
+    },
+    {
+      "epoch": 2.1244444444444444,
+      "grad_norm": 0.43153664469718933,
+      "learning_rate": 9.613411767275209e-05,
+      "loss": 0.2341,
+      "step": 239
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.30483999848365784,
+      "learning_rate": 9.607381059352038e-05,
+      "loss": 0.2655,
+      "step": 240
+    },
+    {
+      "epoch": 2.1422222222222222,
+      "grad_norm": 1.9127657413482666,
+      "learning_rate": 9.601305597366554e-05,
+      "loss": 0.4065,
+      "step": 241
+    },
+    {
+      "epoch": 2.151111111111111,
+      "grad_norm": 0.35035258531570435,
+      "learning_rate": 9.595185440333103e-05,
+      "loss": 0.1855,
+      "step": 242
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.5800159573554993,
+      "learning_rate": 9.589020647700191e-05,
+      "loss": 0.2758,
+      "step": 243
+    },
+    {
+      "epoch": 2.168888888888889,
+      "grad_norm": 0.6490089297294617,
+      "learning_rate": 9.582811279349882e-05,
+      "loss": 0.2891,
+      "step": 244
+    },
+    {
+      "epoch": 2.1777777777777776,
+      "grad_norm": 0.46835729479789734,
+      "learning_rate": 9.576557395597236e-05,
+      "loss": 0.2762,
+      "step": 245
+    },
+    {
+      "epoch": 2.1866666666666665,
+      "grad_norm": 0.3493160605430603,
+      "learning_rate": 9.570259057189717e-05,
+      "loss": 0.2785,
+      "step": 246
+    },
+    {
+      "epoch": 2.1955555555555555,
+      "grad_norm": 0.2647961378097534,
+      "learning_rate": 9.563916325306594e-05,
+      "loss": 0.2389,
+      "step": 247
+    },
+    {
+      "epoch": 2.2044444444444444,
+      "grad_norm": 0.3641429543495178,
+      "learning_rate": 9.557529261558367e-05,
+      "loss": 0.2625,
+      "step": 248
+    },
+    {
+      "epoch": 2.2133333333333334,
+      "grad_norm": 0.3188803195953369,
+      "learning_rate": 9.551097927986144e-05,
+      "loss": 0.2501,
+      "step": 249
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.32517799735069275,
+      "learning_rate": 9.544622387061055e-05,
+      "loss": 0.2333,
+      "step": 250
+    },
+    {
+      "epoch": 2.2311111111111113,
+      "grad_norm": 0.36616072058677673,
+      "learning_rate": 9.538102701683643e-05,
+      "loss": 0.2942,
+      "step": 251
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.34478914737701416,
+      "learning_rate": 9.53153893518325e-05,
+      "loss": 0.2435,
+      "step": 252
+    },
+    {
+      "epoch": 2.2488888888888887,
+      "grad_norm": 0.563099205493927,
+      "learning_rate": 9.5249311513174e-05,
+      "loss": 0.2787,
+      "step": 253
+    },
+    {
+      "epoch": 2.2577777777777777,
+      "grad_norm": 0.4205755293369293,
+      "learning_rate": 9.518279414271183e-05,
+      "loss": 0.2974,
+      "step": 254
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 0.32124075293540955,
+      "learning_rate": 9.511583788656632e-05,
+      "loss": 0.3255,
+      "step": 255
+    },
+    {
+      "epoch": 2.2755555555555556,
+      "grad_norm": 0.34654173254966736,
+      "learning_rate": 9.504844339512095e-05,
+      "loss": 0.272,
+      "step": 256
+    },
+    {
+      "epoch": 2.2844444444444445,
+      "grad_norm": 0.3165460526943207,
+      "learning_rate": 9.498061132301601e-05,
+      "loss": 0.2266,
+      "step": 257
+    },
+    {
+      "epoch": 2.2933333333333334,
+      "grad_norm": 0.3656767010688782,
+      "learning_rate": 9.491234232914221e-05,
+      "loss": 0.3019,
+      "step": 258
+    },
+    {
+      "epoch": 2.3022222222222224,
+      "grad_norm": 0.28812313079833984,
+      "learning_rate": 9.484363707663442e-05,
+      "loss": 0.2631,
+      "step": 259
+    },
+    {
+      "epoch": 2.311111111111111,
+      "grad_norm": 0.5799005031585693,
+      "learning_rate": 9.477449623286505e-05,
+      "loss": 0.2721,
+      "step": 260
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.332253098487854,
+      "learning_rate": 9.470492046943771e-05,
+      "loss": 0.2335,
+      "step": 261
+    },
+    {
+      "epoch": 2.328888888888889,
+      "grad_norm": 0.39480510354042053,
+      "learning_rate": 9.463491046218058e-05,
+      "loss": 0.2565,
+      "step": 262
+    },
+    {
+      "epoch": 2.3377777777777777,
+      "grad_norm": 0.5974012017250061,
+      "learning_rate": 9.456446689113992e-05,
+      "loss": 0.286,
+      "step": 263
+    },
+    {
+      "epoch": 2.3466666666666667,
+      "grad_norm": 0.47316139936447144,
+      "learning_rate": 9.449359044057345e-05,
+      "loss": 0.2111,
+      "step": 264
+    },
+    {
+      "epoch": 2.3555555555555556,
+      "grad_norm": 0.5243270993232727,
+      "learning_rate": 9.442228179894362e-05,
+      "loss": 0.2256,
+      "step": 265
+    },
+    {
+      "epoch": 2.3644444444444446,
+      "grad_norm": 0.6339777112007141,
+      "learning_rate": 9.435054165891109e-05,
+      "loss": 0.2635,
+      "step": 266
+    },
+    {
+      "epoch": 2.3733333333333335,
+      "grad_norm": 0.31498321890830994,
+      "learning_rate": 9.427837071732783e-05,
+      "loss": 0.2504,
+      "step": 267
+    },
+    {
+      "epoch": 2.3822222222222225,
+      "grad_norm": 1.1357451677322388,
+      "learning_rate": 9.420576967523049e-05,
+      "loss": 0.3394,
+      "step": 268
+    },
+    {
+      "epoch": 2.391111111111111,
+      "grad_norm": 0.3476477265357971,
+      "learning_rate": 9.413273923783346e-05,
+      "loss": 0.241,
+      "step": 269
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.3083588778972626,
+      "learning_rate": 9.405928011452211e-05,
+      "loss": 0.2366,
+      "step": 270
+    },
+    {
+      "epoch": 2.408888888888889,
+      "grad_norm": 0.45807912945747375,
+      "learning_rate": 9.398539301884592e-05,
+      "loss": 0.2271,
+      "step": 271
+    },
+    {
+      "epoch": 2.417777777777778,
+      "grad_norm": 0.28123828768730164,
+      "learning_rate": 9.391107866851143e-05,
+      "loss": 0.2591,
+      "step": 272
+    },
+    {
+      "epoch": 2.4266666666666667,
+      "grad_norm": 0.45137813687324524,
+      "learning_rate": 9.38363377853754e-05,
+      "loss": 0.2254,
+      "step": 273
+    },
+    {
+      "epoch": 2.4355555555555557,
+      "grad_norm": 0.38635390996932983,
+      "learning_rate": 9.376117109543769e-05,
+      "loss": 0.2498,
+      "step": 274
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.36426419019699097,
+      "learning_rate": 9.368557932883432e-05,
+      "loss": 0.2346,
+      "step": 275
+    },
+    {
+      "epoch": 2.453333333333333,
+      "grad_norm": 0.5773187875747681,
+      "learning_rate": 9.360956321983028e-05,
+      "loss": 0.2479,
+      "step": 276
+    },
+    {
+      "epoch": 2.462222222222222,
+      "grad_norm": 0.4950138330459595,
+      "learning_rate": 9.353312350681242e-05,
+      "loss": 0.2108,
+      "step": 277
+    },
+    {
+      "epoch": 2.471111111111111,
+      "grad_norm": 0.599851131439209,
+      "learning_rate": 9.345626093228233e-05,
+      "loss": 0.2419,
+      "step": 278
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.6381746530532837,
+      "learning_rate": 9.337897624284906e-05,
+      "loss": 0.2769,
+      "step": 279
+    },
+    {
+      "epoch": 2.488888888888889,
+      "grad_norm": 0.28884226083755493,
+      "learning_rate": 9.330127018922194e-05,
+      "loss": 0.2746,
+      "step": 280
+    },
+    {
+      "epoch": 2.497777777777778,
+      "grad_norm": 0.3661234378814697,
+      "learning_rate": 9.322314352620318e-05,
+      "loss": 0.2438,
+      "step": 281
+    },
+    {
+      "epoch": 2.506666666666667,
+      "grad_norm": 0.4173062741756439,
+      "learning_rate": 9.314459701268065e-05,
+      "loss": 0.2864,
+      "step": 282
+    },
+    {
+      "epoch": 2.5155555555555553,
+      "grad_norm": 0.6054656505584717,
+      "learning_rate": 9.306563141162046e-05,
+      "loss": 0.2368,
+      "step": 283
+    },
+    {
+      "epoch": 2.5244444444444447,
+      "grad_norm": 0.3535010814666748,
+      "learning_rate": 9.298624749005951e-05,
+      "loss": 0.2807,
+      "step": 284
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 0.3705323338508606,
+      "learning_rate": 9.290644601909815e-05,
+      "loss": 0.2116,
+      "step": 285
+    },
+    {
+      "epoch": 2.542222222222222,
+      "grad_norm": 0.3306547701358795,
+      "learning_rate": 9.282622777389258e-05,
+      "loss": 0.2345,
+      "step": 286
+    },
+    {
+      "epoch": 2.551111111111111,
+      "grad_norm": 0.40624353289604187,
+      "learning_rate": 9.274559353364734e-05,
+      "loss": 0.268,
+      "step": 287
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.4179963767528534,
+      "learning_rate": 9.266454408160779e-05,
+      "loss": 0.2457,
+      "step": 288
+    },
+    {
+      "epoch": 2.568888888888889,
+      "grad_norm": 0.3766705393791199,
+      "learning_rate": 9.258308020505247e-05,
+      "loss": 0.2422,
+      "step": 289
+    },
+    {
+      "epoch": 2.5777777777777775,
+      "grad_norm": 0.26529139280319214,
+      "learning_rate": 9.250120269528546e-05,
+      "loss": 0.2562,
+      "step": 290
+    },
+    {
+      "epoch": 2.586666666666667,
+      "grad_norm": 0.5525755286216736,
+      "learning_rate": 9.241891234762869e-05,
+      "loss": 0.2836,
+      "step": 291
+    },
+    {
+      "epoch": 2.5955555555555554,
+      "grad_norm": 0.40089136362075806,
+      "learning_rate": 9.233620996141421e-05,
+      "loss": 0.2623,
+      "step": 292
+    },
+    {
+      "epoch": 2.6044444444444443,
+      "grad_norm": 0.3758887052536011,
+      "learning_rate": 9.225309633997641e-05,
+      "loss": 0.2118,
+      "step": 293
+    },
+    {
+      "epoch": 2.6133333333333333,
+      "grad_norm": 0.2941136062145233,
+      "learning_rate": 9.21695722906443e-05,
+      "loss": 0.2449,
+      "step": 294
+    },
+    {
+      "epoch": 2.6222222222222222,
+      "grad_norm": 0.32856640219688416,
+      "learning_rate": 9.208563862473351e-05,
+      "loss": 0.2308,
+      "step": 295
+    },
+    {
+      "epoch": 2.631111111111111,
+      "grad_norm": 0.3801966905593872,
+      "learning_rate": 9.200129615753859e-05,
+      "loss": 0.2727,
+      "step": 296
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.5766463875770569,
+      "learning_rate": 9.191654570832496e-05,
+      "loss": 0.2068,
+      "step": 297
+    },
+    {
+      "epoch": 2.648888888888889,
+      "grad_norm": 0.2755817174911499,
+      "learning_rate": 9.183138810032099e-05,
+      "loss": 0.2436,
+      "step": 298
+    },
+    {
+      "epoch": 2.6577777777777776,
+      "grad_norm": 0.292676717042923,
+      "learning_rate": 9.174582416071007e-05,
+      "loss": 0.2379,
+      "step": 299
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.5588552951812744,
+      "learning_rate": 9.165985472062246e-05,
+      "loss": 0.3178,
+      "step": 300
+    },
+    {
+      "epoch": 2.6755555555555555,
+      "grad_norm": 0.3965974748134613,
+      "learning_rate": 9.157348061512727e-05,
+      "loss": 0.2575,
+      "step": 301
+    },
+    {
+      "epoch": 2.6844444444444444,
+      "grad_norm": 0.45277366042137146,
+      "learning_rate": 9.148670268322438e-05,
+      "loss": 0.2235,
+      "step": 302
+    },
+    {
+      "epoch": 2.6933333333333334,
+      "grad_norm": 0.34740129113197327,
+      "learning_rate": 9.139952176783626e-05,
+      "loss": 0.2199,
+      "step": 303
+    },
+    {
+      "epoch": 2.7022222222222223,
+      "grad_norm": 0.280239075422287,
+      "learning_rate": 9.131193871579975e-05,
+      "loss": 0.2442,
+      "step": 304
+    },
+    {
+      "epoch": 2.7111111111111112,
+      "grad_norm": 0.3178425133228302,
+      "learning_rate": 9.12239543778579e-05,
+      "loss": 0.2262,
+      "step": 305
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 0.421599417924881,
+      "learning_rate": 9.113556960865167e-05,
+      "loss": 0.2326,
+      "step": 306
+    },
+    {
+      "epoch": 2.728888888888889,
+      "grad_norm": 0.25511476397514343,
+      "learning_rate": 9.104678526671162e-05,
+      "loss": 0.2261,
+      "step": 307
+    },
+    {
+      "epoch": 2.7377777777777776,
+      "grad_norm": 0.48630788922309875,
+      "learning_rate": 9.09576022144496e-05,
+      "loss": 0.2077,
+      "step": 308
+    },
+    {
+      "epoch": 2.7466666666666666,
+      "grad_norm": 0.38153597712516785,
+      "learning_rate": 9.086802131815031e-05,
+      "loss": 0.2797,
+      "step": 309
+    },
+    {
+      "epoch": 2.7555555555555555,
+      "grad_norm": 0.5679035186767578,
+      "learning_rate": 9.077804344796302e-05,
+      "loss": 0.2584,
+      "step": 310
+    },
+    {
+      "epoch": 2.7644444444444445,
+      "grad_norm": 0.5588884949684143,
+      "learning_rate": 9.068766947789292e-05,
+      "loss": 0.2838,
+      "step": 311
+    },
+    {
+      "epoch": 2.7733333333333334,
+      "grad_norm": 0.40492427349090576,
+      "learning_rate": 9.059690028579284e-05,
+      "loss": 0.2037,
+      "step": 312
+    },
+    {
+      "epoch": 2.7822222222222224,
+      "grad_norm": 0.3332842290401459,
+      "learning_rate": 9.050573675335453e-05,
+      "loss": 0.2567,
+      "step": 313
+    },
+    {
+      "epoch": 2.7911111111111113,
+      "grad_norm": 0.29908373951911926,
+      "learning_rate": 9.041417976610027e-05,
+      "loss": 0.2315,
+      "step": 314
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.31926652789115906,
+      "learning_rate": 9.032223021337414e-05,
+      "loss": 0.2138,
+      "step": 315
+    },
+    {
+      "epoch": 2.8088888888888888,
+      "grad_norm": 0.31083905696868896,
+      "learning_rate": 9.022988898833342e-05,
+      "loss": 0.211,
+      "step": 316
+    },
+    {
+      "epoch": 2.8177777777777777,
+      "grad_norm": 0.3902362287044525,
+      "learning_rate": 9.013715698793996e-05,
+      "loss": 0.2503,
+      "step": 317
+    },
+    {
+      "epoch": 2.8266666666666667,
+      "grad_norm": 0.4482881724834442,
+      "learning_rate": 9.004403511295141e-05,
+      "loss": 0.2204,
+      "step": 318
+    },
+    {
+      "epoch": 2.8355555555555556,
+      "grad_norm": 1.0134060382843018,
+      "learning_rate": 8.995052426791247e-05,
+      "loss": 0.2454,
+      "step": 319
+    },
+    {
+      "epoch": 2.8444444444444446,
+      "grad_norm": 0.4079667329788208,
+      "learning_rate": 8.985662536114613e-05,
+      "loss": 0.2865,
+      "step": 320
+    },
+    {
+      "epoch": 2.8533333333333335,
+      "grad_norm": 0.34580424427986145,
+      "learning_rate": 8.976233930474486e-05,
+      "loss": 0.2193,
+      "step": 321
+    },
+    {
+      "epoch": 2.862222222222222,
+      "grad_norm": 0.35853755474090576,
+      "learning_rate": 8.966766701456177e-05,
+      "loss": 0.2596,
+      "step": 322
+    },
+    {
+      "epoch": 2.871111111111111,
+      "grad_norm": 0.6029608249664307,
+      "learning_rate": 8.957260941020154e-05,
+      "loss": 0.264,
+      "step": 323
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.5026920437812805,
+      "learning_rate": 8.947716741501177e-05,
+      "loss": 0.2566,
+      "step": 324
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.5039583444595337,
+      "learning_rate": 8.938134195607377e-05,
+      "loss": 0.2266,
+      "step": 325
+    },
+    {
+      "epoch": 2.897777777777778,
+      "grad_norm": 0.4172382652759552,
+      "learning_rate": 8.928513396419368e-05,
+      "loss": 0.251,
+      "step": 326
+    },
+    {
+      "epoch": 2.9066666666666667,
+      "grad_norm": 0.48238590359687805,
+      "learning_rate": 8.918854437389342e-05,
+      "loss": 0.2583,
+      "step": 327
+    },
+    {
+      "epoch": 2.9155555555555557,
+      "grad_norm": 0.37107643485069275,
+      "learning_rate": 8.90915741234015e-05,
+      "loss": 0.2403,
+      "step": 328
+    },
+    {
+      "epoch": 2.924444444444444,
+      "grad_norm": 0.26561880111694336,
+      "learning_rate": 8.899422415464409e-05,
+      "loss": 0.233,
+      "step": 329
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 0.43001478910446167,
+      "learning_rate": 8.889649541323574e-05,
+      "loss": 0.2598,
+      "step": 330
+    },
+    {
+      "epoch": 2.942222222222222,
+      "grad_norm": 0.3992827832698822,
+      "learning_rate": 8.879838884847025e-05,
+      "loss": 0.2151,
+      "step": 331
+    },
+    {
+      "epoch": 2.951111111111111,
+      "grad_norm": 0.44580769538879395,
+      "learning_rate": 8.869990541331138e-05,
+      "loss": 0.2296,
+      "step": 332
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3314950168132782,
+      "learning_rate": 8.860104606438369e-05,
+      "loss": 0.2331,
+      "step": 333
+    },
+    {
+      "epoch": 2.968888888888889,
+      "grad_norm": 0.6843854784965515,
+      "learning_rate": 8.850181176196315e-05,
+      "loss": 0.3505,
+      "step": 334
+    },
+    {
+      "epoch": 2.977777777777778,
+      "grad_norm": 0.2698711156845093,
+      "learning_rate": 8.840220346996792e-05,
+      "loss": 0.2272,
+      "step": 335
+    },
+    {
+      "epoch": 2.986666666666667,
+      "grad_norm": 0.26144105195999146,
+      "learning_rate": 8.83022221559489e-05,
+      "loss": 0.2438,
+      "step": 336
+    },
+    {
+      "epoch": 2.9955555555555557,
+      "grad_norm": 0.26964858174324036,
+      "learning_rate": 8.820186879108038e-05,
+      "loss": 0.2434,
+      "step": 337
+    },
+    {
+      "epoch": 3.0044444444444443,
+      "grad_norm": 0.2074783891439438,
+      "learning_rate": 8.810114435015054e-05,
+      "loss": 0.2438,
+      "step": 338
+    },
+    {
+      "epoch": 3.013333333333333,
+      "grad_norm": 0.5041800737380981,
+      "learning_rate": 8.800004981155208e-05,
+      "loss": 0.2566,
+      "step": 339
+    },
+    {
+      "epoch": 3.022222222222222,
+      "grad_norm": 0.24667270481586456,
+      "learning_rate": 8.789858615727265e-05,
+      "loss": 0.2366,
+      "step": 340
+    },
+    {
+      "epoch": 3.031111111111111,
+      "grad_norm": 0.3841206729412079,
+      "learning_rate": 8.779675437288532e-05,
+      "loss": 0.2371,
+      "step": 341
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.23301474750041962,
+      "learning_rate": 8.7694555447539e-05,
+      "loss": 0.1969,
+      "step": 342
+    },
+    {
+      "epoch": 3.048888888888889,
+      "grad_norm": 0.26055586338043213,
+      "learning_rate": 8.759199037394887e-05,
+      "loss": 0.2064,
+      "step": 343
+    },
+    {
+      "epoch": 3.057777777777778,
+      "grad_norm": 0.4642762243747711,
+      "learning_rate": 8.748906014838672e-05,
+      "loss": 0.1877,
+      "step": 344
+    },
+    {
+      "epoch": 3.066666666666667,
+      "grad_norm": 0.28164124488830566,
+      "learning_rate": 8.738576577067122e-05,
+      "loss": 0.2482,
+      "step": 345
+    },
+    {
+      "epoch": 3.0755555555555554,
+      "grad_norm": 0.6319877505302429,
+      "learning_rate": 8.728210824415827e-05,
+      "loss": 0.196,
+      "step": 346
+    },
+    {
+      "epoch": 3.0844444444444443,
+      "grad_norm": 0.3142436146736145,
+      "learning_rate": 8.717808857573131e-05,
+      "loss": 0.1907,
+      "step": 347
+    },
+    {
+      "epoch": 3.0933333333333333,
+      "grad_norm": 0.3424491882324219,
+      "learning_rate": 8.707370777579133e-05,
+      "loss": 0.228,
+      "step": 348
+    },
+    {
+      "epoch": 3.102222222222222,
+      "grad_norm": 0.34540361166000366,
+      "learning_rate": 8.696896685824731e-05,
+      "loss": 0.2292,
+      "step": 349
+    },
+    {
+      "epoch": 3.111111111111111,
+      "grad_norm": 0.38283613324165344,
+      "learning_rate": 8.68638668405062e-05,
+      "loss": 0.1662,
+      "step": 350
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.4507913887500763,
+      "learning_rate": 8.67584087434631e-05,
+      "loss": 0.2408,
+      "step": 351
+    },
+    {
+      "epoch": 3.128888888888889,
+      "grad_norm": 1.011723518371582,
+      "learning_rate": 8.665259359149132e-05,
+      "loss": 0.2772,
+      "step": 352
+    },
+    {
+      "epoch": 3.137777777777778,
+      "grad_norm": 0.5270500779151917,
+      "learning_rate": 8.654642241243247e-05,
+      "loss": 0.2017,
+      "step": 353
+    },
+    {
+      "epoch": 3.1466666666666665,
+      "grad_norm": 0.8714268207550049,
+      "learning_rate": 8.643989623758643e-05,
+      "loss": 0.2311,
+      "step": 354
+    },
+    {
+      "epoch": 3.1555555555555554,
+      "grad_norm": 0.2898414134979248,
+      "learning_rate": 8.633301610170135e-05,
+      "loss": 0.2443,
+      "step": 355
+    },
+    {
+      "epoch": 3.1644444444444444,
+      "grad_norm": 0.6308351159095764,
+      "learning_rate": 8.622578304296364e-05,
+      "loss": 0.2517,
+      "step": 356
+    },
+    {
+      "epoch": 3.1733333333333333,
+      "grad_norm": 0.5109901428222656,
+      "learning_rate": 8.611819810298778e-05,
+      "loss": 0.2269,
+      "step": 357
+    },
+    {
+      "epoch": 3.1822222222222223,
+      "grad_norm": 0.432049959897995,
+      "learning_rate": 8.601026232680634e-05,
+      "loss": 0.1978,
+      "step": 358
+    },
+    {
+      "epoch": 3.1911111111111112,
+      "grad_norm": 0.3804490566253662,
+      "learning_rate": 8.59019767628597e-05,
+      "loss": 0.2082,
+      "step": 359
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.31701037287712097,
+      "learning_rate": 8.579334246298593e-05,
+      "loss": 0.2356,
+      "step": 360
+    },
+    {
+      "epoch": 3.2088888888888887,
+      "grad_norm": 0.25454384088516235,
+      "learning_rate": 8.56843604824106e-05,
+      "loss": 0.1967,
+      "step": 361
+    },
+    {
+      "epoch": 3.2177777777777776,
+      "grad_norm": 0.3782944977283478,
+      "learning_rate": 8.557503187973651e-05,
+      "loss": 0.239,
+      "step": 362
+    },
+    {
+      "epoch": 3.2266666666666666,
+      "grad_norm": 0.6260755062103271,
+      "learning_rate": 8.546535771693334e-05,
+      "loss": 0.2061,
+      "step": 363
+    },
+    {
+      "epoch": 3.2355555555555555,
+      "grad_norm": 0.3981928825378418,
+      "learning_rate": 8.535533905932738e-05,
+      "loss": 0.2094,
+      "step": 364
+    },
+    {
+      "epoch": 3.2444444444444445,
+      "grad_norm": 0.5317679643630981,
+      "learning_rate": 8.524497697559126e-05,
+      "loss": 0.2448,
+      "step": 365
+    },
+    {
+      "epoch": 3.2533333333333334,
+      "grad_norm": 0.45160210132598877,
+      "learning_rate": 8.513427253773346e-05,
+      "loss": 0.2451,
+      "step": 366
+    },
+    {
+      "epoch": 3.2622222222222224,
+      "grad_norm": 0.5067645907402039,
+      "learning_rate": 8.502322682108792e-05,
+      "loss": 0.2036,
+      "step": 367
+    },
+    {
+      "epoch": 3.2711111111111113,
+      "grad_norm": 0.3842930495738983,
+      "learning_rate": 8.491184090430364e-05,
+      "loss": 0.2056,
+      "step": 368
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 0.33578169345855713,
+      "learning_rate": 8.480011586933418e-05,
+      "loss": 0.2292,
+      "step": 369
+    },
+    {
+      "epoch": 3.2888888888888888,
+      "grad_norm": 0.4589829444885254,
+      "learning_rate": 8.468805280142709e-05,
+      "loss": 0.2026,
+      "step": 370
+    },
+    {
+      "epoch": 3.2977777777777777,
+      "grad_norm": 0.666786789894104,
+      "learning_rate": 8.457565278911348e-05,
+      "loss": 0.2031,
+      "step": 371
+    },
+    {
+      "epoch": 3.3066666666666666,
+      "grad_norm": 0.3466854393482208,
+      "learning_rate": 8.446291692419736e-05,
+      "loss": 0.2024,
+      "step": 372
+    },
+    {
+      "epoch": 3.3155555555555556,
+      "grad_norm": 0.4151984453201294,
+      "learning_rate": 8.434984630174509e-05,
+      "loss": 0.2071,
+      "step": 373
+    },
+    {
+      "epoch": 3.3244444444444445,
+      "grad_norm": 0.3907124102115631,
+      "learning_rate": 8.423644202007467e-05,
+      "loss": 0.199,
+      "step": 374
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.7014510035514832,
+      "learning_rate": 8.412270518074518e-05,
+      "loss": 0.2345,
+      "step": 375
+    },
+    {
+      "epoch": 3.3422222222222224,
+      "grad_norm": 0.3714005649089813,
+      "learning_rate": 8.400863688854597e-05,
+      "loss": 0.2272,
+      "step": 376
+    },
+    {
+      "epoch": 3.351111111111111,
+      "grad_norm": 0.9573063254356384,
+      "learning_rate": 8.389423825148598e-05,
+      "loss": 0.2531,
+      "step": 377
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.5541326999664307,
+      "learning_rate": 8.377951038078302e-05,
+      "loss": 0.2667,
+      "step": 378
+    },
+    {
+      "epoch": 3.368888888888889,
+      "grad_norm": 0.5363870859146118,
+      "learning_rate": 8.366445439085286e-05,
+      "loss": 0.2028,
+      "step": 379
+    },
+    {
+      "epoch": 3.3777777777777778,
+      "grad_norm": 0.4728504717350006,
+      "learning_rate": 8.354907139929851e-05,
+      "loss": 0.1955,
+      "step": 380
+    },
+    {
+      "epoch": 3.3866666666666667,
+      "grad_norm": 0.46801096200942993,
+      "learning_rate": 8.343336252689935e-05,
+      "loss": 0.1981,
+      "step": 381
+    },
+    {
+      "epoch": 3.3955555555555557,
+      "grad_norm": 0.528943657875061,
+      "learning_rate": 8.33173288976002e-05,
+      "loss": 0.2185,
+      "step": 382
+    },
+    {
+      "epoch": 3.4044444444444446,
+      "grad_norm": 0.7669292092323303,
+      "learning_rate": 8.320097163850043e-05,
+      "loss": 0.264,
+      "step": 383
+    },
+    {
+      "epoch": 3.413333333333333,
+      "grad_norm": 0.46919432282447815,
+      "learning_rate": 8.308429187984297e-05,
+      "loss": 0.1932,
+      "step": 384
+    },
+    {
+      "epoch": 3.422222222222222,
+      "grad_norm": 0.5307047367095947,
+      "learning_rate": 8.296729075500344e-05,
+      "loss": 0.1948,
+      "step": 385
+    },
+    {
+      "epoch": 3.431111111111111,
+      "grad_norm": 0.4287688136100769,
+      "learning_rate": 8.284996940047903e-05,
+      "loss": 0.2424,
+      "step": 386
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.44700077176094055,
+      "learning_rate": 8.273232895587748e-05,
+      "loss": 0.2041,
+      "step": 387
+    },
+    {
+      "epoch": 3.448888888888889,
+      "grad_norm": 0.5288101434707642,
+      "learning_rate": 8.261437056390606e-05,
+      "loss": 0.2205,
+      "step": 388
+    },
+    {
+      "epoch": 3.457777777777778,
+      "grad_norm": 0.4757843613624573,
+      "learning_rate": 8.249609537036043e-05,
+      "loss": 0.2398,
+      "step": 389
+    },
+    {
+      "epoch": 3.466666666666667,
+      "grad_norm": 0.2801768481731415,
+      "learning_rate": 8.237750452411353e-05,
+      "loss": 0.187,
+      "step": 390
+    },
+    {
+      "epoch": 3.4755555555555557,
+      "grad_norm": 0.35519644618034363,
+      "learning_rate": 8.225859917710439e-05,
+      "loss": 0.2045,
+      "step": 391
+    },
+    {
+      "epoch": 3.4844444444444447,
+      "grad_norm": 0.4472619891166687,
+      "learning_rate": 8.213938048432697e-05,
+      "loss": 0.21,
+      "step": 392
+    },
+    {
+      "epoch": 3.493333333333333,
+      "grad_norm": 0.7717186808586121,
+      "learning_rate": 8.201984960381894e-05,
+      "loss": 0.2442,
+      "step": 393
+    },
+    {
+      "epoch": 3.502222222222222,
+      "grad_norm": 0.5858293771743774,
+      "learning_rate": 8.190000769665044e-05,
+      "loss": 0.2072,
+      "step": 394
+    },
+    {
+      "epoch": 3.511111111111111,
+      "grad_norm": 0.4433729648590088,
+      "learning_rate": 8.177985592691272e-05,
+      "loss": 0.2933,
+      "step": 395
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.47395846247673035,
+      "learning_rate": 8.1659395461707e-05,
+      "loss": 0.218,
+      "step": 396
+    },
+    {
+      "epoch": 3.528888888888889,
+      "grad_norm": 0.4193160831928253,
+      "learning_rate": 8.153862747113292e-05,
+      "loss": 0.2046,
+      "step": 397
+    },
+    {
+      "epoch": 3.537777777777778,
+      "grad_norm": 0.40366363525390625,
+      "learning_rate": 8.141755312827736e-05,
+      "loss": 0.2022,
+      "step": 398
+    },
+    {
+      "epoch": 3.546666666666667,
+      "grad_norm": 0.543988823890686,
+      "learning_rate": 8.129617360920296e-05,
+      "loss": 0.1818,
+      "step": 399
+    },
+    {
+      "epoch": 3.5555555555555554,
+      "grad_norm": 0.3089783191680908,
+      "learning_rate": 8.117449009293668e-05,
+      "loss": 0.2196,
+      "step": 400
+    },
+    {
+      "epoch": 3.5644444444444443,
+      "grad_norm": 0.399312287569046,
+      "learning_rate": 8.10525037614584e-05,
+      "loss": 0.1889,
+      "step": 401
+    },
+    {
+      "epoch": 3.5733333333333333,
+      "grad_norm": 0.3407880663871765,
+      "learning_rate": 8.093021579968941e-05,
+      "loss": 0.2562,
+      "step": 402
+    },
+    {
+      "epoch": 3.582222222222222,
+      "grad_norm": 0.8019405603408813,
+      "learning_rate": 8.080762739548089e-05,
+      "loss": 0.2453,
+      "step": 403
+    },
+    {
+      "epoch": 3.591111111111111,
+      "grad_norm": 0.308278352022171,
+      "learning_rate": 8.068473973960238e-05,
+      "loss": 0.2129,
+      "step": 404
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.5972309112548828,
+      "learning_rate": 8.056155402573024e-05,
+      "loss": 0.2369,
+      "step": 405
+    },
+    {
+      "epoch": 3.608888888888889,
+      "grad_norm": 0.40940365195274353,
+      "learning_rate": 8.043807145043604e-05,
+      "loss": 0.2144,
+      "step": 406
+    },
+    {
+      "epoch": 3.6177777777777775,
+      "grad_norm": 0.3109559714794159,
+      "learning_rate": 8.03142932131749e-05,
+      "loss": 0.2189,
+      "step": 407
+    },
+    {
+      "epoch": 3.626666666666667,
+      "grad_norm": 0.3607446551322937,
+      "learning_rate": 8.019022051627388e-05,
+      "loss": 0.2197,
+      "step": 408
+    },
+    {
+      "epoch": 3.6355555555555554,
+      "grad_norm": 0.4839872419834137,
+      "learning_rate": 8.006585456492029e-05,
+      "loss": 0.2797,
+      "step": 409
+    },
+    {
+      "epoch": 3.6444444444444444,
+      "grad_norm": 0.33101025223731995,
+      "learning_rate": 7.994119656715002e-05,
+      "loss": 0.2042,
+      "step": 410
+    },
+    {
+      "epoch": 3.6533333333333333,
+      "grad_norm": 0.558871865272522,
+      "learning_rate": 7.981624773383572e-05,
+      "loss": 0.2372,
+      "step": 411
+    },
+    {
+      "epoch": 3.6622222222222223,
+      "grad_norm": 0.29825037717819214,
+      "learning_rate": 7.969100927867507e-05,
+      "loss": 0.2156,
+      "step": 412
+    },
+    {
+      "epoch": 3.671111111111111,
+      "grad_norm": 0.4881444573402405,
+      "learning_rate": 7.956548241817912e-05,
+      "loss": 0.177,
+      "step": 413
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.3237267732620239,
+      "learning_rate": 7.943966837166023e-05,
+      "loss": 0.2089,
+      "step": 414
+    },
+    {
+      "epoch": 3.688888888888889,
+      "grad_norm": 0.32327884435653687,
+      "learning_rate": 7.931356836122046e-05,
+      "loss": 0.2082,
+      "step": 415
+    },
+    {
+      "epoch": 3.6977777777777776,
+      "grad_norm": 0.4469964802265167,
+      "learning_rate": 7.91871836117395e-05,
+      "loss": 0.1687,
+      "step": 416
+    },
+    {
+      "epoch": 3.7066666666666666,
+      "grad_norm": 0.29894670844078064,
+      "learning_rate": 7.906051535086296e-05,
+      "loss": 0.1816,
+      "step": 417
+    },
+    {
+      "epoch": 3.7155555555555555,
+      "grad_norm": 0.3552427291870117,
+      "learning_rate": 7.89335648089903e-05,
+      "loss": 0.2157,
+      "step": 418
+    },
+    {
+      "epoch": 3.7244444444444444,
+      "grad_norm": 0.6162396669387817,
+      "learning_rate": 7.880633321926294e-05,
+      "loss": 0.1988,
+      "step": 419
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 0.4064711332321167,
+      "learning_rate": 7.86788218175523e-05,
+      "loss": 0.2438,
+      "step": 420
+    },
+    {
+      "epoch": 3.7422222222222223,
+      "grad_norm": 0.634732723236084,
+      "learning_rate": 7.855103184244776e-05,
+      "loss": 0.266,
+      "step": 421
+    },
+    {
+      "epoch": 3.7511111111111113,
+      "grad_norm": 1.4341474771499634,
+      "learning_rate": 7.842296453524463e-05,
+      "loss": 0.3107,
+      "step": 422
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.49320435523986816,
+      "learning_rate": 7.829462113993207e-05,
+      "loss": 0.2334,
+      "step": 423
+    },
+    {
+      "epoch": 3.7688888888888887,
+      "grad_norm": 0.39640945196151733,
+      "learning_rate": 7.81660029031811e-05,
+      "loss": 0.2418,
+      "step": 424
+    },
+    {
+      "epoch": 3.7777777777777777,
+      "grad_norm": 0.6148217916488647,
+      "learning_rate": 7.80371110743324e-05,
+      "loss": 0.2023,
+      "step": 425
+    },
+    {
+      "epoch": 3.7866666666666666,
+      "grad_norm": 0.3857884109020233,
+      "learning_rate": 7.79079469053842e-05,
+      "loss": 0.2146,
+      "step": 426
+    },
+    {
+      "epoch": 3.7955555555555556,
+      "grad_norm": 0.38891321420669556,
+      "learning_rate": 7.777851165098012e-05,
+      "loss": 0.1895,
+      "step": 427
+    },
+    {
+      "epoch": 3.8044444444444445,
+      "grad_norm": 0.3949791491031647,
+      "learning_rate": 7.764880656839696e-05,
+      "loss": 0.1818,
+      "step": 428
+    },
+    {
+      "epoch": 3.8133333333333335,
+      "grad_norm": 0.36369502544403076,
+      "learning_rate": 7.751883291753262e-05,
+      "loss": 0.2442,
+      "step": 429
+    },
+    {
+      "epoch": 3.822222222222222,
+      "grad_norm": 0.39478230476379395,
+      "learning_rate": 7.738859196089358e-05,
+      "loss": 0.1618,
+      "step": 430
+    },
+    {
+      "epoch": 3.8311111111111114,
+      "grad_norm": 0.34515222907066345,
+      "learning_rate": 7.725808496358295e-05,
+      "loss": 0.225,
+      "step": 431
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.5892112851142883,
+      "learning_rate": 7.712731319328798e-05,
+      "loss": 0.2617,
+      "step": 432
+    },
+    {
+      "epoch": 3.848888888888889,
+      "grad_norm": 0.44167014956474304,
+      "learning_rate": 7.699627792026783e-05,
+      "loss": 0.2044,
+      "step": 433
+    },
+    {
+      "epoch": 3.8577777777777778,
+      "grad_norm": 0.3467167615890503,
+      "learning_rate": 7.68649804173412e-05,
+      "loss": 0.2429,
+      "step": 434
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "grad_norm": 0.5901280045509338,
+      "learning_rate": 7.673342195987397e-05,
+      "loss": 0.1852,
+      "step": 435
+    },
+    {
+      "epoch": 3.8755555555555556,
+      "grad_norm": 0.47540798783302307,
+      "learning_rate": 7.660160382576683e-05,
+      "loss": 0.1873,
+      "step": 436
+    },
+    {
+      "epoch": 3.8844444444444446,
+      "grad_norm": 0.546599268913269,
+      "learning_rate": 7.646952729544284e-05,
+      "loss": 0.2246,
+      "step": 437
+    },
+    {
+      "epoch": 3.8933333333333335,
+      "grad_norm": 0.3890751898288727,
+      "learning_rate": 7.633719365183504e-05,
+      "loss": 0.2376,
+      "step": 438
+    },
+    {
+      "epoch": 3.902222222222222,
+      "grad_norm": 0.5732626914978027,
+      "learning_rate": 7.620460418037388e-05,
+      "loss": 0.2012,
+      "step": 439
+    },
+    {
+      "epoch": 3.911111111111111,
+      "grad_norm": 0.5902274250984192,
+      "learning_rate": 7.60717601689749e-05,
+      "loss": 0.2548,
+      "step": 440
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.5472671985626221,
+      "learning_rate": 7.593866290802608e-05,
+      "loss": 0.203,
+      "step": 441
+    },
+    {
+      "epoch": 3.928888888888889,
+      "grad_norm": 1.1805832386016846,
+      "learning_rate": 7.580531369037533e-05,
+      "loss": 0.2373,
+      "step": 442
+    },
+    {
+      "epoch": 3.937777777777778,
+      "grad_norm": 0.4900795817375183,
+      "learning_rate": 7.567171381131802e-05,
+      "loss": 0.2181,
+      "step": 443
+    },
+    {
+      "epoch": 3.9466666666666668,
+      "grad_norm": 0.35954034328460693,
+      "learning_rate": 7.553786456858429e-05,
+      "loss": 0.205,
+      "step": 444
+    },
+    {
+      "epoch": 3.9555555555555557,
+      "grad_norm": 0.6151618361473083,
+      "learning_rate": 7.540376726232648e-05,
+      "loss": 0.216,
+      "step": 445
+    },
+    {
+      "epoch": 3.964444444444444,
+      "grad_norm": 0.6646876335144043,
+      "learning_rate": 7.526942319510655e-05,
+      "loss": 0.2201,
+      "step": 446
+    },
+    {
+      "epoch": 3.9733333333333336,
+      "grad_norm": 0.3377375602722168,
+      "learning_rate": 7.513483367188335e-05,
+      "loss": 0.2397,
+      "step": 447
+    },
+    {
+      "epoch": 3.982222222222222,
+      "grad_norm": 0.44786104559898376,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.2128,
+      "step": 448
+    },
+    {
+      "epoch": 3.991111111111111,
+      "grad_norm": 0.8776792287826538,
+      "learning_rate": 7.48649234891712e-05,
+      "loss": 0.2508,
+      "step": 449
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.3274805247783661,
+      "learning_rate": 7.472960545147038e-05,
+      "loss": 0.2039,
+      "step": 450
+    },
+    {
+      "epoch": 4.0088888888888885,
+      "grad_norm": 0.29022541642189026,
+      "learning_rate": 7.459404720131718e-05,
+      "loss": 0.2154,
+      "step": 451
+    },
+    {
+      "epoch": 4.017777777777778,
+      "grad_norm": 0.27060315012931824,
+      "learning_rate": 7.445825005546448e-05,
+      "loss": 0.1818,
+      "step": 452
+    },
+    {
+      "epoch": 4.026666666666666,
+      "grad_norm": 0.37638726830482483,
+      "learning_rate": 7.432221533298569e-05,
+      "loss": 0.1706,
+      "step": 453
+    },
+    {
+      "epoch": 4.035555555555556,
+      "grad_norm": 0.506497323513031,
+      "learning_rate": 7.4185944355262e-05,
+      "loss": 0.188,
+      "step": 454
+    },
+    {
+      "epoch": 4.044444444444444,
+      "grad_norm": 0.3924923241138458,
+      "learning_rate": 7.404943844596939e-05,
+      "loss": 0.234,
+      "step": 455
+    },
+    {
+      "epoch": 4.053333333333334,
+      "grad_norm": 0.4585055112838745,
+      "learning_rate": 7.391269893106592e-05,
+      "loss": 0.1821,
+      "step": 456
+    },
+    {
+      "epoch": 4.062222222222222,
+      "grad_norm": 0.41875752806663513,
+      "learning_rate": 7.377572713877877e-05,
+      "loss": 0.1349,
+      "step": 457
+    },
+    {
+      "epoch": 4.071111111111111,
+      "grad_norm": 0.33695676922798157,
+      "learning_rate": 7.363852439959135e-05,
+      "loss": 0.1754,
+      "step": 458
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.739496648311615,
+      "learning_rate": 7.350109204623043e-05,
+      "loss": 0.1928,
+      "step": 459
+    },
+    {
+      "epoch": 4.088888888888889,
+      "grad_norm": 0.7458643317222595,
+      "learning_rate": 7.33634314136531e-05,
+      "loss": 0.2129,
+      "step": 460
+    },
+    {
+      "epoch": 4.097777777777778,
+      "grad_norm": 0.6029854416847229,
+      "learning_rate": 7.322554383903388e-05,
+      "loss": 0.1585,
+      "step": 461
+    },
+    {
+      "epoch": 4.1066666666666665,
+      "grad_norm": 0.5195834636688232,
+      "learning_rate": 7.308743066175172e-05,
+      "loss": 0.1809,
+      "step": 462
+    },
+    {
+      "epoch": 4.115555555555556,
+      "grad_norm": 0.5776547193527222,
+      "learning_rate": 7.294909322337689e-05,
+      "loss": 0.1918,
+      "step": 463
+    },
+    {
+      "epoch": 4.124444444444444,
+      "grad_norm": 0.394884318113327,
+      "learning_rate": 7.281053286765815e-05,
+      "loss": 0.2107,
+      "step": 464
+    },
+    {
+      "epoch": 4.133333333333334,
+      "grad_norm": 0.46265098452568054,
+      "learning_rate": 7.267175094050952e-05,
+      "loss": 0.2289,
+      "step": 465
+    },
+    {
+      "epoch": 4.142222222222222,
+      "grad_norm": 0.7043814063072205,
+      "learning_rate": 7.253274878999727e-05,
+      "loss": 0.185,
+      "step": 466
+    },
+    {
+      "epoch": 4.151111111111111,
+      "grad_norm": 0.6173180341720581,
+      "learning_rate": 7.239352776632681e-05,
+      "loss": 0.1913,
+      "step": 467
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.6549043655395508,
+      "learning_rate": 7.225408922182961e-05,
+      "loss": 0.2055,
+      "step": 468
+    },
+    {
+      "epoch": 4.168888888888889,
+      "grad_norm": 0.488744854927063,
+      "learning_rate": 7.211443451095007e-05,
+      "loss": 0.1934,
+      "step": 469
+    },
+    {
+      "epoch": 4.177777777777778,
+      "grad_norm": 0.8572131395339966,
+      "learning_rate": 7.197456499023225e-05,
+      "loss": 0.166,
+      "step": 470
+    },
+    {
+      "epoch": 4.1866666666666665,
+      "grad_norm": 0.4277536869049072,
+      "learning_rate": 7.183448201830685e-05,
+      "loss": 0.1799,
+      "step": 471
+    },
+    {
+      "epoch": 4.195555555555556,
+      "grad_norm": 0.37053030729293823,
+      "learning_rate": 7.169418695587791e-05,
+      "loss": 0.1943,
+      "step": 472
+    },
+    {
+      "epoch": 4.204444444444444,
+      "grad_norm": 0.5450714826583862,
+      "learning_rate": 7.155368116570962e-05,
+      "loss": 0.1985,
+      "step": 473
+    },
+    {
+      "epoch": 4.213333333333333,
+      "grad_norm": 0.5906587243080139,
+      "learning_rate": 7.141296601261314e-05,
+      "loss": 0.2086,
+      "step": 474
+    },
+    {
+      "epoch": 4.222222222222222,
+      "grad_norm": 0.601839542388916,
+      "learning_rate": 7.127204286343321e-05,
+      "loss": 0.1699,
+      "step": 475
+    },
+    {
+      "epoch": 4.231111111111111,
+      "grad_norm": 0.4745982587337494,
+      "learning_rate": 7.113091308703498e-05,
+      "loss": 0.1757,
+      "step": 476
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.5195015072822571,
+      "learning_rate": 7.098957805429072e-05,
+      "loss": 0.196,
+      "step": 477
+    },
+    {
+      "epoch": 4.248888888888889,
+      "grad_norm": 0.34185004234313965,
+      "learning_rate": 7.084803913806641e-05,
+      "loss": 0.1876,
+      "step": 478
+    },
+    {
+      "epoch": 4.257777777777778,
+      "grad_norm": 0.8930533528327942,
+      "learning_rate": 7.070629771320852e-05,
+      "loss": 0.1915,
+      "step": 479
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 0.5491169095039368,
+      "learning_rate": 7.056435515653059e-05,
+      "loss": 0.221,
+      "step": 480
+    },
+    {
+      "epoch": 4.275555555555556,
+      "grad_norm": 0.36910513043403625,
+      "learning_rate": 7.042221284679982e-05,
+      "loss": 0.183,
+      "step": 481
+    },
+    {
+      "epoch": 4.2844444444444445,
+      "grad_norm": 0.961035430431366,
+      "learning_rate": 7.027987216472377e-05,
+      "loss": 0.1886,
+      "step": 482
+    },
+    {
+      "epoch": 4.293333333333333,
+      "grad_norm": 0.7092379927635193,
+      "learning_rate": 7.013733449293687e-05,
+      "loss": 0.1477,
+      "step": 483
+    },
+    {
+      "epoch": 4.302222222222222,
+      "grad_norm": 0.45441004633903503,
+      "learning_rate": 6.999460121598704e-05,
+      "loss": 0.1962,
+      "step": 484
+    },
+    {
+      "epoch": 4.311111111111111,
+      "grad_norm": 0.43963754177093506,
+      "learning_rate": 6.985167372032225e-05,
+      "loss": 0.1689,
+      "step": 485
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.1201061010360718,
+      "learning_rate": 6.970855339427698e-05,
+      "loss": 0.176,
+      "step": 486
+    },
+    {
+      "epoch": 4.328888888888889,
+      "grad_norm": 0.7198124527931213,
+      "learning_rate": 6.956524162805875e-05,
+      "loss": 0.2368,
+      "step": 487
+    },
+    {
+      "epoch": 4.337777777777778,
+      "grad_norm": 0.7982850670814514,
+      "learning_rate": 6.942173981373474e-05,
+      "loss": 0.1913,
+      "step": 488
+    },
+    {
+      "epoch": 4.346666666666667,
+      "grad_norm": 0.6484464406967163,
+      "learning_rate": 6.92780493452181e-05,
+      "loss": 0.175,
+      "step": 489
+    },
+    {
+      "epoch": 4.355555555555555,
+      "grad_norm": 0.594021737575531,
+      "learning_rate": 6.91341716182545e-05,
+      "loss": 0.1777,
+      "step": 490
+    },
+    {
+      "epoch": 4.364444444444445,
+      "grad_norm": 0.6100602149963379,
+      "learning_rate": 6.899010803040857e-05,
+      "loss": 0.1492,
+      "step": 491
+    },
+    {
+      "epoch": 4.373333333333333,
+      "grad_norm": 0.5219585299491882,
+      "learning_rate": 6.884585998105026e-05,
+      "loss": 0.2231,
+      "step": 492
+    },
+    {
+      "epoch": 4.3822222222222225,
+      "grad_norm": 0.790718138217926,
+      "learning_rate": 6.870142887134141e-05,
+      "loss": 0.1844,
+      "step": 493
+    },
+    {
+      "epoch": 4.391111111111111,
+      "grad_norm": 0.44537198543548584,
+      "learning_rate": 6.855681610422189e-05,
+      "loss": 0.1866,
+      "step": 494
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.3384419083595276,
+      "learning_rate": 6.841202308439623e-05,
+      "loss": 0.1788,
+      "step": 495
+    },
+    {
+      "epoch": 4.408888888888889,
+      "grad_norm": 0.4067346453666687,
+      "learning_rate": 6.826705121831976e-05,
+      "loss": 0.2291,
+      "step": 496
+    },
+    {
+      "epoch": 4.417777777777777,
+      "grad_norm": 0.9824737310409546,
+      "learning_rate": 6.812190191418508e-05,
+      "loss": 0.1777,
+      "step": 497
+    },
+    {
+      "epoch": 4.426666666666667,
+      "grad_norm": 0.963668704032898,
+      "learning_rate": 6.797657658190839e-05,
+      "loss": 0.1765,
+      "step": 498
+    },
+    {
+      "epoch": 4.435555555555555,
+      "grad_norm": 0.3285558223724365,
+      "learning_rate": 6.783107663311565e-05,
+      "loss": 0.1971,
+      "step": 499
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 0.38868242502212524,
+      "learning_rate": 6.768540348112907e-05,
+      "loss": 0.2064,
+      "step": 500
+    },
+    {
+      "epoch": 4.444444444444445,
+      "eval_loss": 0.3316148519515991,
+      "eval_runtime": 44.4048,
+      "eval_samples_per_second": 2.252,
+      "eval_steps_per_second": 2.252,
+      "step": 500
+    },
+    {
+      "epoch": 4.453333333333333,
+      "grad_norm": 0.8600643873214722,
+      "learning_rate": 6.753955854095323e-05,
+      "loss": 0.1599,
+      "step": 501
+    },
+    {
+      "epoch": 4.4622222222222225,
+      "grad_norm": 0.6059608459472656,
+      "learning_rate": 6.739354322926136e-05,
+      "loss": 0.1915,
+      "step": 502
+    },
+    {
+      "epoch": 4.471111111111111,
+      "grad_norm": 0.8490309715270996,
+      "learning_rate": 6.724735896438167e-05,
+      "loss": 0.2769,
+      "step": 503
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.3413046896457672,
+      "learning_rate": 6.710100716628344e-05,
+      "loss": 0.1927,
+      "step": 504
+    },
+    {
+      "epoch": 4.488888888888889,
+      "grad_norm": 0.5042988657951355,
+      "learning_rate": 6.695448925656333e-05,
+      "loss": 0.146,
+      "step": 505
+    },
+    {
+      "epoch": 4.497777777777777,
+      "grad_norm": 0.46430540084838867,
+      "learning_rate": 6.680780665843155e-05,
+      "loss": 0.1989,
+      "step": 506
+    },
+    {
+      "epoch": 4.506666666666667,
+      "grad_norm": 0.45499563217163086,
+      "learning_rate": 6.666096079669797e-05,
+      "loss": 0.2019,
+      "step": 507
+    },
+    {
+      "epoch": 4.515555555555555,
+      "grad_norm": 0.33445480465888977,
+      "learning_rate": 6.651395309775837e-05,
+      "loss": 0.2001,
+      "step": 508
+    },
+    {
+      "epoch": 4.524444444444445,
+      "grad_norm": 0.6296830773353577,
+      "learning_rate": 6.636678498958052e-05,
+      "loss": 0.1525,
+      "step": 509
+    },
+    {
+      "epoch": 4.533333333333333,
+      "grad_norm": 0.49194592237472534,
+      "learning_rate": 6.621945790169036e-05,
+      "loss": 0.1748,
+      "step": 510
+    },
+    {
+      "epoch": 4.542222222222223,
+      "grad_norm": 0.3655872344970703,
+      "learning_rate": 6.607197326515808e-05,
+      "loss": 0.1935,
+      "step": 511
+    },
+    {
+      "epoch": 4.551111111111111,
+      "grad_norm": 0.6454001069068909,
+      "learning_rate": 6.592433251258423e-05,
+      "loss": 0.1625,
+      "step": 512
+    },
+    {
+      "epoch": 4.5600000000000005,
+      "grad_norm": 0.5047568082809448,
+      "learning_rate": 6.577653707808577e-05,
+      "loss": 0.1964,
+      "step": 513
+    },
+    {
+      "epoch": 4.568888888888889,
+      "grad_norm": 0.8591883778572083,
+      "learning_rate": 6.562858839728223e-05,
+      "loss": 0.2091,
+      "step": 514
+    },
+    {
+      "epoch": 4.5777777777777775,
+      "grad_norm": 0.3702365756034851,
+      "learning_rate": 6.548048790728165e-05,
+      "loss": 0.2029,
+      "step": 515
+    },
+    {
+      "epoch": 4.586666666666667,
+      "grad_norm": 0.4821636378765106,
+      "learning_rate": 6.533223704666672e-05,
+      "loss": 0.1799,
+      "step": 516
+    },
+    {
+      "epoch": 4.595555555555555,
+      "grad_norm": 0.4567244350910187,
+      "learning_rate": 6.518383725548074e-05,
+      "loss": 0.1867,
+      "step": 517
+    },
+    {
+      "epoch": 4.604444444444445,
+      "grad_norm": 0.613053560256958,
+      "learning_rate": 6.503528997521366e-05,
+      "loss": 0.1594,
+      "step": 518
+    },
+    {
+      "epoch": 4.613333333333333,
+      "grad_norm": 0.34238895773887634,
+      "learning_rate": 6.488659664878808e-05,
+      "loss": 0.1931,
+      "step": 519
+    },
+    {
+      "epoch": 4.622222222222222,
+      "grad_norm": 0.3325413465499878,
+      "learning_rate": 6.473775872054521e-05,
+      "loss": 0.1665,
+      "step": 520
+    },
+    {
+      "epoch": 4.631111111111111,
+      "grad_norm": 0.518052875995636,
+      "learning_rate": 6.458877763623089e-05,
+      "loss": 0.1491,
+      "step": 521
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.398858904838562,
+      "learning_rate": 6.44396548429815e-05,
+      "loss": 0.1651,
+      "step": 522
+    },
+    {
+      "epoch": 4.648888888888889,
+      "grad_norm": 0.44876357913017273,
+      "learning_rate": 6.42903917893099e-05,
+      "loss": 0.1933,
+      "step": 523
+    },
+    {
+      "epoch": 4.657777777777778,
+      "grad_norm": 0.58448725938797,
+      "learning_rate": 6.414098992509138e-05,
+      "loss": 0.1924,
+      "step": 524
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 0.5056392550468445,
+      "learning_rate": 6.399145070154961e-05,
+      "loss": 0.1667,
+      "step": 525
+    },
+    {
+      "epoch": 4.6755555555555555,
+      "grad_norm": 0.6365034580230713,
+      "learning_rate": 6.384177557124247e-05,
+      "loss": 0.182,
+      "step": 526
+    },
+    {
+      "epoch": 4.684444444444445,
+      "grad_norm": 0.3645959496498108,
+      "learning_rate": 6.369196598804801e-05,
+      "loss": 0.2032,
+      "step": 527
+    },
+    {
+      "epoch": 4.693333333333333,
+      "grad_norm": 0.42011165618896484,
+      "learning_rate": 6.354202340715026e-05,
+      "loss": 0.2085,
+      "step": 528
+    },
+    {
+      "epoch": 4.702222222222222,
+      "grad_norm": 0.4557260274887085,
+      "learning_rate": 6.339194928502517e-05,
+      "loss": 0.2283,
+      "step": 529
+    },
+    {
+      "epoch": 4.711111111111111,
+      "grad_norm": 0.41336071491241455,
+      "learning_rate": 6.324174507942637e-05,
+      "loss": 0.1633,
+      "step": 530
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.3430007994174957,
+      "learning_rate": 6.309141224937111e-05,
+      "loss": 0.179,
+      "step": 531
+    },
+    {
+      "epoch": 4.728888888888889,
+      "grad_norm": 0.5025080442428589,
+      "learning_rate": 6.294095225512603e-05,
+      "loss": 0.1826,
+      "step": 532
+    },
+    {
+      "epoch": 4.737777777777778,
+      "grad_norm": 0.6088576316833496,
+      "learning_rate": 6.279036655819302e-05,
+      "loss": 0.2157,
+      "step": 533
+    },
+    {
+      "epoch": 4.746666666666667,
+      "grad_norm": 0.4910157322883606,
+      "learning_rate": 6.263965662129487e-05,
+      "loss": 0.1971,
+      "step": 534
+    },
+    {
+      "epoch": 4.7555555555555555,
+      "grad_norm": 0.3384090065956116,
+      "learning_rate": 6.248882390836135e-05,
+      "loss": 0.1443,
+      "step": 535
+    },
+    {
+      "epoch": 4.764444444444445,
+      "grad_norm": 0.40888798236846924,
+      "learning_rate": 6.233786988451468e-05,
+      "loss": 0.134,
+      "step": 536
+    },
+    {
+      "epoch": 4.773333333333333,
+      "grad_norm": 0.411403626203537,
+      "learning_rate": 6.218679601605554e-05,
+      "loss": 0.2239,
+      "step": 537
+    },
+    {
+      "epoch": 4.782222222222222,
+      "grad_norm": 0.5049718618392944,
+      "learning_rate": 6.203560377044866e-05,
+      "loss": 0.1874,
+      "step": 538
+    },
+    {
+      "epoch": 4.791111111111111,
+      "grad_norm": 0.35621222853660583,
+      "learning_rate": 6.188429461630866e-05,
+      "loss": 0.1915,
+      "step": 539
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.7713626027107239,
+      "learning_rate": 6.173287002338577e-05,
+      "loss": 0.2097,
+      "step": 540
+    },
+    {
+      "epoch": 4.808888888888889,
+      "grad_norm": 1.0501298904418945,
+      "learning_rate": 6.158133146255153e-05,
+      "loss": 0.2026,
+      "step": 541
+    },
+    {
+      "epoch": 4.817777777777778,
+      "grad_norm": 0.7429870367050171,
+      "learning_rate": 6.142968040578449e-05,
+      "loss": 0.2041,
+      "step": 542
+    },
+    {
+      "epoch": 4.826666666666666,
+      "grad_norm": 0.4614238739013672,
+      "learning_rate": 6.127791832615598e-05,
+      "loss": 0.1685,
+      "step": 543
+    },
+    {
+      "epoch": 4.835555555555556,
+      "grad_norm": 0.63481205701828,
+      "learning_rate": 6.112604669781572e-05,
+      "loss": 0.1836,
+      "step": 544
+    },
+    {
+      "epoch": 4.844444444444444,
+      "grad_norm": 0.5263459086418152,
+      "learning_rate": 6.097406699597759e-05,
+      "loss": 0.162,
+      "step": 545
+    },
+    {
+      "epoch": 4.8533333333333335,
+      "grad_norm": 0.4739071726799011,
+      "learning_rate": 6.0821980696905146e-05,
+      "loss": 0.2085,
+      "step": 546
+    },
+    {
+      "epoch": 4.862222222222222,
+      "grad_norm": 0.4817952811717987,
+      "learning_rate": 6.0669789277897507e-05,
+      "loss": 0.1907,
+      "step": 547
+    },
+    {
+      "epoch": 4.871111111111111,
+      "grad_norm": 0.540302574634552,
+      "learning_rate": 6.0517494217274794e-05,
+      "loss": 0.1743,
+      "step": 548
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.48975828289985657,
+      "learning_rate": 6.036509699436389e-05,
+      "loss": 0.1769,
+      "step": 549
+    },
+    {
+      "epoch": 4.888888888888889,
+      "grad_norm": 0.4697459042072296,
+      "learning_rate": 6.021259908948402e-05,
+      "loss": 0.1811,
+      "step": 550
+    },
+    {
+      "epoch": 4.897777777777778,
+      "grad_norm": 0.8186047077178955,
+      "learning_rate": 6.0060001983932425e-05,
+      "loss": 0.225,
+      "step": 551
+    },
+    {
+      "epoch": 4.906666666666666,
+      "grad_norm": 0.3741462528705597,
+      "learning_rate": 5.9907307159969884e-05,
+      "loss": 0.1601,
+      "step": 552
+    },
+    {
+      "epoch": 4.915555555555556,
+      "grad_norm": 0.46077340841293335,
+      "learning_rate": 5.9754516100806423e-05,
+      "loss": 0.1887,
+      "step": 553
+    },
+    {
+      "epoch": 4.924444444444444,
+      "grad_norm": 0.5414716601371765,
+      "learning_rate": 5.960163029058682e-05,
+      "loss": 0.1665,
+      "step": 554
+    },
+    {
+      "epoch": 4.933333333333334,
+      "grad_norm": 0.5361833572387695,
+      "learning_rate": 5.944865121437622e-05,
+      "loss": 0.1925,
+      "step": 555
+    },
+    {
+      "epoch": 4.942222222222222,
+      "grad_norm": 0.35004904866218567,
+      "learning_rate": 5.9295580358145744e-05,
+      "loss": 0.1933,
+      "step": 556
+    },
+    {
+      "epoch": 4.9511111111111115,
+      "grad_norm": 1.0410363674163818,
+      "learning_rate": 5.914241920875798e-05,
+      "loss": 0.2299,
+      "step": 557
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.5200708508491516,
+      "learning_rate": 5.898916925395264e-05,
+      "loss": 0.1673,
+      "step": 558
+    },
+    {
+      "epoch": 4.968888888888889,
+      "grad_norm": 0.534602165222168,
+      "learning_rate": 5.8835831982332015e-05,
+      "loss": 0.1815,
+      "step": 559
+    },
+    {
+      "epoch": 4.977777777777778,
+      "grad_norm": 0.4768090546131134,
+      "learning_rate": 5.868240888334653e-05,
+      "loss": 0.2138,
+      "step": 560
+    },
+    {
+      "epoch": 4.986666666666666,
+      "grad_norm": 0.38369208574295044,
+      "learning_rate": 5.852890144728034e-05,
+      "loss": 0.1509,
+      "step": 561
+    },
+    {
+      "epoch": 4.995555555555556,
+      "grad_norm": 0.35721951723098755,
+      "learning_rate": 5.837531116523682e-05,
+      "loss": 0.1485,
+      "step": 562
+    },
+    {
+      "epoch": 5.004444444444444,
+      "grad_norm": 0.44299712777137756,
+      "learning_rate": 5.822163952912404e-05,
+      "loss": 0.1504,
+      "step": 563
+    },
+    {
+      "epoch": 5.013333333333334,
+      "grad_norm": 0.47305387258529663,
+      "learning_rate": 5.806788803164034e-05,
+      "loss": 0.1497,
+      "step": 564
+    },
+    {
+      "epoch": 5.022222222222222,
+      "grad_norm": 0.5253143906593323,
+      "learning_rate": 5.791405816625975e-05,
+      "loss": 0.1726,
+      "step": 565
+    },
+    {
+      "epoch": 5.0311111111111115,
+      "grad_norm": 0.5787915587425232,
+      "learning_rate": 5.7760151427217576e-05,
+      "loss": 0.1603,
+      "step": 566
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.7631460428237915,
+      "learning_rate": 5.7606169309495836e-05,
+      "loss": 0.2153,
+      "step": 567
+    },
+    {
+      "epoch": 5.0488888888888885,
+      "grad_norm": 0.472397118806839,
+      "learning_rate": 5.745211330880872e-05,
+      "loss": 0.1887,
+      "step": 568
+    },
+    {
+      "epoch": 5.057777777777778,
+      "grad_norm": 0.4452255368232727,
+      "learning_rate": 5.729798492158811e-05,
+      "loss": 0.2073,
+      "step": 569
+    },
+    {
+      "epoch": 5.066666666666666,
+      "grad_norm": 0.4919035732746124,
+      "learning_rate": 5.714378564496901e-05,
+      "loss": 0.1449,
+      "step": 570
+    },
+    {
+      "epoch": 5.075555555555556,
+      "grad_norm": 0.6700975894927979,
+      "learning_rate": 5.698951697677498e-05,
+      "loss": 0.1024,
+      "step": 571
+    },
+    {
+      "epoch": 5.084444444444444,
+      "grad_norm": 0.6321210265159607,
+      "learning_rate": 5.683518041550368e-05,
+      "loss": 0.1584,
+      "step": 572
+    },
+    {
+      "epoch": 5.093333333333334,
+      "grad_norm": 0.7015148997306824,
+      "learning_rate": 5.668077746031219e-05,
+      "loss": 0.1387,
+      "step": 573
+    },
+    {
+      "epoch": 5.102222222222222,
+      "grad_norm": 0.6382136344909668,
+      "learning_rate": 5.6526309611002594e-05,
+      "loss": 0.1706,
+      "step": 574
+    },
+    {
+      "epoch": 5.111111111111111,
+      "grad_norm": 0.45012837648391724,
+      "learning_rate": 5.637177836800722e-05,
+      "loss": 0.155,
+      "step": 575
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.47716137766838074,
+      "learning_rate": 5.621718523237427e-05,
+      "loss": 0.1303,
+      "step": 576
+    },
+    {
+      "epoch": 5.128888888888889,
+      "grad_norm": 0.7658069729804993,
+      "learning_rate": 5.6062531705753075e-05,
+      "loss": 0.1448,
+      "step": 577
+    },
+    {
+      "epoch": 5.137777777777778,
+      "grad_norm": 0.43680229783058167,
+      "learning_rate": 5.590781929037965e-05,
+      "loss": 0.1443,
+      "step": 578
+    },
+    {
+      "epoch": 5.1466666666666665,
+      "grad_norm": 0.834297239780426,
+      "learning_rate": 5.575304948906194e-05,
+      "loss": 0.0925,
+      "step": 579
+    },
+    {
+      "epoch": 5.155555555555556,
+      "grad_norm": 0.405964195728302,
+      "learning_rate": 5.559822380516539e-05,
+      "loss": 0.1427,
+      "step": 580
+    },
+    {
+      "epoch": 5.164444444444444,
+      "grad_norm": 0.49563100934028625,
+      "learning_rate": 5.544334374259823e-05,
+      "loss": 0.1492,
+      "step": 581
+    },
+    {
+      "epoch": 5.173333333333334,
+      "grad_norm": 0.5021004676818848,
+      "learning_rate": 5.5288410805796895e-05,
+      "loss": 0.1556,
+      "step": 582
+    },
+    {
+      "epoch": 5.182222222222222,
+      "grad_norm": 0.5862144827842712,
+      "learning_rate": 5.5133426499711425e-05,
+      "loss": 0.142,
+      "step": 583
+    },
+    {
+      "epoch": 5.191111111111111,
+      "grad_norm": 0.43614262342453003,
+      "learning_rate": 5.497839232979084e-05,
+      "loss": 0.1626,
+      "step": 584
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.5935210585594177,
+      "learning_rate": 5.4823309801968516e-05,
+      "loss": 0.1267,
+      "step": 585
+    },
+    {
+      "epoch": 5.208888888888889,
+      "grad_norm": 0.4941434860229492,
+      "learning_rate": 5.466818042264753e-05,
+      "loss": 0.1144,
+      "step": 586
+    },
+    {
+      "epoch": 5.217777777777778,
+      "grad_norm": 0.6197287440299988,
+      "learning_rate": 5.451300569868612e-05,
+      "loss": 0.1654,
+      "step": 587
+    },
+    {
+      "epoch": 5.226666666666667,
+      "grad_norm": 0.38996899127960205,
+      "learning_rate": 5.435778713738292e-05,
+      "loss": 0.1925,
+      "step": 588
+    },
+    {
+      "epoch": 5.235555555555556,
+      "grad_norm": 1.012840986251831,
+      "learning_rate": 5.420252624646238e-05,
+      "loss": 0.1575,
+      "step": 589
+    },
+    {
+      "epoch": 5.2444444444444445,
+      "grad_norm": 0.33670493960380554,
+      "learning_rate": 5.404722453406017e-05,
+      "loss": 0.1635,
+      "step": 590
+    },
+    {
+      "epoch": 5.253333333333333,
+      "grad_norm": 0.3880348205566406,
+      "learning_rate": 5.3891883508708444e-05,
+      "loss": 0.1301,
+      "step": 591
+    },
+    {
+      "epoch": 5.262222222222222,
+      "grad_norm": 0.8689696192741394,
+      "learning_rate": 5.373650467932122e-05,
+      "loss": 0.1621,
+      "step": 592
+    },
+    {
+      "epoch": 5.271111111111111,
+      "grad_norm": 0.4196907877922058,
+      "learning_rate": 5.3581089555179754e-05,
+      "loss": 0.1501,
+      "step": 593
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.9811832904815674,
+      "learning_rate": 5.3425639645917834e-05,
+      "loss": 0.1124,
+      "step": 594
+    },
+    {
+      "epoch": 5.288888888888889,
+      "grad_norm": 0.7768471240997314,
+      "learning_rate": 5.327015646150716e-05,
+      "loss": 0.1264,
+      "step": 595
+    },
+    {
+      "epoch": 5.297777777777778,
+      "grad_norm": 0.5367645025253296,
+      "learning_rate": 5.311464151224261e-05,
+      "loss": 0.1095,
+      "step": 596
+    },
+    {
+      "epoch": 5.306666666666667,
+      "grad_norm": 0.5028023719787598,
+      "learning_rate": 5.295909630872764e-05,
+      "loss": 0.1205,
+      "step": 597
+    },
+    {
+      "epoch": 5.315555555555555,
+      "grad_norm": 0.39689093828201294,
+      "learning_rate": 5.2803522361859594e-05,
+      "loss": 0.1511,
+      "step": 598
+    },
+    {
+      "epoch": 5.3244444444444445,
+      "grad_norm": 0.7014448642730713,
+      "learning_rate": 5.264792118281498e-05,
+      "loss": 0.1379,
+      "step": 599
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 0.7733895778656006,
+      "learning_rate": 5.249229428303486e-05,
+      "loss": 0.1538,
+      "step": 600
+    },
+    {
+      "epoch": 5.342222222222222,
+      "grad_norm": 0.46061837673187256,
+      "learning_rate": 5.233664317421012e-05,
+      "loss": 0.188,
+      "step": 601
+    },
+    {
+      "epoch": 5.351111111111111,
+      "grad_norm": 0.45391765236854553,
+      "learning_rate": 5.218096936826681e-05,
+      "loss": 0.2143,
+      "step": 602
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.4227662682533264,
+      "learning_rate": 5.202527437735143e-05,
+      "loss": 0.1974,
+      "step": 603
+    },
+    {
+      "epoch": 5.368888888888889,
+      "grad_norm": 0.5953728556632996,
+      "learning_rate": 5.18695597138163e-05,
+      "loss": 0.131,
+      "step": 604
+    },
+    {
+      "epoch": 5.377777777777778,
+      "grad_norm": 0.7897362112998962,
+      "learning_rate": 5.1713826890204764e-05,
+      "loss": 0.151,
+      "step": 605
+    },
+    {
+      "epoch": 5.386666666666667,
+      "grad_norm": 0.611005425453186,
+      "learning_rate": 5.155807741923666e-05,
+      "loss": 0.1771,
+      "step": 606
+    },
+    {
+      "epoch": 5.395555555555555,
+      "grad_norm": 0.746476411819458,
+      "learning_rate": 5.140231281379345e-05,
+      "loss": 0.1676,
+      "step": 607
+    },
+    {
+      "epoch": 5.404444444444445,
+      "grad_norm": 0.4701192378997803,
+      "learning_rate": 5.124653458690365e-05,
+      "loss": 0.1661,
+      "step": 608
+    },
+    {
+      "epoch": 5.413333333333333,
+      "grad_norm": 0.6008636355400085,
+      "learning_rate": 5.1090744251728064e-05,
+      "loss": 0.1462,
+      "step": 609
+    },
+    {
+      "epoch": 5.4222222222222225,
+      "grad_norm": 0.5137926340103149,
+      "learning_rate": 5.0934943321545115e-05,
+      "loss": 0.1682,
+      "step": 610
+    },
+    {
+      "epoch": 5.431111111111111,
+      "grad_norm": 0.42210161685943604,
+      "learning_rate": 5.077913330973616e-05,
+      "loss": 0.1599,
+      "step": 611
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.805438220500946,
+      "learning_rate": 5.062331572977076e-05,
+      "loss": 0.1416,
+      "step": 612
+    },
+    {
+      "epoch": 5.448888888888889,
+      "grad_norm": 0.696622371673584,
+      "learning_rate": 5.046749209519197e-05,
+      "loss": 0.1785,
+      "step": 613
+    },
+    {
+      "epoch": 5.457777777777777,
+      "grad_norm": 0.4268178939819336,
+      "learning_rate": 5.031166391960168e-05,
+      "loss": 0.183,
+      "step": 614
+    },
+    {
+      "epoch": 5.466666666666667,
+      "grad_norm": 0.8230905532836914,
+      "learning_rate": 5.015583271664589e-05,
+      "loss": 0.1309,
+      "step": 615
+    },
+    {
+      "epoch": 5.475555555555555,
+      "grad_norm": 0.4906931221485138,
+      "learning_rate": 5e-05,
+      "loss": 0.1683,
+      "step": 616
+    },
+    {
+      "epoch": 5.484444444444445,
+      "grad_norm": 0.6042024493217468,
+      "learning_rate": 4.984416728335412e-05,
+      "loss": 0.1486,
+      "step": 617
+    },
+    {
+      "epoch": 5.493333333333333,
+      "grad_norm": 0.39481180906295776,
+      "learning_rate": 4.968833608039832e-05,
+      "loss": 0.1548,
+      "step": 618
+    },
+    {
+      "epoch": 5.502222222222223,
+      "grad_norm": 0.5813857913017273,
+      "learning_rate": 4.953250790480805e-05,
+      "loss": 0.1513,
+      "step": 619
+    },
+    {
+      "epoch": 5.511111111111111,
+      "grad_norm": 0.3986724019050598,
+      "learning_rate": 4.9376684270229254e-05,
+      "loss": 0.1728,
+      "step": 620
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.5897936224937439,
+      "learning_rate": 4.922086669026385e-05,
+      "loss": 0.1607,
+      "step": 621
+    },
+    {
+      "epoch": 5.528888888888889,
+      "grad_norm": 0.4015739858150482,
+      "learning_rate": 4.9065056678454904e-05,
+      "loss": 0.1542,
+      "step": 622
+    },
+    {
+      "epoch": 5.5377777777777775,
+      "grad_norm": 0.5246159434318542,
+      "learning_rate": 4.890925574827195e-05,
+      "loss": 0.1894,
+      "step": 623
+    },
+    {
+      "epoch": 5.546666666666667,
+      "grad_norm": 0.6899661421775818,
+      "learning_rate": 4.875346541309637e-05,
+      "loss": 0.1172,
+      "step": 624
+    },
+    {
+      "epoch": 5.555555555555555,
+      "grad_norm": 0.3335443139076233,
+      "learning_rate": 4.859768718620656e-05,
+      "loss": 0.1272,
+      "step": 625
+    },
+    {
+      "epoch": 5.564444444444445,
+      "grad_norm": 0.5370382070541382,
+      "learning_rate": 4.844192258076336e-05,
+      "loss": 0.1597,
+      "step": 626
+    },
+    {
+      "epoch": 5.573333333333333,
+      "grad_norm": 0.7130039930343628,
+      "learning_rate": 4.8286173109795255e-05,
+      "loss": 0.1795,
+      "step": 627
+    },
+    {
+      "epoch": 5.582222222222223,
+      "grad_norm": 0.6044110059738159,
+      "learning_rate": 4.813044028618373e-05,
+      "loss": 0.1402,
+      "step": 628
+    },
+    {
+      "epoch": 5.591111111111111,
+      "grad_norm": 0.5665392875671387,
+      "learning_rate": 4.7974725622648595e-05,
+      "loss": 0.1339,
+      "step": 629
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.49484479427337646,
+      "learning_rate": 4.781903063173321e-05,
+      "loss": 0.2101,
+      "step": 630
+    },
+    {
+      "epoch": 5.608888888888889,
+      "grad_norm": 0.39191365242004395,
+      "learning_rate": 4.76633568257899e-05,
+      "loss": 0.1823,
+      "step": 631
+    },
+    {
+      "epoch": 5.6177777777777775,
+      "grad_norm": 0.5061699151992798,
+      "learning_rate": 4.750770571696514e-05,
+      "loss": 0.127,
+      "step": 632
+    },
+    {
+      "epoch": 5.626666666666667,
+      "grad_norm": 0.5720147490501404,
+      "learning_rate": 4.735207881718502e-05,
+      "loss": 0.1226,
+      "step": 633
+    },
+    {
+      "epoch": 5.635555555555555,
+      "grad_norm": 0.5010076761245728,
+      "learning_rate": 4.7196477638140404e-05,
+      "loss": 0.1503,
+      "step": 634
+    },
+    {
+      "epoch": 5.644444444444445,
+      "grad_norm": 0.6409832239151001,
+      "learning_rate": 4.704090369127236e-05,
+      "loss": 0.1718,
+      "step": 635
+    },
+    {
+      "epoch": 5.653333333333333,
+      "grad_norm": 0.5019122958183289,
+      "learning_rate": 4.68853584877574e-05,
+      "loss": 0.1898,
+      "step": 636
+    },
+    {
+      "epoch": 5.662222222222223,
+      "grad_norm": 0.4418942630290985,
+      "learning_rate": 4.6729843538492847e-05,
+      "loss": 0.1961,
+      "step": 637
+    },
+    {
+      "epoch": 5.671111111111111,
+      "grad_norm": 0.4697255790233612,
+      "learning_rate": 4.657436035408217e-05,
+      "loss": 0.1161,
+      "step": 638
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.6954605579376221,
+      "learning_rate": 4.6418910444820244e-05,
+      "loss": 0.1298,
+      "step": 639
+    },
+    {
+      "epoch": 5.688888888888889,
+      "grad_norm": 0.7173737287521362,
+      "learning_rate": 4.626349532067879e-05,
+      "loss": 0.1488,
+      "step": 640
+    },
+    {
+      "epoch": 5.697777777777778,
+      "grad_norm": 0.4419032633304596,
+      "learning_rate": 4.6108116491291575e-05,
+      "loss": 0.1844,
+      "step": 641
+    },
+    {
+      "epoch": 5.706666666666667,
+      "grad_norm": 0.4070740044116974,
+      "learning_rate": 4.595277546593984e-05,
+      "loss": 0.1796,
+      "step": 642
+    },
+    {
+      "epoch": 5.7155555555555555,
+      "grad_norm": 0.40745702385902405,
+      "learning_rate": 4.579747375353763e-05,
+      "loss": 0.1655,
+      "step": 643
+    },
+    {
+      "epoch": 5.724444444444444,
+      "grad_norm": 0.7508057951927185,
+      "learning_rate": 4.564221286261709e-05,
+      "loss": 0.1376,
+      "step": 644
+    },
+    {
+      "epoch": 5.733333333333333,
+      "grad_norm": 0.7357051968574524,
+      "learning_rate": 4.548699430131388e-05,
+      "loss": 0.1365,
+      "step": 645
+    },
+    {
+      "epoch": 5.742222222222222,
+      "grad_norm": 1.2087818384170532,
+      "learning_rate": 4.5331819577352474e-05,
+      "loss": 0.1381,
+      "step": 646
+    },
+    {
+      "epoch": 5.751111111111111,
+      "grad_norm": 0.690025269985199,
+      "learning_rate": 4.51766901980315e-05,
+      "loss": 0.1401,
+      "step": 647
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.41388073563575745,
+      "learning_rate": 4.502160767020918e-05,
+      "loss": 0.1285,
+      "step": 648
+    },
+    {
+      "epoch": 5.768888888888889,
+      "grad_norm": 0.5141646862030029,
+      "learning_rate": 4.486657350028859e-05,
+      "loss": 0.1308,
+      "step": 649
+    },
+    {
+      "epoch": 5.777777777777778,
+      "grad_norm": 1.4173223972320557,
+      "learning_rate": 4.471158919420312e-05,
+      "loss": 0.1514,
+      "step": 650
+    },
+    {
+      "epoch": 5.786666666666667,
+      "grad_norm": 0.5400106906890869,
+      "learning_rate": 4.4556656257401786e-05,
+      "loss": 0.1632,
+      "step": 651
+    },
+    {
+      "epoch": 5.795555555555556,
+      "grad_norm": 0.8285782337188721,
+      "learning_rate": 4.4401776194834613e-05,
+      "loss": 0.1529,
+      "step": 652
+    },
+    {
+      "epoch": 5.804444444444444,
+      "grad_norm": 0.49138960242271423,
+      "learning_rate": 4.424695051093807e-05,
+      "loss": 0.1557,
+      "step": 653
+    },
+    {
+      "epoch": 5.8133333333333335,
+      "grad_norm": 0.4443909525871277,
+      "learning_rate": 4.409218070962036e-05,
+      "loss": 0.1424,
+      "step": 654
+    },
+    {
+      "epoch": 5.822222222222222,
+      "grad_norm": 0.5284453630447388,
+      "learning_rate": 4.393746829424693e-05,
+      "loss": 0.1889,
+      "step": 655
+    },
+    {
+      "epoch": 5.831111111111111,
+      "grad_norm": 0.5384455919265747,
+      "learning_rate": 4.378281476762576e-05,
+      "loss": 0.1727,
+      "step": 656
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.6882079839706421,
+      "learning_rate": 4.362822163199279e-05,
+      "loss": 0.1511,
+      "step": 657
+    },
+    {
+      "epoch": 5.848888888888889,
+      "grad_norm": 0.4634445309638977,
+      "learning_rate": 4.347369038899744e-05,
+      "loss": 0.1261,
+      "step": 658
+    },
+    {
+      "epoch": 5.857777777777778,
+      "grad_norm": 0.5075255036354065,
+      "learning_rate": 4.331922253968781e-05,
+      "loss": 0.1919,
+      "step": 659
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 1.4190889596939087,
+      "learning_rate": 4.316481958449634e-05,
+      "loss": 0.2282,
+      "step": 660
+    },
+    {
+      "epoch": 5.875555555555556,
+      "grad_norm": 0.6407334804534912,
+      "learning_rate": 4.3010483023225045e-05,
+      "loss": 0.1497,
+      "step": 661
+    },
+    {
+      "epoch": 5.884444444444444,
+      "grad_norm": 0.5419467687606812,
+      "learning_rate": 4.285621435503101e-05,
+      "loss": 0.1311,
+      "step": 662
+    },
+    {
+      "epoch": 5.8933333333333335,
+      "grad_norm": 0.8229005336761475,
+      "learning_rate": 4.2702015078411906e-05,
+      "loss": 0.128,
+      "step": 663
+    },
+    {
+      "epoch": 5.902222222222222,
+      "grad_norm": 0.33451828360557556,
+      "learning_rate": 4.254788669119127e-05,
+      "loss": 0.15,
+      "step": 664
+    },
+    {
+      "epoch": 5.911111111111111,
+      "grad_norm": 0.5506479740142822,
+      "learning_rate": 4.239383069050417e-05,
+      "loss": 0.1371,
+      "step": 665
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.41086164116859436,
+      "learning_rate": 4.223984857278242e-05,
+      "loss": 0.1652,
+      "step": 666
+    },
+    {
+      "epoch": 5.928888888888888,
+      "grad_norm": 0.41804587841033936,
+      "learning_rate": 4.208594183374026e-05,
+      "loss": 0.1461,
+      "step": 667
+    },
+    {
+      "epoch": 5.937777777777778,
+      "grad_norm": 0.7241559028625488,
+      "learning_rate": 4.1932111968359664e-05,
+      "loss": 0.1429,
+      "step": 668
+    },
+    {
+      "epoch": 5.946666666666666,
+      "grad_norm": 0.7180150747299194,
+      "learning_rate": 4.177836047087595e-05,
+      "loss": 0.147,
+      "step": 669
+    },
+    {
+      "epoch": 5.955555555555556,
+      "grad_norm": 0.668580174446106,
+      "learning_rate": 4.162468883476319e-05,
+      "loss": 0.1562,
+      "step": 670
+    },
+    {
+      "epoch": 5.964444444444444,
+      "grad_norm": 0.46259158849716187,
+      "learning_rate": 4.147109855271966e-05,
+      "loss": 0.1453,
+      "step": 671
+    },
+    {
+      "epoch": 5.973333333333334,
+      "grad_norm": 0.5375617146492004,
+      "learning_rate": 4.131759111665349e-05,
+      "loss": 0.1581,
+      "step": 672
+    },
+    {
+      "epoch": 5.982222222222222,
+      "grad_norm": 0.39930492639541626,
+      "learning_rate": 4.1164168017668e-05,
+      "loss": 0.1394,
+      "step": 673
+    },
+    {
+      "epoch": 5.9911111111111115,
+      "grad_norm": 0.400290846824646,
+      "learning_rate": 4.101083074604737e-05,
+      "loss": 0.153,
+      "step": 674
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.504225492477417,
+      "learning_rate": 4.085758079124203e-05,
+      "loss": 0.1532,
+      "step": 675
+    },
+    {
+      "epoch": 6.0088888888888885,
+      "grad_norm": 0.47197291254997253,
+      "learning_rate": 4.0704419641854274e-05,
+      "loss": 0.0869,
+      "step": 676
+    },
+    {
+      "epoch": 6.017777777777778,
+      "grad_norm": 1.4398659467697144,
+      "learning_rate": 4.05513487856238e-05,
+      "loss": 0.1247,
+      "step": 677
+    },
+    {
+      "epoch": 6.026666666666666,
+      "grad_norm": 0.4351118803024292,
+      "learning_rate": 4.03983697094132e-05,
+      "loss": 0.1621,
+      "step": 678
+    },
+    {
+      "epoch": 6.035555555555556,
+      "grad_norm": 0.4250251352787018,
+      "learning_rate": 4.0245483899193595e-05,
+      "loss": 0.1137,
+      "step": 679
+    },
+    {
+      "epoch": 6.044444444444444,
+      "grad_norm": 0.5068423748016357,
+      "learning_rate": 4.0092692840030134e-05,
+      "loss": 0.1503,
+      "step": 680
+    },
+    {
+      "epoch": 6.053333333333334,
+      "grad_norm": 0.8082426190376282,
+      "learning_rate": 3.993999801606759e-05,
+      "loss": 0.1423,
+      "step": 681
+    },
+    {
+      "epoch": 6.062222222222222,
+      "grad_norm": 0.4672413170337677,
+      "learning_rate": 3.978740091051599e-05,
+      "loss": 0.12,
+      "step": 682
+    },
+    {
+      "epoch": 6.071111111111111,
+      "grad_norm": 0.8545836806297302,
+      "learning_rate": 3.963490300563613e-05,
+      "loss": 0.1151,
+      "step": 683
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 0.4437183141708374,
+      "learning_rate": 3.9482505782725224e-05,
+      "loss": 0.1586,
+      "step": 684
+    },
+    {
+      "epoch": 6.088888888888889,
+      "grad_norm": 0.6166926622390747,
+      "learning_rate": 3.933021072210251e-05,
+      "loss": 0.1711,
+      "step": 685
+    },
+    {
+      "epoch": 6.097777777777778,
+      "grad_norm": 0.8207690715789795,
+      "learning_rate": 3.917801930309486e-05,
+      "loss": 0.1137,
+      "step": 686
+    },
+    {
+      "epoch": 6.1066666666666665,
+      "grad_norm": 1.1031180620193481,
+      "learning_rate": 3.902593300402244e-05,
+      "loss": 0.1194,
+      "step": 687
+    },
+    {
+      "epoch": 6.115555555555556,
+      "grad_norm": 0.5387362837791443,
+      "learning_rate": 3.887395330218429e-05,
+      "loss": 0.1407,
+      "step": 688
+    },
+    {
+      "epoch": 6.124444444444444,
+      "grad_norm": 0.48524364829063416,
+      "learning_rate": 3.872208167384404e-05,
+      "loss": 0.0863,
+      "step": 689
+    },
+    {
+      "epoch": 6.133333333333334,
+      "grad_norm": 0.5386446714401245,
+      "learning_rate": 3.857031959421553e-05,
+      "loss": 0.1376,
+      "step": 690
+    },
+    {
+      "epoch": 6.142222222222222,
+      "grad_norm": 1.6886932849884033,
+      "learning_rate": 3.8418668537448495e-05,
+      "loss": 0.1503,
+      "step": 691
+    },
+    {
+      "epoch": 6.151111111111111,
+      "grad_norm": 0.4677202105522156,
+      "learning_rate": 3.8267129976614254e-05,
+      "loss": 0.1255,
+      "step": 692
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.4135400652885437,
+      "learning_rate": 3.8115705383691355e-05,
+      "loss": 0.1174,
+      "step": 693
+    },
+    {
+      "epoch": 6.168888888888889,
+      "grad_norm": 0.6193229556083679,
+      "learning_rate": 3.7964396229551364e-05,
+      "loss": 0.1114,
+      "step": 694
+    },
+    {
+      "epoch": 6.177777777777778,
+      "grad_norm": 0.6773757338523865,
+      "learning_rate": 3.781320398394446e-05,
+      "loss": 0.1079,
+      "step": 695
+    },
+    {
+      "epoch": 6.1866666666666665,
+      "grad_norm": 0.6738713979721069,
+      "learning_rate": 3.7662130115485314e-05,
+      "loss": 0.1335,
+      "step": 696
+    },
+    {
+      "epoch": 6.195555555555556,
+      "grad_norm": 0.7145228385925293,
+      "learning_rate": 3.7511176091638653e-05,
+      "loss": 0.1191,
+      "step": 697
+    },
+    {
+      "epoch": 6.204444444444444,
+      "grad_norm": 0.5776643753051758,
+      "learning_rate": 3.7360343378705124e-05,
+      "loss": 0.1195,
+      "step": 698
+    },
+    {
+      "epoch": 6.213333333333333,
+      "grad_norm": 0.5871068239212036,
+      "learning_rate": 3.7209633441807e-05,
+      "loss": 0.1329,
+      "step": 699
+    },
+    {
+      "epoch": 6.222222222222222,
+      "grad_norm": 0.5175588130950928,
+      "learning_rate": 3.705904774487396e-05,
+      "loss": 0.124,
+      "step": 700
+    },
+    {
+      "epoch": 6.231111111111111,
+      "grad_norm": 0.4751562774181366,
+      "learning_rate": 3.69085877506289e-05,
+      "loss": 0.1393,
+      "step": 701
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.48501744866371155,
+      "learning_rate": 3.675825492057364e-05,
+      "loss": 0.1205,
+      "step": 702
+    },
+    {
+      "epoch": 6.248888888888889,
+      "grad_norm": 0.43550392985343933,
+      "learning_rate": 3.660805071497485e-05,
+      "loss": 0.156,
+      "step": 703
+    },
+    {
+      "epoch": 6.257777777777778,
+      "grad_norm": 0.761576235294342,
+      "learning_rate": 3.6457976592849754e-05,
+      "loss": 0.1583,
+      "step": 704
+    },
+    {
+      "epoch": 6.266666666666667,
+      "grad_norm": 0.6814683079719543,
+      "learning_rate": 3.6308034011952e-05,
+      "loss": 0.1124,
+      "step": 705
+    },
+    {
+      "epoch": 6.275555555555556,
+      "grad_norm": 0.5146603584289551,
+      "learning_rate": 3.6158224428757535e-05,
+      "loss": 0.1126,
+      "step": 706
+    },
+    {
+      "epoch": 6.2844444444444445,
+      "grad_norm": 0.5021145343780518,
+      "learning_rate": 3.60085492984504e-05,
+      "loss": 0.0965,
+      "step": 707
+    },
+    {
+      "epoch": 6.293333333333333,
+      "grad_norm": 0.5636239647865295,
+      "learning_rate": 3.585901007490863e-05,
+      "loss": 0.1003,
+      "step": 708
+    },
+    {
+      "epoch": 6.302222222222222,
+      "grad_norm": 0.783494770526886,
+      "learning_rate": 3.5709608210690125e-05,
+      "loss": 0.0869,
+      "step": 709
+    },
+    {
+      "epoch": 6.311111111111111,
+      "grad_norm": 0.5414121150970459,
+      "learning_rate": 3.556034515701852e-05,
+      "loss": 0.1572,
+      "step": 710
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.7618036270141602,
+      "learning_rate": 3.541122236376911e-05,
+      "loss": 0.1396,
+      "step": 711
+    },
+    {
+      "epoch": 6.328888888888889,
+      "grad_norm": 0.48708009719848633,
+      "learning_rate": 3.5262241279454785e-05,
+      "loss": 0.1151,
+      "step": 712
+    },
+    {
+      "epoch": 6.337777777777778,
+      "grad_norm": 0.4765779376029968,
+      "learning_rate": 3.5113403351211935e-05,
+      "loss": 0.1119,
+      "step": 713
+    },
+    {
+      "epoch": 6.346666666666667,
+      "grad_norm": 0.5376085042953491,
+      "learning_rate": 3.4964710024786354e-05,
+      "loss": 0.1194,
+      "step": 714
+    },
+    {
+      "epoch": 6.355555555555555,
+      "grad_norm": 0.7036638259887695,
+      "learning_rate": 3.4816162744519263e-05,
+      "loss": 0.1572,
+      "step": 715
+    },
+    {
+      "epoch": 6.364444444444445,
+      "grad_norm": 0.6036682724952698,
+      "learning_rate": 3.4667762953333295e-05,
+      "loss": 0.107,
+      "step": 716
+    },
+    {
+      "epoch": 6.373333333333333,
+      "grad_norm": 0.843076229095459,
+      "learning_rate": 3.4519512092718354e-05,
+      "loss": 0.1489,
+      "step": 717
+    },
+    {
+      "epoch": 6.3822222222222225,
+      "grad_norm": 0.9905326962471008,
+      "learning_rate": 3.4371411602717784e-05,
+      "loss": 0.1203,
+      "step": 718
+    },
+    {
+      "epoch": 6.391111111111111,
+      "grad_norm": 0.4549235999584198,
+      "learning_rate": 3.422346292191424e-05,
+      "loss": 0.1171,
+      "step": 719
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.5710920095443726,
+      "learning_rate": 3.4075667487415785e-05,
+      "loss": 0.162,
+      "step": 720
+    },
+    {
+      "epoch": 6.408888888888889,
+      "grad_norm": 0.6092873215675354,
+      "learning_rate": 3.392802673484193e-05,
+      "loss": 0.1448,
+      "step": 721
+    },
+    {
+      "epoch": 6.417777777777777,
+      "grad_norm": 0.5983030796051025,
+      "learning_rate": 3.3780542098309654e-05,
+      "loss": 0.1172,
+      "step": 722
+    },
+    {
+      "epoch": 6.426666666666667,
+      "grad_norm": 0.6922889351844788,
+      "learning_rate": 3.36332150104195e-05,
+      "loss": 0.1098,
+      "step": 723
+    },
+    {
+      "epoch": 6.435555555555555,
+      "grad_norm": 0.6339572072029114,
+      "learning_rate": 3.3486046902241664e-05,
+      "loss": 0.1398,
+      "step": 724
+    },
+    {
+      "epoch": 6.444444444444445,
+      "grad_norm": 0.5848474502563477,
+      "learning_rate": 3.333903920330205e-05,
+      "loss": 0.1581,
+      "step": 725
+    },
+    {
+      "epoch": 6.453333333333333,
+      "grad_norm": 0.48497921228408813,
+      "learning_rate": 3.319219334156847e-05,
+      "loss": 0.1454,
+      "step": 726
+    },
+    {
+      "epoch": 6.4622222222222225,
+      "grad_norm": 0.5115544199943542,
+      "learning_rate": 3.3045510743436665e-05,
+      "loss": 0.1105,
+      "step": 727
+    },
+    {
+      "epoch": 6.471111111111111,
+      "grad_norm": 0.7270033359527588,
+      "learning_rate": 3.289899283371657e-05,
+      "loss": 0.1041,
+      "step": 728
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.6745510697364807,
+      "learning_rate": 3.275264103561834e-05,
+      "loss": 0.1349,
+      "step": 729
+    },
+    {
+      "epoch": 6.488888888888889,
+      "grad_norm": 0.4949975609779358,
+      "learning_rate": 3.2606456770738636e-05,
+      "loss": 0.1741,
+      "step": 730
+    },
+    {
+      "epoch": 6.497777777777777,
+      "grad_norm": 0.5432770252227783,
+      "learning_rate": 3.246044145904677e-05,
+      "loss": 0.1235,
+      "step": 731
+    },
+    {
+      "epoch": 6.506666666666667,
+      "grad_norm": 0.583692729473114,
+      "learning_rate": 3.231459651887093e-05,
+      "loss": 0.1151,
+      "step": 732
+    },
+    {
+      "epoch": 6.515555555555555,
+      "grad_norm": 0.4701627194881439,
+      "learning_rate": 3.216892336688435e-05,
+      "loss": 0.0989,
+      "step": 733
+    },
+    {
+      "epoch": 6.524444444444445,
+      "grad_norm": 0.6878033876419067,
+      "learning_rate": 3.2023423418091626e-05,
+      "loss": 0.1207,
+      "step": 734
+    },
+    {
+      "epoch": 6.533333333333333,
+      "grad_norm": 0.7487357258796692,
+      "learning_rate": 3.1878098085814924e-05,
+      "loss": 0.1652,
+      "step": 735
+    },
+    {
+      "epoch": 6.542222222222223,
+      "grad_norm": 0.8228577971458435,
+      "learning_rate": 3.173294878168025e-05,
+      "loss": 0.097,
+      "step": 736
+    },
+    {
+      "epoch": 6.551111111111111,
+      "grad_norm": 0.9791572690010071,
+      "learning_rate": 3.158797691560378e-05,
+      "loss": 0.1128,
+      "step": 737
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "grad_norm": 0.4378013610839844,
+      "learning_rate": 3.1443183895778105e-05,
+      "loss": 0.101,
+      "step": 738
+    },
+    {
+      "epoch": 6.568888888888889,
+      "grad_norm": 0.5853977799415588,
+      "learning_rate": 3.129857112865859e-05,
+      "loss": 0.1425,
+      "step": 739
+    },
+    {
+      "epoch": 6.5777777777777775,
+      "grad_norm": 0.5795565247535706,
+      "learning_rate": 3.115414001894974e-05,
+      "loss": 0.1255,
+      "step": 740
+    },
+    {
+      "epoch": 6.586666666666667,
+      "grad_norm": 0.5457404851913452,
+      "learning_rate": 3.100989196959145e-05,
+      "loss": 0.1008,
+      "step": 741
+    },
+    {
+      "epoch": 6.595555555555555,
+      "grad_norm": 0.7424979209899902,
+      "learning_rate": 3.086582838174551e-05,
+      "loss": 0.1573,
+      "step": 742
+    },
+    {
+      "epoch": 6.604444444444445,
+      "grad_norm": 0.7952689528465271,
+      "learning_rate": 3.072195065478192e-05,
+      "loss": 0.1342,
+      "step": 743
+    },
+    {
+      "epoch": 6.613333333333333,
+      "grad_norm": 0.5833268761634827,
+      "learning_rate": 3.0578260186265265e-05,
+      "loss": 0.0916,
+      "step": 744
+    },
+    {
+      "epoch": 6.622222222222222,
+      "grad_norm": 0.6129670739173889,
+      "learning_rate": 3.0434758371941258e-05,
+      "loss": 0.0802,
+      "step": 745
+    },
+    {
+      "epoch": 6.631111111111111,
+      "grad_norm": 0.6553269028663635,
+      "learning_rate": 3.029144660572304e-05,
+      "loss": 0.1206,
+      "step": 746
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 0.7755371928215027,
+      "learning_rate": 3.014832627967775e-05,
+      "loss": 0.1356,
+      "step": 747
+    },
+    {
+      "epoch": 6.648888888888889,
+      "grad_norm": 0.5356115102767944,
+      "learning_rate": 3.000539878401296e-05,
+      "loss": 0.1344,
+      "step": 748
+    },
+    {
+      "epoch": 6.657777777777778,
+      "grad_norm": 0.6588380932807922,
+      "learning_rate": 2.9862665507063147e-05,
+      "loss": 0.1449,
+      "step": 749
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 0.5725284218788147,
+      "learning_rate": 2.9720127835276256e-05,
+      "loss": 0.1202,
+      "step": 750
+    },
+    {
+      "epoch": 6.6755555555555555,
+      "grad_norm": 0.6672924160957336,
+      "learning_rate": 2.9577787153200197e-05,
+      "loss": 0.0947,
+      "step": 751
+    },
+    {
+      "epoch": 6.684444444444445,
+      "grad_norm": 0.6932110786437988,
+      "learning_rate": 2.9435644843469436e-05,
+      "loss": 0.1466,
+      "step": 752
+    },
+    {
+      "epoch": 6.693333333333333,
+      "grad_norm": 0.807898998260498,
+      "learning_rate": 2.929370228679149e-05,
+      "loss": 0.1349,
+      "step": 753
+    },
+    {
+      "epoch": 6.702222222222222,
+      "grad_norm": 0.4961835443973541,
+      "learning_rate": 2.9151960861933614e-05,
+      "loss": 0.1696,
+      "step": 754
+    },
+    {
+      "epoch": 6.711111111111111,
+      "grad_norm": 1.539179801940918,
+      "learning_rate": 2.901042194570931e-05,
+      "loss": 0.129,
+      "step": 755
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.46866491436958313,
+      "learning_rate": 2.886908691296504e-05,
+      "loss": 0.1299,
+      "step": 756
+    },
+    {
+      "epoch": 6.728888888888889,
+      "grad_norm": 0.5181779265403748,
+      "learning_rate": 2.8727957136566823e-05,
+      "loss": 0.1131,
+      "step": 757
+    },
+    {
+      "epoch": 6.737777777777778,
+      "grad_norm": 0.785124659538269,
+      "learning_rate": 2.858703398738686e-05,
+      "loss": 0.1055,
+      "step": 758
+    },
+    {
+      "epoch": 6.746666666666667,
+      "grad_norm": 0.4711576998233795,
+      "learning_rate": 2.8446318834290358e-05,
+      "loss": 0.1572,
+      "step": 759
+    },
+    {
+      "epoch": 6.7555555555555555,
+      "grad_norm": 0.5323825478553772,
+      "learning_rate": 2.8305813044122097e-05,
+      "loss": 0.1385,
+      "step": 760
+    },
+    {
+      "epoch": 6.764444444444445,
+      "grad_norm": 0.6163938641548157,
+      "learning_rate": 2.8165517981693157e-05,
+      "loss": 0.1445,
+      "step": 761
+    },
+    {
+      "epoch": 6.773333333333333,
+      "grad_norm": 0.6116064190864563,
+      "learning_rate": 2.8025435009767747e-05,
+      "loss": 0.0994,
+      "step": 762
+    },
+    {
+      "epoch": 6.782222222222222,
+      "grad_norm": 0.6490322947502136,
+      "learning_rate": 2.7885565489049946e-05,
+      "loss": 0.1374,
+      "step": 763
+    },
+    {
+      "epoch": 6.791111111111111,
+      "grad_norm": 0.8776857852935791,
+      "learning_rate": 2.774591077817038e-05,
+      "loss": 0.1,
+      "step": 764
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.44012898206710815,
+      "learning_rate": 2.7606472233673186e-05,
+      "loss": 0.1526,
+      "step": 765
+    },
+    {
+      "epoch": 6.808888888888889,
+      "grad_norm": 0.5889332890510559,
+      "learning_rate": 2.746725121000273e-05,
+      "loss": 0.1418,
+      "step": 766
+    },
+    {
+      "epoch": 6.817777777777778,
+      "grad_norm": 0.5039485692977905,
+      "learning_rate": 2.732824905949049e-05,
+      "loss": 0.175,
+      "step": 767
+    },
+    {
+      "epoch": 6.826666666666666,
+      "grad_norm": 0.5044254660606384,
+      "learning_rate": 2.718946713234185e-05,
+      "loss": 0.1538,
+      "step": 768
+    },
+    {
+      "epoch": 6.835555555555556,
+      "grad_norm": 0.43058761954307556,
+      "learning_rate": 2.705090677662311e-05,
+      "loss": 0.1376,
+      "step": 769
+    },
+    {
+      "epoch": 6.844444444444444,
+      "grad_norm": 0.7214236259460449,
+      "learning_rate": 2.6912569338248315e-05,
+      "loss": 0.1195,
+      "step": 770
+    },
+    {
+      "epoch": 6.8533333333333335,
+      "grad_norm": 1.1327557563781738,
+      "learning_rate": 2.6774456160966122e-05,
+      "loss": 0.1503,
+      "step": 771
+    },
+    {
+      "epoch": 6.862222222222222,
+      "grad_norm": 0.45260992646217346,
+      "learning_rate": 2.66365685863469e-05,
+      "loss": 0.1429,
+      "step": 772
+    },
+    {
+      "epoch": 6.871111111111111,
+      "grad_norm": 0.46228495240211487,
+      "learning_rate": 2.6498907953769582e-05,
+      "loss": 0.0988,
+      "step": 773
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 0.5020700097084045,
+      "learning_rate": 2.636147560040866e-05,
+      "loss": 0.1107,
+      "step": 774
+    },
+    {
+      "epoch": 6.888888888888889,
+      "grad_norm": 0.45958662033081055,
+      "learning_rate": 2.6224272861221243e-05,
+      "loss": 0.1518,
+      "step": 775
+    },
+    {
+      "epoch": 6.897777777777778,
+      "grad_norm": 0.5302269458770752,
+      "learning_rate": 2.6087301068934106e-05,
+      "loss": 0.1464,
+      "step": 776
+    },
+    {
+      "epoch": 6.906666666666666,
+      "grad_norm": 0.432018518447876,
+      "learning_rate": 2.595056155403063e-05,
+      "loss": 0.1781,
+      "step": 777
+    },
+    {
+      "epoch": 6.915555555555556,
+      "grad_norm": 0.48671501874923706,
+      "learning_rate": 2.581405564473801e-05,
+      "loss": 0.1276,
+      "step": 778
+    },
+    {
+      "epoch": 6.924444444444444,
+      "grad_norm": 0.4488458037376404,
+      "learning_rate": 2.567778466701431e-05,
+      "loss": 0.1344,
+      "step": 779
+    },
+    {
+      "epoch": 6.933333333333334,
+      "grad_norm": 0.47056058049201965,
+      "learning_rate": 2.5541749944535554e-05,
+      "loss": 0.1317,
+      "step": 780
+    },
+    {
+      "epoch": 6.942222222222222,
+      "grad_norm": 0.6379369497299194,
+      "learning_rate": 2.5405952798682842e-05,
+      "loss": 0.1521,
+      "step": 781
+    },
+    {
+      "epoch": 6.9511111111111115,
+      "grad_norm": 0.8954217433929443,
+      "learning_rate": 2.527039454852963e-05,
+      "loss": 0.1355,
+      "step": 782
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 0.6195422410964966,
+      "learning_rate": 2.5135076510828837e-05,
+      "loss": 0.1059,
+      "step": 783
+    },
+    {
+      "epoch": 6.968888888888889,
+      "grad_norm": 0.6319138407707214,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.1139,
+      "step": 784
+    },
+    {
+      "epoch": 6.977777777777778,
+      "grad_norm": 0.49685609340667725,
+      "learning_rate": 2.486516632811666e-05,
+      "loss": 0.1271,
+      "step": 785
+    },
+    {
+      "epoch": 6.986666666666666,
+      "grad_norm": 0.7997812628746033,
+      "learning_rate": 2.473057680489348e-05,
+      "loss": 0.1291,
+      "step": 786
+    },
+    {
+      "epoch": 6.995555555555556,
+      "grad_norm": 0.609612226486206,
+      "learning_rate": 2.459623273767354e-05,
+      "loss": 0.1428,
+      "step": 787
+    },
+    {
+      "epoch": 7.004444444444444,
+      "grad_norm": 0.4185393452644348,
+      "learning_rate": 2.4462135431415733e-05,
+      "loss": 0.1231,
+      "step": 788
+    },
+    {
+      "epoch": 7.013333333333334,
+      "grad_norm": 0.6450554132461548,
+      "learning_rate": 2.4328286188681987e-05,
+      "loss": 0.101,
+      "step": 789
+    },
+    {
+      "epoch": 7.022222222222222,
+      "grad_norm": 0.4551238417625427,
+      "learning_rate": 2.4194686309624663e-05,
+      "loss": 0.1042,
+      "step": 790
+    },
+    {
+      "epoch": 7.0311111111111115,
+      "grad_norm": 0.7651455998420715,
+      "learning_rate": 2.4061337091973918e-05,
+      "loss": 0.1206,
+      "step": 791
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.46599751710891724,
+      "learning_rate": 2.39282398310251e-05,
+      "loss": 0.1477,
+      "step": 792
+    },
+    {
+      "epoch": 7.0488888888888885,
+      "grad_norm": 0.46619459986686707,
+      "learning_rate": 2.3795395819626114e-05,
+      "loss": 0.1203,
+      "step": 793
+    },
+    {
+      "epoch": 7.057777777777778,
+      "grad_norm": 1.243935465812683,
+      "learning_rate": 2.366280634816496e-05,
+      "loss": 0.1135,
+      "step": 794
+    },
+    {
+      "epoch": 7.066666666666666,
+      "grad_norm": 0.5829083919525146,
+      "learning_rate": 2.3530472704557145e-05,
+      "loss": 0.0728,
+      "step": 795
+    },
+    {
+      "epoch": 7.075555555555556,
+      "grad_norm": 0.5312824845314026,
+      "learning_rate": 2.3398396174233178e-05,
+      "loss": 0.0761,
+      "step": 796
+    },
+    {
+      "epoch": 7.084444444444444,
+      "grad_norm": 0.6193990111351013,
+      "learning_rate": 2.3266578040126036e-05,
+      "loss": 0.0875,
+      "step": 797
+    },
+    {
+      "epoch": 7.093333333333334,
+      "grad_norm": 0.4868319034576416,
+      "learning_rate": 2.3135019582658802e-05,
+      "loss": 0.0876,
+      "step": 798
+    },
+    {
+      "epoch": 7.102222222222222,
+      "grad_norm": 0.4497816860675812,
+      "learning_rate": 2.300372207973219e-05,
+      "loss": 0.0989,
+      "step": 799
+    },
+    {
+      "epoch": 7.111111111111111,
+      "grad_norm": 0.4699987769126892,
+      "learning_rate": 2.2872686806712035e-05,
+      "loss": 0.1211,
+      "step": 800
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 0.7330834865570068,
+      "learning_rate": 2.274191503641706e-05,
+      "loss": 0.0784,
+      "step": 801
+    },
+    {
+      "epoch": 7.128888888888889,
+      "grad_norm": 0.3836953938007355,
+      "learning_rate": 2.261140803910644e-05,
+      "loss": 0.0889,
+      "step": 802
+    },
+    {
+      "epoch": 7.137777777777778,
+      "grad_norm": 0.528725266456604,
+      "learning_rate": 2.2481167082467403e-05,
+      "loss": 0.1309,
+      "step": 803
+    },
+    {
+      "epoch": 7.1466666666666665,
+      "grad_norm": 0.5461883544921875,
+      "learning_rate": 2.235119343160303e-05,
+      "loss": 0.1064,
+      "step": 804
+    },
+    {
+      "epoch": 7.155555555555556,
+      "grad_norm": 0.47903427481651306,
+      "learning_rate": 2.2221488349019903e-05,
+      "loss": 0.1347,
+      "step": 805
+    },
+    {
+      "epoch": 7.164444444444444,
+      "grad_norm": 0.43539464473724365,
+      "learning_rate": 2.2092053094615813e-05,
+      "loss": 0.0761,
+      "step": 806
+    },
+    {
+      "epoch": 7.173333333333334,
+      "grad_norm": 0.5986665487289429,
+      "learning_rate": 2.1962888925667603e-05,
+      "loss": 0.0729,
+      "step": 807
+    },
+    {
+      "epoch": 7.182222222222222,
+      "grad_norm": 0.5698485970497131,
+      "learning_rate": 2.1833997096818898e-05,
+      "loss": 0.1272,
+      "step": 808
+    },
+    {
+      "epoch": 7.191111111111111,
+      "grad_norm": 0.5316697359085083,
+      "learning_rate": 2.1705378860067944e-05,
+      "loss": 0.1396,
+      "step": 809
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.5823459029197693,
+      "learning_rate": 2.157703546475539e-05,
+      "loss": 0.1466,
+      "step": 810
+    },
+    {
+      "epoch": 7.208888888888889,
+      "grad_norm": 0.4261781573295593,
+      "learning_rate": 2.144896815755224e-05,
+      "loss": 0.0868,
+      "step": 811
+    },
+    {
+      "epoch": 7.217777777777778,
+      "grad_norm": 0.47086384892463684,
+      "learning_rate": 2.132117818244771e-05,
+      "loss": 0.15,
+      "step": 812
+    },
+    {
+      "epoch": 7.226666666666667,
+      "grad_norm": 0.43166688084602356,
+      "learning_rate": 2.119366678073707e-05,
+      "loss": 0.0989,
+      "step": 813
+    },
+    {
+      "epoch": 7.235555555555556,
+      "grad_norm": 0.7396053075790405,
+      "learning_rate": 2.1066435191009715e-05,
+      "loss": 0.0756,
+      "step": 814
+    },
+    {
+      "epoch": 7.2444444444444445,
+      "grad_norm": 0.4421260952949524,
+      "learning_rate": 2.093948464913707e-05,
+      "loss": 0.1058,
+      "step": 815
+    },
+    {
+      "epoch": 7.253333333333333,
+      "grad_norm": 0.9845119714736938,
+      "learning_rate": 2.0812816388260518e-05,
+      "loss": 0.1217,
+      "step": 816
+    },
+    {
+      "epoch": 7.262222222222222,
+      "grad_norm": 0.5246585607528687,
+      "learning_rate": 2.0686431638779564e-05,
+      "loss": 0.0857,
+      "step": 817
+    },
+    {
+      "epoch": 7.271111111111111,
+      "grad_norm": 0.4285733997821808,
+      "learning_rate": 2.056033162833977e-05,
+      "loss": 0.0885,
+      "step": 818
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 1.0954898595809937,
+      "learning_rate": 2.0434517581820896e-05,
+      "loss": 0.1127,
+      "step": 819
+    },
+    {
+      "epoch": 7.288888888888889,
+      "grad_norm": 0.6235244274139404,
+      "learning_rate": 2.0308990721324927e-05,
+      "loss": 0.1219,
+      "step": 820
+    },
+    {
+      "epoch": 7.297777777777778,
+      "grad_norm": 1.4151173830032349,
+      "learning_rate": 2.01837522661643e-05,
+      "loss": 0.1308,
+      "step": 821
+    },
+    {
+      "epoch": 7.306666666666667,
+      "grad_norm": 0.6299641132354736,
+      "learning_rate": 2.0058803432849987e-05,
+      "loss": 0.1403,
+      "step": 822
+    },
+    {
+      "epoch": 7.315555555555555,
+      "grad_norm": 0.5858526825904846,
+      "learning_rate": 1.9934145435079702e-05,
+      "loss": 0.0698,
+      "step": 823
+    },
+    {
+      "epoch": 7.3244444444444445,
+      "grad_norm": 0.43147486448287964,
+      "learning_rate": 1.980977948372612e-05,
+      "loss": 0.1261,
+      "step": 824
+    },
+    {
+      "epoch": 7.333333333333333,
+      "grad_norm": 0.6104679703712463,
+      "learning_rate": 1.9685706786825114e-05,
+      "loss": 0.0826,
+      "step": 825
+    },
+    {
+      "epoch": 7.342222222222222,
+      "grad_norm": 0.48661118745803833,
+      "learning_rate": 1.9561928549563968e-05,
+      "loss": 0.1047,
+      "step": 826
+    },
+    {
+      "epoch": 7.351111111111111,
+      "grad_norm": 0.5598762035369873,
+      "learning_rate": 1.9438445974269754e-05,
+      "loss": 0.1111,
+      "step": 827
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.6354761719703674,
+      "learning_rate": 1.931526026039764e-05,
+      "loss": 0.1158,
+      "step": 828
+    },
+    {
+      "epoch": 7.368888888888889,
+      "grad_norm": 0.59492427110672,
+      "learning_rate": 1.9192372604519127e-05,
+      "loss": 0.1495,
+      "step": 829
+    },
+    {
+      "epoch": 7.377777777777778,
+      "grad_norm": 0.5921106934547424,
+      "learning_rate": 1.906978420031059e-05,
+      "loss": 0.1477,
+      "step": 830
+    },
+    {
+      "epoch": 7.386666666666667,
+      "grad_norm": 0.5231952667236328,
+      "learning_rate": 1.8947496238541612e-05,
+      "loss": 0.072,
+      "step": 831
+    },
+    {
+      "epoch": 7.395555555555555,
+      "grad_norm": 0.7299399971961975,
+      "learning_rate": 1.8825509907063327e-05,
+      "loss": 0.0799,
+      "step": 832
+    },
+    {
+      "epoch": 7.404444444444445,
+      "grad_norm": 0.545674204826355,
+      "learning_rate": 1.8703826390797048e-05,
+      "loss": 0.1354,
+      "step": 833
+    },
+    {
+      "epoch": 7.413333333333333,
+      "grad_norm": 0.5333763360977173,
+      "learning_rate": 1.8582446871722636e-05,
+      "loss": 0.0966,
+      "step": 834
+    },
+    {
+      "epoch": 7.4222222222222225,
+      "grad_norm": 0.6203194260597229,
+      "learning_rate": 1.8461372528867093e-05,
+      "loss": 0.1415,
+      "step": 835
+    },
+    {
+      "epoch": 7.431111111111111,
+      "grad_norm": 0.39000070095062256,
+      "learning_rate": 1.8340604538293015e-05,
+      "loss": 0.1476,
+      "step": 836
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 0.8264181613922119,
+      "learning_rate": 1.822014407308727e-05,
+      "loss": 0.1139,
+      "step": 837
+    },
+    {
+      "epoch": 7.448888888888889,
+      "grad_norm": 0.4161231815814972,
+      "learning_rate": 1.8099992303349577e-05,
+      "loss": 0.1079,
+      "step": 838
+    },
+    {
+      "epoch": 7.457777777777777,
+      "grad_norm": 0.48988258838653564,
+      "learning_rate": 1.798015039618106e-05,
+      "loss": 0.0761,
+      "step": 839
+    },
+    {
+      "epoch": 7.466666666666667,
+      "grad_norm": 0.573543906211853,
+      "learning_rate": 1.7860619515673033e-05,
+      "loss": 0.1233,
+      "step": 840
+    },
+    {
+      "epoch": 7.475555555555555,
+      "grad_norm": 1.0791929960250854,
+      "learning_rate": 1.774140082289563e-05,
+      "loss": 0.0833,
+      "step": 841
+    },
+    {
+      "epoch": 7.484444444444445,
+      "grad_norm": 0.7512285113334656,
+      "learning_rate": 1.7622495475886487e-05,
+      "loss": 0.1107,
+      "step": 842
+    },
+    {
+      "epoch": 7.493333333333333,
+      "grad_norm": 0.4891441762447357,
+      "learning_rate": 1.7503904629639574e-05,
+      "loss": 0.0675,
+      "step": 843
+    },
+    {
+      "epoch": 7.502222222222223,
+      "grad_norm": 0.3364730775356293,
+      "learning_rate": 1.738562943609396e-05,
+      "loss": 0.0884,
+      "step": 844
+    },
+    {
+      "epoch": 7.511111111111111,
+      "grad_norm": 1.6392979621887207,
+      "learning_rate": 1.7267671044122534e-05,
+      "loss": 0.1564,
+      "step": 845
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.6125866770744324,
+      "learning_rate": 1.7150030599520984e-05,
+      "loss": 0.1495,
+      "step": 846
+    },
+    {
+      "epoch": 7.528888888888889,
+      "grad_norm": 0.6755143404006958,
+      "learning_rate": 1.703270924499656e-05,
+      "loss": 0.0962,
+      "step": 847
+    },
+    {
+      "epoch": 7.5377777777777775,
+      "grad_norm": 0.5426997542381287,
+      "learning_rate": 1.691570812015704e-05,
+      "loss": 0.0945,
+      "step": 848
+    },
+    {
+      "epoch": 7.546666666666667,
+      "grad_norm": 0.6936843395233154,
+      "learning_rate": 1.679902836149959e-05,
+      "loss": 0.1192,
+      "step": 849
+    },
+    {
+      "epoch": 7.555555555555555,
+      "grad_norm": 0.5444396138191223,
+      "learning_rate": 1.6682671102399805e-05,
+      "loss": 0.0844,
+      "step": 850
+    },
+    {
+      "epoch": 7.564444444444445,
+      "grad_norm": 0.5248525142669678,
+      "learning_rate": 1.6566637473100667e-05,
+      "loss": 0.1279,
+      "step": 851
+    },
+    {
+      "epoch": 7.573333333333333,
+      "grad_norm": 0.488179087638855,
+      "learning_rate": 1.6450928600701504e-05,
+      "loss": 0.1259,
+      "step": 852
+    },
+    {
+      "epoch": 7.582222222222223,
+      "grad_norm": 0.4678577184677124,
+      "learning_rate": 1.6335545609147142e-05,
+      "loss": 0.0997,
+      "step": 853
+    },
+    {
+      "epoch": 7.591111111111111,
+      "grad_norm": 0.5950666666030884,
+      "learning_rate": 1.622048961921699e-05,
+      "loss": 0.105,
+      "step": 854
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.7520263195037842,
+      "learning_rate": 1.6105761748514007e-05,
+      "loss": 0.143,
+      "step": 855
+    },
+    {
+      "epoch": 7.608888888888889,
+      "grad_norm": 0.45088163018226624,
+      "learning_rate": 1.599136311145402e-05,
+      "loss": 0.1162,
+      "step": 856
+    },
+    {
+      "epoch": 7.6177777777777775,
+      "grad_norm": 0.46655556559562683,
+      "learning_rate": 1.5877294819254822e-05,
+      "loss": 0.1449,
+      "step": 857
+    },
+    {
+      "epoch": 7.626666666666667,
+      "grad_norm": 0.5721983313560486,
+      "learning_rate": 1.5763557979925324e-05,
+      "loss": 0.1131,
+      "step": 858
+    },
+    {
+      "epoch": 7.635555555555555,
+      "grad_norm": 0.5537592172622681,
+      "learning_rate": 1.5650153698254916e-05,
+      "loss": 0.1206,
+      "step": 859
+    },
+    {
+      "epoch": 7.644444444444445,
+      "grad_norm": 0.4676898419857025,
+      "learning_rate": 1.553708307580265e-05,
+      "loss": 0.1245,
+      "step": 860
+    },
+    {
+      "epoch": 7.653333333333333,
+      "grad_norm": 0.9704411625862122,
+      "learning_rate": 1.5424347210886538e-05,
+      "loss": 0.0839,
+      "step": 861
+    },
+    {
+      "epoch": 7.662222222222223,
+      "grad_norm": 0.5594210624694824,
+      "learning_rate": 1.531194719857292e-05,
+      "loss": 0.0942,
+      "step": 862
+    },
+    {
+      "epoch": 7.671111111111111,
+      "grad_norm": 0.8229500651359558,
+      "learning_rate": 1.5199884130665831e-05,
+      "loss": 0.0732,
+      "step": 863
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.5628749132156372,
+      "learning_rate": 1.5088159095696363e-05,
+      "loss": 0.1224,
+      "step": 864
+    },
+    {
+      "epoch": 7.688888888888889,
+      "grad_norm": 0.6322095394134521,
+      "learning_rate": 1.4976773178912084e-05,
+      "loss": 0.0854,
+      "step": 865
+    },
+    {
+      "epoch": 7.697777777777778,
+      "grad_norm": 0.7371559143066406,
+      "learning_rate": 1.4865727462266543e-05,
+      "loss": 0.1024,
+      "step": 866
+    },
+    {
+      "epoch": 7.706666666666667,
+      "grad_norm": 1.2312216758728027,
+      "learning_rate": 1.4755023024408748e-05,
+      "loss": 0.14,
+      "step": 867
+    },
+    {
+      "epoch": 7.7155555555555555,
+      "grad_norm": 0.7259622812271118,
+      "learning_rate": 1.4644660940672627e-05,
+      "loss": 0.0879,
+      "step": 868
+    },
+    {
+      "epoch": 7.724444444444444,
+      "grad_norm": 0.6174317002296448,
+      "learning_rate": 1.453464228306668e-05,
+      "loss": 0.0905,
+      "step": 869
+    },
+    {
+      "epoch": 7.733333333333333,
+      "grad_norm": 0.46135640144348145,
+      "learning_rate": 1.4424968120263504e-05,
+      "loss": 0.1019,
+      "step": 870
+    },
+    {
+      "epoch": 7.742222222222222,
+      "grad_norm": 0.983392596244812,
+      "learning_rate": 1.4315639517589397e-05,
+      "loss": 0.1439,
+      "step": 871
+    },
+    {
+      "epoch": 7.751111111111111,
+      "grad_norm": 0.5525241494178772,
+      "learning_rate": 1.4206657537014079e-05,
+      "loss": 0.1483,
+      "step": 872
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 0.4846337139606476,
+      "learning_rate": 1.4098023237140334e-05,
+      "loss": 0.145,
+      "step": 873
+    },
+    {
+      "epoch": 7.768888888888889,
+      "grad_norm": 0.5754955410957336,
+      "learning_rate": 1.398973767319368e-05,
+      "loss": 0.0999,
+      "step": 874
+    },
+    {
+      "epoch": 7.777777777777778,
+      "grad_norm": 0.567977786064148,
+      "learning_rate": 1.3881801897012225e-05,
+      "loss": 0.1267,
+      "step": 875
+    },
+    {
+      "epoch": 7.786666666666667,
+      "grad_norm": 0.6116244792938232,
+      "learning_rate": 1.3774216957036367e-05,
+      "loss": 0.0912,
+      "step": 876
+    },
+    {
+      "epoch": 7.795555555555556,
+      "grad_norm": 0.5989839434623718,
+      "learning_rate": 1.3666983898298657e-05,
+      "loss": 0.097,
+      "step": 877
+    },
+    {
+      "epoch": 7.804444444444444,
+      "grad_norm": 0.6098581552505493,
+      "learning_rate": 1.3560103762413584e-05,
+      "loss": 0.0789,
+      "step": 878
+    },
+    {
+      "epoch": 7.8133333333333335,
+      "grad_norm": 0.4235369563102722,
+      "learning_rate": 1.345357758756754e-05,
+      "loss": 0.1303,
+      "step": 879
+    },
+    {
+      "epoch": 7.822222222222222,
+      "grad_norm": 0.5483813881874084,
+      "learning_rate": 1.3347406408508695e-05,
+      "loss": 0.0967,
+      "step": 880
+    },
+    {
+      "epoch": 7.831111111111111,
+      "grad_norm": 0.5754316449165344,
+      "learning_rate": 1.3241591256536917e-05,
+      "loss": 0.0806,
+      "step": 881
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 1.0643495321273804,
+      "learning_rate": 1.3136133159493802e-05,
+      "loss": 0.0715,
+      "step": 882
+    },
+    {
+      "epoch": 7.848888888888889,
+      "grad_norm": 0.6817154884338379,
+      "learning_rate": 1.3031033141752702e-05,
+      "loss": 0.1125,
+      "step": 883
+    },
+    {
+      "epoch": 7.857777777777778,
+      "grad_norm": 0.5241005420684814,
+      "learning_rate": 1.2926292224208664e-05,
+      "loss": 0.0945,
+      "step": 884
+    },
+    {
+      "epoch": 7.866666666666667,
+      "grad_norm": 0.5982441902160645,
+      "learning_rate": 1.282191142426869e-05,
+      "loss": 0.1098,
+      "step": 885
+    },
+    {
+      "epoch": 7.875555555555556,
+      "grad_norm": 0.49455299973487854,
+      "learning_rate": 1.2717891755841722e-05,
+      "loss": 0.0987,
+      "step": 886
+    },
+    {
+      "epoch": 7.884444444444444,
+      "grad_norm": 0.5417470335960388,
+      "learning_rate": 1.2614234229328792e-05,
+      "loss": 0.0931,
+      "step": 887
+    },
+    {
+      "epoch": 7.8933333333333335,
+      "grad_norm": 0.6845139265060425,
+      "learning_rate": 1.2510939851613285e-05,
+      "loss": 0.1042,
+      "step": 888
+    },
+    {
+      "epoch": 7.902222222222222,
+      "grad_norm": 0.585252583026886,
+      "learning_rate": 1.2408009626051137e-05,
+      "loss": 0.1561,
+      "step": 889
+    },
+    {
+      "epoch": 7.911111111111111,
+      "grad_norm": 1.172561764717102,
+      "learning_rate": 1.230544455246101e-05,
+      "loss": 0.1128,
+      "step": 890
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.5593766570091248,
+      "learning_rate": 1.2203245627114685e-05,
+      "loss": 0.1115,
+      "step": 891
+    },
+    {
+      "epoch": 7.928888888888888,
+      "grad_norm": 0.7291374802589417,
+      "learning_rate": 1.2101413842727345e-05,
+      "loss": 0.0952,
+      "step": 892
+    },
+    {
+      "epoch": 7.937777777777778,
+      "grad_norm": 0.4604465961456299,
+      "learning_rate": 1.1999950188447922e-05,
+      "loss": 0.1461,
+      "step": 893
+    },
+    {
+      "epoch": 7.946666666666666,
+      "grad_norm": 0.43182215094566345,
+      "learning_rate": 1.1898855649849461e-05,
+      "loss": 0.1466,
+      "step": 894
+    },
+    {
+      "epoch": 7.955555555555556,
+      "grad_norm": 0.4441966116428375,
+      "learning_rate": 1.1798131208919627e-05,
+      "loss": 0.1075,
+      "step": 895
+    },
+    {
+      "epoch": 7.964444444444444,
+      "grad_norm": 0.5411422252655029,
+      "learning_rate": 1.1697777844051105e-05,
+      "loss": 0.1224,
+      "step": 896
+    },
+    {
+      "epoch": 7.973333333333334,
+      "grad_norm": 0.4580567479133606,
+      "learning_rate": 1.1597796530032085e-05,
+      "loss": 0.1341,
+      "step": 897
+    },
+    {
+      "epoch": 7.982222222222222,
+      "grad_norm": 0.5973615646362305,
+      "learning_rate": 1.1498188238036861e-05,
+      "loss": 0.0929,
+      "step": 898
+    },
+    {
+      "epoch": 7.9911111111111115,
+      "grad_norm": 0.6847777366638184,
+      "learning_rate": 1.1398953935616336e-05,
+      "loss": 0.0985,
+      "step": 899
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.6161903142929077,
+      "learning_rate": 1.130009458668863e-05,
+      "loss": 0.1104,
+      "step": 900
+    },
+    {
+      "epoch": 8.008888888888889,
+      "grad_norm": 0.46788787841796875,
+      "learning_rate": 1.1201611151529756e-05,
+      "loss": 0.0999,
+      "step": 901
+    },
+    {
+      "epoch": 8.017777777777777,
+      "grad_norm": 0.3665621876716614,
+      "learning_rate": 1.1103504586764263e-05,
+      "loss": 0.0823,
+      "step": 902
+    },
+    {
+      "epoch": 8.026666666666667,
+      "grad_norm": 0.4475909173488617,
+      "learning_rate": 1.100577584535592e-05,
+      "loss": 0.1003,
+      "step": 903
+    },
+    {
+      "epoch": 8.035555555555556,
+      "grad_norm": 0.37525415420532227,
+      "learning_rate": 1.090842587659851e-05,
+      "loss": 0.1119,
+      "step": 904
+    },
+    {
+      "epoch": 8.044444444444444,
+      "grad_norm": 0.4087243676185608,
+      "learning_rate": 1.0811455626106599e-05,
+      "loss": 0.0782,
+      "step": 905
+    },
+    {
+      "epoch": 8.053333333333333,
+      "grad_norm": 0.3830251693725586,
+      "learning_rate": 1.0714866035806326e-05,
+      "loss": 0.1224,
+      "step": 906
+    },
+    {
+      "epoch": 8.062222222222223,
+      "grad_norm": 0.49408161640167236,
+      "learning_rate": 1.0618658043926233e-05,
+      "loss": 0.0887,
+      "step": 907
+    },
+    {
+      "epoch": 8.071111111111112,
+      "grad_norm": 0.4445248246192932,
+      "learning_rate": 1.0522832584988234e-05,
+      "loss": 0.1233,
+      "step": 908
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.7690616250038147,
+      "learning_rate": 1.0427390589798469e-05,
+      "loss": 0.1152,
+      "step": 909
+    },
+    {
+      "epoch": 8.088888888888889,
+      "grad_norm": 0.4401998221874237,
+      "learning_rate": 1.0332332985438248e-05,
+      "loss": 0.1095,
+      "step": 910
+    },
+    {
+      "epoch": 8.097777777777777,
+      "grad_norm": 0.4737964868545532,
+      "learning_rate": 1.023766069525513e-05,
+      "loss": 0.1122,
+      "step": 911
+    },
+    {
+      "epoch": 8.106666666666667,
+      "grad_norm": 0.464690625667572,
+      "learning_rate": 1.0143374638853891e-05,
+      "loss": 0.0811,
+      "step": 912
+    },
+    {
+      "epoch": 8.115555555555556,
+      "grad_norm": 0.5359342694282532,
+      "learning_rate": 1.004947573208756e-05,
+      "loss": 0.076,
+      "step": 913
+    },
+    {
+      "epoch": 8.124444444444444,
+      "grad_norm": 0.641705334186554,
+      "learning_rate": 9.955964887048607e-06,
+      "loss": 0.1052,
+      "step": 914
+    },
+    {
+      "epoch": 8.133333333333333,
+      "grad_norm": 0.4826803505420685,
+      "learning_rate": 9.862843012060047e-06,
+      "loss": 0.0885,
+      "step": 915
+    },
+    {
+      "epoch": 8.142222222222221,
+      "grad_norm": 0.41360554099082947,
+      "learning_rate": 9.770111011666583e-06,
+      "loss": 0.0934,
+      "step": 916
+    },
+    {
+      "epoch": 8.151111111111112,
+      "grad_norm": 0.44290822744369507,
+      "learning_rate": 9.677769786625867e-06,
+      "loss": 0.1115,
+      "step": 917
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 1.0914057493209839,
+      "learning_rate": 9.58582023389974e-06,
+      "loss": 0.0933,
+      "step": 918
+    },
+    {
+      "epoch": 8.168888888888889,
+      "grad_norm": 1.4697805643081665,
+      "learning_rate": 9.494263246645474e-06,
+      "loss": 0.154,
+      "step": 919
+    },
+    {
+      "epoch": 8.177777777777777,
+      "grad_norm": 0.4320299029350281,
+      "learning_rate": 9.403099714207175e-06,
+      "loss": 0.0527,
+      "step": 920
+    },
+    {
+      "epoch": 8.186666666666667,
+      "grad_norm": 0.5250918865203857,
+      "learning_rate": 9.312330522107076e-06,
+      "loss": 0.1327,
+      "step": 921
+    },
+    {
+      "epoch": 8.195555555555556,
+      "grad_norm": 0.49150344729423523,
+      "learning_rate": 9.221956552036992e-06,
+      "loss": 0.06,
+      "step": 922
+    },
+    {
+      "epoch": 8.204444444444444,
+      "grad_norm": 0.4893438518047333,
+      "learning_rate": 9.131978681849684e-06,
+      "loss": 0.1078,
+      "step": 923
+    },
+    {
+      "epoch": 8.213333333333333,
+      "grad_norm": 0.6123852133750916,
+      "learning_rate": 9.042397785550405e-06,
+      "loss": 0.0889,
+      "step": 924
+    },
+    {
+      "epoch": 8.222222222222221,
+      "grad_norm": 0.4057047963142395,
+      "learning_rate": 8.953214733288383e-06,
+      "loss": 0.0751,
+      "step": 925
+    },
+    {
+      "epoch": 8.231111111111112,
+      "grad_norm": 0.5074282288551331,
+      "learning_rate": 8.864430391348332e-06,
+      "loss": 0.092,
+      "step": 926
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 0.6459361910820007,
+      "learning_rate": 8.776045622142098e-06,
+      "loss": 0.0532,
+      "step": 927
+    },
+    {
+      "epoch": 8.248888888888889,
+      "grad_norm": 0.4968688189983368,
+      "learning_rate": 8.688061284200266e-06,
+      "loss": 0.0794,
+      "step": 928
+    },
+    {
+      "epoch": 8.257777777777777,
+      "grad_norm": 0.5141646265983582,
+      "learning_rate": 8.600478232163755e-06,
+      "loss": 0.1339,
+      "step": 929
+    },
+    {
+      "epoch": 8.266666666666667,
+      "grad_norm": 0.864693820476532,
+      "learning_rate": 8.513297316775625e-06,
+      "loss": 0.0959,
+      "step": 930
+    },
+    {
+      "epoch": 8.275555555555556,
+      "grad_norm": 0.42371752858161926,
+      "learning_rate": 8.426519384872733e-06,
+      "loss": 0.1039,
+      "step": 931
+    },
+    {
+      "epoch": 8.284444444444444,
+      "grad_norm": 0.6697008609771729,
+      "learning_rate": 8.34014527937756e-06,
+      "loss": 0.0884,
+      "step": 932
+    },
+    {
+      "epoch": 8.293333333333333,
+      "grad_norm": 0.47926008701324463,
+      "learning_rate": 8.254175839289941e-06,
+      "loss": 0.0839,
+      "step": 933
+    },
+    {
+      "epoch": 8.302222222222222,
+      "grad_norm": 0.5035369396209717,
+      "learning_rate": 8.168611899679013e-06,
+      "loss": 0.1043,
+      "step": 934
+    },
+    {
+      "epoch": 8.311111111111112,
+      "grad_norm": 0.5710785984992981,
+      "learning_rate": 8.08345429167507e-06,
+      "loss": 0.0937,
+      "step": 935
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 0.5553144216537476,
+      "learning_rate": 7.998703842461431e-06,
+      "loss": 0.0776,
+      "step": 936
+    },
+    {
+      "epoch": 8.328888888888889,
+      "grad_norm": 0.8414843678474426,
+      "learning_rate": 7.914361375266504e-06,
+      "loss": 0.0924,
+      "step": 937
+    },
+    {
+      "epoch": 8.337777777777777,
+      "grad_norm": 0.4151730537414551,
+      "learning_rate": 7.830427709355725e-06,
+      "loss": 0.1119,
+      "step": 938
+    },
+    {
+      "epoch": 8.346666666666668,
+      "grad_norm": 0.7383652329444885,
+      "learning_rate": 7.74690366002359e-06,
+      "loss": 0.0704,
+      "step": 939
+    },
+    {
+      "epoch": 8.355555555555556,
+      "grad_norm": 0.5272123217582703,
+      "learning_rate": 7.663790038585793e-06,
+      "loss": 0.0665,
+      "step": 940
+    },
+    {
+      "epoch": 8.364444444444445,
+      "grad_norm": 0.4467993378639221,
+      "learning_rate": 7.581087652371316e-06,
+      "loss": 0.0703,
+      "step": 941
+    },
+    {
+      "epoch": 8.373333333333333,
+      "grad_norm": 0.45421063899993896,
+      "learning_rate": 7.498797304714544e-06,
+      "loss": 0.1272,
+      "step": 942
+    },
+    {
+      "epoch": 8.382222222222222,
+      "grad_norm": 0.5808837413787842,
+      "learning_rate": 7.416919794947536e-06,
+      "loss": 0.094,
+      "step": 943
+    },
+    {
+      "epoch": 8.391111111111112,
+      "grad_norm": 0.4749268889427185,
+      "learning_rate": 7.33545591839222e-06,
+      "loss": 0.079,
+      "step": 944
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.6306861042976379,
+      "learning_rate": 7.2544064663526815e-06,
+      "loss": 0.1138,
+      "step": 945
+    },
+    {
+      "epoch": 8.408888888888889,
+      "grad_norm": 0.4660092890262604,
+      "learning_rate": 7.173772226107434e-06,
+      "loss": 0.0533,
+      "step": 946
+    },
+    {
+      "epoch": 8.417777777777777,
+      "grad_norm": 0.3973066210746765,
+      "learning_rate": 7.093553980901852e-06,
+      "loss": 0.0845,
+      "step": 947
+    },
+    {
+      "epoch": 8.426666666666666,
+      "grad_norm": 0.5049480199813843,
+      "learning_rate": 7.013752509940485e-06,
+      "loss": 0.1119,
+      "step": 948
+    },
+    {
+      "epoch": 8.435555555555556,
+      "grad_norm": 0.4741290211677551,
+      "learning_rate": 6.934368588379553e-06,
+      "loss": 0.0845,
+      "step": 949
+    },
+    {
+      "epoch": 8.444444444444445,
+      "grad_norm": 0.6338300108909607,
+      "learning_rate": 6.855402987319348e-06,
+      "loss": 0.1058,
+      "step": 950
+    },
+    {
+      "epoch": 8.453333333333333,
+      "grad_norm": 0.8661510944366455,
+      "learning_rate": 6.776856473796828e-06,
+      "loss": 0.1208,
+      "step": 951
+    },
+    {
+      "epoch": 8.462222222222222,
+      "grad_norm": 0.40579918026924133,
+      "learning_rate": 6.698729810778065e-06,
+      "loss": 0.0766,
+      "step": 952
+    },
+    {
+      "epoch": 8.471111111111112,
+      "grad_norm": 0.5100427269935608,
+      "learning_rate": 6.621023757150929e-06,
+      "loss": 0.0978,
+      "step": 953
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.6719834804534912,
+      "learning_rate": 6.54373906771768e-06,
+      "loss": 0.0954,
+      "step": 954
+    },
+    {
+      "epoch": 8.488888888888889,
+      "grad_norm": 0.9201704263687134,
+      "learning_rate": 6.46687649318759e-06,
+      "loss": 0.0886,
+      "step": 955
+    },
+    {
+      "epoch": 8.497777777777777,
+      "grad_norm": 0.5468865633010864,
+      "learning_rate": 6.390436780169734e-06,
+      "loss": 0.0887,
+      "step": 956
+    },
+    {
+      "epoch": 8.506666666666666,
+      "grad_norm": 0.35717347264289856,
+      "learning_rate": 6.314420671165694e-06,
+      "loss": 0.0997,
+      "step": 957
+    },
+    {
+      "epoch": 8.515555555555556,
+      "grad_norm": 0.5313848257064819,
+      "learning_rate": 6.238828904562316e-06,
+      "loss": 0.1077,
+      "step": 958
+    },
+    {
+      "epoch": 8.524444444444445,
+      "grad_norm": 0.5072668194770813,
+      "learning_rate": 6.163662214624616e-06,
+      "loss": 0.0988,
+      "step": 959
+    },
+    {
+      "epoch": 8.533333333333333,
+      "grad_norm": 0.7123374938964844,
+      "learning_rate": 6.088921331488568e-06,
+      "loss": 0.0783,
+      "step": 960
+    },
+    {
+      "epoch": 8.542222222222222,
+      "grad_norm": 0.584987223148346,
+      "learning_rate": 6.014606981154086e-06,
+      "loss": 0.0501,
+      "step": 961
+    },
+    {
+      "epoch": 8.551111111111112,
+      "grad_norm": 0.5454912185668945,
+      "learning_rate": 5.94071988547788e-06,
+      "loss": 0.0881,
+      "step": 962
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.6294575333595276,
+      "learning_rate": 5.867260762166543e-06,
+      "loss": 0.1615,
+      "step": 963
+    },
+    {
+      "epoch": 8.568888888888889,
+      "grad_norm": 0.440383642911911,
+      "learning_rate": 5.794230324769517e-06,
+      "loss": 0.0822,
+      "step": 964
+    },
+    {
+      "epoch": 8.577777777777778,
+      "grad_norm": 0.49135321378707886,
+      "learning_rate": 5.721629282672164e-06,
+      "loss": 0.1226,
+      "step": 965
+    },
+    {
+      "epoch": 8.586666666666666,
+      "grad_norm": 0.3769822120666504,
+      "learning_rate": 5.649458341088915e-06,
+      "loss": 0.1169,
+      "step": 966
+    },
+    {
+      "epoch": 8.595555555555556,
+      "grad_norm": 0.45154932141304016,
+      "learning_rate": 5.577718201056392e-06,
+      "loss": 0.1008,
+      "step": 967
+    },
+    {
+      "epoch": 8.604444444444445,
+      "grad_norm": 0.48603206872940063,
+      "learning_rate": 5.506409559426573e-06,
+      "loss": 0.0974,
+      "step": 968
+    },
+    {
+      "epoch": 8.613333333333333,
+      "grad_norm": 0.9174118638038635,
+      "learning_rate": 5.435533108860086e-06,
+      "loss": 0.0739,
+      "step": 969
+    },
+    {
+      "epoch": 8.622222222222222,
+      "grad_norm": 0.7738699913024902,
+      "learning_rate": 5.365089537819434e-06,
+      "loss": 0.1017,
+      "step": 970
+    },
+    {
+      "epoch": 8.63111111111111,
+      "grad_norm": 0.4519662857055664,
+      "learning_rate": 5.295079530562302e-06,
+      "loss": 0.058,
+      "step": 971
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 0.5471189618110657,
+      "learning_rate": 5.2255037671349535e-06,
+      "loss": 0.089,
+      "step": 972
+    },
+    {
+      "epoch": 8.648888888888889,
+      "grad_norm": 0.40861138701438904,
+      "learning_rate": 5.156362923365588e-06,
+      "loss": 0.1123,
+      "step": 973
+    },
+    {
+      "epoch": 8.657777777777778,
+      "grad_norm": 0.5626519322395325,
+      "learning_rate": 5.087657670857798e-06,
+      "loss": 0.0654,
+      "step": 974
+    },
+    {
+      "epoch": 8.666666666666666,
+      "grad_norm": 0.7317439317703247,
+      "learning_rate": 5.019388676984005e-06,
+      "loss": 0.0452,
+      "step": 975
+    },
+    {
+      "epoch": 8.675555555555556,
+      "grad_norm": 0.454049289226532,
+      "learning_rate": 4.951556604879048e-06,
+      "loss": 0.0782,
+      "step": 976
+    },
+    {
+      "epoch": 8.684444444444445,
+      "grad_norm": 0.6130763292312622,
+      "learning_rate": 4.8841621134336765e-06,
+      "loss": 0.0874,
+      "step": 977
+    },
+    {
+      "epoch": 8.693333333333333,
+      "grad_norm": 0.5617997646331787,
+      "learning_rate": 4.8172058572881765e-06,
+      "loss": 0.065,
+      "step": 978
+    },
+    {
+      "epoch": 8.702222222222222,
+      "grad_norm": 0.5074090957641602,
+      "learning_rate": 4.7506884868259995e-06,
+      "loss": 0.0944,
+      "step": 979
+    },
+    {
+      "epoch": 8.71111111111111,
+      "grad_norm": 2.4399478435516357,
+      "learning_rate": 4.684610648167503e-06,
+      "loss": 0.118,
+      "step": 980
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 0.47364744544029236,
+      "learning_rate": 4.618972983163566e-06,
+      "loss": 0.1199,
+      "step": 981
+    },
+    {
+      "epoch": 8.72888888888889,
+      "grad_norm": 0.5814787149429321,
+      "learning_rate": 4.5537761293894535e-06,
+      "loss": 0.1464,
+      "step": 982
+    },
+    {
+      "epoch": 8.737777777777778,
+      "grad_norm": 0.4782449007034302,
+      "learning_rate": 4.489020720138582e-06,
+      "loss": 0.1058,
+      "step": 983
+    },
+    {
+      "epoch": 8.746666666666666,
+      "grad_norm": 0.4726293981075287,
+      "learning_rate": 4.424707384416344e-06,
+      "loss": 0.0855,
+      "step": 984
+    },
+    {
+      "epoch": 8.755555555555556,
+      "grad_norm": 0.8018205761909485,
+      "learning_rate": 4.360836746934055e-06,
+      "loss": 0.0956,
+      "step": 985
+    },
+    {
+      "epoch": 8.764444444444445,
+      "grad_norm": 0.44194820523262024,
+      "learning_rate": 4.29740942810285e-06,
+      "loss": 0.1166,
+      "step": 986
+    },
+    {
+      "epoch": 8.773333333333333,
+      "grad_norm": 0.5051525235176086,
+      "learning_rate": 4.234426044027645e-06,
+      "loss": 0.1071,
+      "step": 987
+    },
+    {
+      "epoch": 8.782222222222222,
+      "grad_norm": 0.539904773235321,
+      "learning_rate": 4.1718872065011904e-06,
+      "loss": 0.0965,
+      "step": 988
+    },
+    {
+      "epoch": 8.79111111111111,
+      "grad_norm": 0.5747020840644836,
+      "learning_rate": 4.109793522998101e-06,
+      "loss": 0.0666,
+      "step": 989
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.4732227027416229,
+      "learning_rate": 4.048145596668967e-06,
+      "loss": 0.1377,
+      "step": 990
+    },
+    {
+      "epoch": 8.80888888888889,
+      "grad_norm": 0.6295973658561707,
+      "learning_rate": 3.9869440263344714e-06,
+      "loss": 0.0956,
+      "step": 991
+    },
+    {
+      "epoch": 8.817777777777778,
+      "grad_norm": 0.46836766600608826,
+      "learning_rate": 3.9261894064796135e-06,
+      "loss": 0.1131,
+      "step": 992
+    },
+    {
+      "epoch": 8.826666666666666,
+      "grad_norm": 0.47983092069625854,
+      "learning_rate": 3.865882327247916e-06,
+      "loss": 0.1138,
+      "step": 993
+    },
+    {
+      "epoch": 8.835555555555555,
+      "grad_norm": 0.5129906535148621,
+      "learning_rate": 3.8060233744356633e-06,
+      "loss": 0.0959,
+      "step": 994
+    },
+    {
+      "epoch": 8.844444444444445,
+      "grad_norm": 0.4921208322048187,
+      "learning_rate": 3.7466131294862595e-06,
+      "loss": 0.1029,
+      "step": 995
+    },
+    {
+      "epoch": 8.853333333333333,
+      "grad_norm": 0.4332411289215088,
+      "learning_rate": 3.687652169484568e-06,
+      "loss": 0.1143,
+      "step": 996
+    },
+    {
+      "epoch": 8.862222222222222,
+      "grad_norm": 0.4102513790130615,
+      "learning_rate": 3.6291410671512594e-06,
+      "loss": 0.0641,
+      "step": 997
+    },
+    {
+      "epoch": 8.87111111111111,
+      "grad_norm": 0.5347101092338562,
+      "learning_rate": 3.5710803908373224e-06,
+      "loss": 0.102,
+      "step": 998
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 0.5151996612548828,
+      "learning_rate": 3.5134707045184955e-06,
+      "loss": 0.0896,
+      "step": 999
+    },
+    {
+      "epoch": 8.88888888888889,
+      "grad_norm": 0.5282700061798096,
+      "learning_rate": 3.4563125677897932e-06,
+      "loss": 0.0743,
+      "step": 1000
+    },
+    {
+      "epoch": 8.88888888888889,
+      "eval_loss": 0.4198705554008484,
+      "eval_runtime": 44.3807,
+      "eval_samples_per_second": 2.253,
+      "eval_steps_per_second": 2.253,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1120,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.141377228159058e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}