diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.172839506172839,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0030864197530864196,
+      "grad_norm": 184.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 27.5986,
+      "step": 1
+    },
+    {
+      "epoch": 0.006172839506172839,
+      "grad_norm": 182.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 27.4133,
+      "step": 2
+    },
+    {
+      "epoch": 0.009259259259259259,
+      "grad_norm": 188.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 28.0722,
+      "step": 3
+    },
+    {
+      "epoch": 0.012345679012345678,
+      "grad_norm": 185.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 28.1167,
+      "step": 4
+    },
+    {
+      "epoch": 0.015432098765432098,
+      "grad_norm": 183.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 27.3655,
+      "step": 5
+    },
+    {
+      "epoch": 0.018518518518518517,
+      "grad_norm": 185.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 27.6192,
+      "step": 6
+    },
+    {
+      "epoch": 0.021604938271604937,
+      "grad_norm": 181.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 27.1146,
+      "step": 7
+    },
+    {
+      "epoch": 0.024691358024691357,
+      "grad_norm": 183.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 27.2559,
+      "step": 8
+    },
+    {
+      "epoch": 0.027777777777777776,
+      "grad_norm": 183.0,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 27.3459,
+      "step": 9
+    },
+    {
+      "epoch": 0.030864197530864196,
+      "grad_norm": 184.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 27.4518,
+      "step": 10
+    },
+    {
+      "epoch": 0.033950617283950615,
+      "grad_norm": 184.0,
+      "learning_rate": 2.2e-06,
+      "loss": 27.3792,
+      "step": 11
+    },
+    {
+      "epoch": 0.037037037037037035,
+      "grad_norm": 184.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 26.8609,
+      "step": 12
+    },
+    {
+      "epoch": 0.040123456790123455,
+      "grad_norm": 187.0,
+      "learning_rate": 2.6e-06,
+      "loss": 26.5152,
+      "step": 13
+    },
+    {
+      "epoch": 0.043209876543209874,
+      "grad_norm": 184.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 26.2923,
+      "step": 14
+    },
+    {
+      "epoch": 0.046296296296296294,
+      "grad_norm": 184.0,
+      "learning_rate": 3e-06,
+      "loss": 25.9474,
+      "step": 15
+    },
+    {
+      "epoch": 0.04938271604938271,
+      "grad_norm": 187.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 25.7775,
+      "step": 16
+    },
+    {
+      "epoch": 0.05246913580246913,
+      "grad_norm": 181.0,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 25.2894,
+      "step": 17
+    },
+    {
+      "epoch": 0.05555555555555555,
+      "grad_norm": 186.0,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 25.2786,
+      "step": 18
+    },
+    {
+      "epoch": 0.05864197530864197,
+      "grad_norm": 184.0,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 24.8447,
+      "step": 19
+    },
+    {
+      "epoch": 0.06172839506172839,
+      "grad_norm": 183.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 24.0526,
+      "step": 20
+    },
+    {
+      "epoch": 0.06481481481481481,
+      "grad_norm": 183.0,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 23.7467,
+      "step": 21
+    },
+    {
+      "epoch": 0.06790123456790123,
+      "grad_norm": 192.0,
+      "learning_rate": 4.4e-06,
+      "loss": 22.9038,
+      "step": 22
+    },
+    {
+      "epoch": 0.07098765432098765,
+      "grad_norm": 182.0,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 21.792,
+      "step": 23
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 176.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 20.9423,
+      "step": 24
+    },
+    {
+      "epoch": 0.07716049382716049,
+      "grad_norm": 174.0,
+      "learning_rate": 5e-06,
+      "loss": 20.1287,
+      "step": 25
+    },
+    {
+      "epoch": 0.08024691358024691,
+      "grad_norm": 173.0,
+      "learning_rate": 5.2e-06,
+      "loss": 19.0291,
+      "step": 26
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 172.0,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 18.1876,
+      "step": 27
+    },
+    {
+      "epoch": 0.08641975308641975,
+      "grad_norm": 173.0,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 17.2964,
+      "step": 28
+    },
+    {
+      "epoch": 0.08950617283950617,
+      "grad_norm": 176.0,
+      "learning_rate": 5.8e-06,
+      "loss": 16.4037,
+      "step": 29
+    },
+    {
+      "epoch": 0.09259259259259259,
+      "grad_norm": 169.0,
+      "learning_rate": 6e-06,
+      "loss": 15.3495,
+      "step": 30
+    },
+    {
+      "epoch": 0.09567901234567901,
+      "grad_norm": 164.0,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 14.3221,
+      "step": 31
+    },
+    {
+      "epoch": 0.09876543209876543,
+      "grad_norm": 162.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 13.1121,
+      "step": 32
+    },
+    {
+      "epoch": 0.10185185185185185,
+      "grad_norm": 142.0,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 11.7731,
+      "step": 33
+    },
+    {
+      "epoch": 0.10493827160493827,
+      "grad_norm": 120.0,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 10.4971,
+      "step": 34
+    },
+    {
+      "epoch": 0.10802469135802469,
+      "grad_norm": 86.5,
+      "learning_rate": 7e-06,
+      "loss": 9.3055,
+      "step": 35
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 72.0,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 8.6972,
+      "step": 36
+    },
+    {
+      "epoch": 0.11419753086419752,
+      "grad_norm": 64.0,
+      "learning_rate": 7.4e-06,
+      "loss": 8.3069,
+      "step": 37
+    },
+    {
+      "epoch": 0.11728395061728394,
+      "grad_norm": 56.25,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 8.0132,
+      "step": 38
+    },
+    {
+      "epoch": 0.12037037037037036,
+      "grad_norm": 53.0,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 7.6036,
+      "step": 39
+    },
+    {
+      "epoch": 0.12345679012345678,
+      "grad_norm": 42.25,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 7.2739,
+      "step": 40
+    },
+    {
+      "epoch": 0.12654320987654322,
+      "grad_norm": 40.0,
+      "learning_rate": 8.2e-06,
+      "loss": 7.1157,
+      "step": 41
+    },
+    {
+      "epoch": 0.12962962962962962,
+      "grad_norm": 38.5,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 6.9532,
+      "step": 42
+    },
+    {
+      "epoch": 0.13271604938271606,
+      "grad_norm": 35.5,
+      "learning_rate": 8.6e-06,
+      "loss": 6.6702,
+      "step": 43
+    },
+    {
+      "epoch": 0.13580246913580246,
+      "grad_norm": 28.25,
+      "learning_rate": 8.8e-06,
+      "loss": 6.5143,
+      "step": 44
+    },
+    {
+      "epoch": 0.1388888888888889,
+      "grad_norm": 34.25,
+      "learning_rate": 9e-06,
+      "loss": 6.378,
+      "step": 45
+    },
+    {
+      "epoch": 0.1419753086419753,
+      "grad_norm": 26.375,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 6.2246,
+      "step": 46
+    },
+    {
+      "epoch": 0.14506172839506173,
+      "grad_norm": 26.75,
+      "learning_rate": 9.4e-06,
+      "loss": 6.06,
+      "step": 47
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 30.25,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 5.8962,
+      "step": 48
+    },
+    {
+      "epoch": 0.15123456790123457,
+      "grad_norm": 22.875,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 5.7839,
+      "step": 49
+    },
+    {
+      "epoch": 0.15432098765432098,
+      "grad_norm": 25.375,
+      "learning_rate": 1e-05,
+      "loss": 5.5753,
+      "step": 50
+    },
+    {
+      "epoch": 0.1574074074074074,
+      "grad_norm": 21.125,
+      "learning_rate": 1.02e-05,
+      "loss": 5.5293,
+      "step": 51
+    },
+    {
+      "epoch": 0.16049382716049382,
+      "grad_norm": 21.375,
+      "learning_rate": 1.04e-05,
+      "loss": 5.3325,
+      "step": 52
+    },
+    {
+      "epoch": 0.16358024691358025,
+      "grad_norm": 21.25,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 5.4129,
+      "step": 53
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 20.0,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 5.1563,
+      "step": 54
+    },
+    {
+      "epoch": 0.1697530864197531,
+      "grad_norm": 24.125,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 5.1243,
+      "step": 55
+    },
+    {
+      "epoch": 0.1728395061728395,
+      "grad_norm": 26.375,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 4.9199,
+      "step": 56
+    },
+    {
+      "epoch": 0.17592592592592593,
+      "grad_norm": 25.125,
+      "learning_rate": 1.14e-05,
+      "loss": 4.8811,
+      "step": 57
+    },
+    {
+      "epoch": 0.17901234567901234,
+      "grad_norm": 40.0,
+      "learning_rate": 1.16e-05,
+      "loss": 4.7498,
+      "step": 58
+    },
+    {
+      "epoch": 0.18209876543209877,
+      "grad_norm": 21.875,
+      "learning_rate": 1.18e-05,
+      "loss": 4.6488,
+      "step": 59
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 23.0,
+      "learning_rate": 1.2e-05,
+      "loss": 4.4789,
+      "step": 60
+    },
+    {
+      "epoch": 0.1882716049382716,
+      "grad_norm": 20.875,
+      "learning_rate": 1.22e-05,
+      "loss": 4.3335,
+      "step": 61
+    },
+    {
+      "epoch": 0.19135802469135801,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 4.0636,
+      "step": 62
+    },
+    {
+      "epoch": 0.19444444444444445,
+      "grad_norm": 20.875,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 3.8986,
+      "step": 63
+    },
+    {
+      "epoch": 0.19753086419753085,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 3.8057,
+      "step": 64
+    },
+    {
+      "epoch": 0.2006172839506173,
+      "grad_norm": 19.375,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 3.6571,
+      "step": 65
+    },
+    {
+      "epoch": 0.2037037037037037,
+      "grad_norm": 25.625,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 3.5601,
+      "step": 66
+    },
+    {
+      "epoch": 0.20679012345679013,
+      "grad_norm": 32.5,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 3.47,
+      "step": 67
+    },
+    {
+      "epoch": 0.20987654320987653,
+      "grad_norm": 42.75,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 3.3042,
+      "step": 68
+    },
+    {
+      "epoch": 0.21296296296296297,
+      "grad_norm": 22.875,
+      "learning_rate": 1.38e-05,
+      "loss": 3.1864,
+      "step": 69
+    },
+    {
+      "epoch": 0.21604938271604937,
+      "grad_norm": 20.625,
+      "learning_rate": 1.4e-05,
+      "loss": 3.1025,
+      "step": 70
+    },
+    {
+      "epoch": 0.2191358024691358,
+      "grad_norm": 20.875,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 3.0647,
+      "step": 71
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 2.8714,
+      "step": 72
+    },
+    {
+      "epoch": 0.22530864197530864,
+      "grad_norm": 18.0,
+      "learning_rate": 1.46e-05,
+      "loss": 2.8563,
+      "step": 73
+    },
+    {
+      "epoch": 0.22839506172839505,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.48e-05,
+      "loss": 2.7964,
+      "step": 74
+    },
+    {
+      "epoch": 0.23148148148148148,
+      "grad_norm": 12.625,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 2.8715,
+      "step": 75
+    },
+    {
+      "epoch": 0.2345679012345679,
+      "grad_norm": 16.5,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 2.7654,
+      "step": 76
+    },
+    {
+      "epoch": 0.23765432098765432,
+      "grad_norm": 17.75,
+      "learning_rate": 1.54e-05,
+      "loss": 2.7336,
+      "step": 77
+    },
+    {
+      "epoch": 0.24074074074074073,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 2.6593,
+      "step": 78
+    },
+    {
+      "epoch": 0.24382716049382716,
+      "grad_norm": 11.75,
+      "learning_rate": 1.58e-05,
+      "loss": 2.4657,
+      "step": 79
+    },
+    {
+      "epoch": 0.24691358024691357,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.4154,
+      "step": 80
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 11.875,
+      "learning_rate": 1.62e-05,
+      "loss": 2.4599,
+      "step": 81
+    },
+    {
+      "epoch": 0.25308641975308643,
+      "grad_norm": 18.0,
+      "learning_rate": 1.64e-05,
+      "loss": 2.3133,
+      "step": 82
+    },
+    {
+      "epoch": 0.25617283950617287,
+      "grad_norm": 10.5,
+      "learning_rate": 1.66e-05,
+      "loss": 2.2728,
+      "step": 83
+    },
+    {
+      "epoch": 0.25925925925925924,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 2.2739,
+      "step": 84
+    },
+    {
+      "epoch": 0.2623456790123457,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7e-05,
+      "loss": 2.2353,
+      "step": 85
+    },
+    {
+      "epoch": 0.2654320987654321,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.72e-05,
+      "loss": 2.1565,
+      "step": 86
+    },
+    {
+      "epoch": 0.26851851851851855,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 2.1032,
+      "step": 87
+    },
+    {
+      "epoch": 0.2716049382716049,
+      "grad_norm": 19.25,
+      "learning_rate": 1.76e-05,
+      "loss": 2.1014,
+      "step": 88
+    },
+    {
+      "epoch": 0.27469135802469136,
+      "grad_norm": 13.25,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 2.0393,
+      "step": 89
+    },
+    {
+      "epoch": 0.2777777777777778,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8e-05,
+      "loss": 2.019,
+      "step": 90
+    },
+    {
+      "epoch": 0.2808641975308642,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 1.9641,
+      "step": 91
+    },
+    {
+      "epoch": 0.2839506172839506,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.9475,
+      "step": 92
+    },
+    {
+      "epoch": 0.28703703703703703,
+      "grad_norm": 21.125,
+      "learning_rate": 1.86e-05,
+      "loss": 1.9671,
+      "step": 93
+    },
+    {
+      "epoch": 0.29012345679012347,
+      "grad_norm": 19.875,
+      "learning_rate": 1.88e-05,
+      "loss": 1.9702,
+      "step": 94
+    },
+    {
+      "epoch": 0.2932098765432099,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.9e-05,
+      "loss": 1.8942,
+      "step": 95
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.7709,
+      "step": 96
+    },
+    {
+      "epoch": 0.2993827160493827,
+      "grad_norm": 11.875,
+      "learning_rate": 1.94e-05,
+      "loss": 1.7907,
+      "step": 97
+    },
+    {
+      "epoch": 0.30246913580246915,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.8636,
+      "step": 98
+    },
+    {
+      "epoch": 0.3055555555555556,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.98e-05,
+      "loss": 1.7745,
+      "step": 99
+    },
+    {
+      "epoch": 0.30864197530864196,
+      "grad_norm": 11.5625,
+      "learning_rate": 2e-05,
+      "loss": 1.7629,
+      "step": 100
+    },
+    {
+      "epoch": 0.3117283950617284,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9989473684210526e-05,
+      "loss": 1.7629,
+      "step": 101
+    },
+    {
+      "epoch": 0.3148148148148148,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.9978947368421054e-05,
+      "loss": 1.7434,
+      "step": 102
+    },
+    {
+      "epoch": 0.31790123456790126,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9968421052631582e-05,
+      "loss": 1.7646,
+      "step": 103
+    },
+    {
+      "epoch": 0.32098765432098764,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9957894736842107e-05,
+      "loss": 1.6937,
+      "step": 104
+    },
+    {
+      "epoch": 0.32407407407407407,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.994736842105263e-05,
+      "loss": 1.5772,
+      "step": 105
+    },
+    {
+      "epoch": 0.3271604938271605,
+      "grad_norm": 20.875,
+      "learning_rate": 1.993684210526316e-05,
+      "loss": 1.6787,
+      "step": 106
+    },
+    {
+      "epoch": 0.33024691358024694,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9926315789473688e-05,
+      "loss": 1.6397,
+      "step": 107
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9915789473684212e-05,
+      "loss": 1.6089,
+      "step": 108
+    },
+    {
+      "epoch": 0.33641975308641975,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.990526315789474e-05,
+      "loss": 1.6292,
+      "step": 109
+    },
+    {
+      "epoch": 0.3395061728395062,
+      "grad_norm": 11.25,
+      "learning_rate": 1.9894736842105265e-05,
+      "loss": 1.5789,
+      "step": 110
+    },
+    {
+      "epoch": 0.3425925925925926,
+      "grad_norm": 12.625,
+      "learning_rate": 1.988421052631579e-05,
+      "loss": 1.5793,
+      "step": 111
+    },
+    {
+      "epoch": 0.345679012345679,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.9873684210526318e-05,
+      "loss": 1.5839,
+      "step": 112
+    },
+    {
+      "epoch": 0.3487654320987654,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9863157894736846e-05,
+      "loss": 1.5303,
+      "step": 113
+    },
+    {
+      "epoch": 0.35185185185185186,
+      "grad_norm": 10.625,
+      "learning_rate": 1.985263157894737e-05,
+      "loss": 1.5435,
+      "step": 114
+    },
+    {
+      "epoch": 0.3549382716049383,
+      "grad_norm": 10.25,
+      "learning_rate": 1.9842105263157895e-05,
+      "loss": 1.4994,
+      "step": 115
+    },
+    {
+      "epoch": 0.35802469135802467,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9831578947368423e-05,
+      "loss": 1.6159,
+      "step": 116
+    },
+    {
+      "epoch": 0.3611111111111111,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.982105263157895e-05,
+      "loss": 1.5704,
+      "step": 117
+    },
+    {
+      "epoch": 0.36419753086419754,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.9810526315789476e-05,
+      "loss": 1.4551,
+      "step": 118
+    },
+    {
+      "epoch": 0.36728395061728397,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.98e-05,
+      "loss": 1.5297,
+      "step": 119
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 20.0,
+      "learning_rate": 1.9789473684210528e-05,
+      "loss": 1.5677,
+      "step": 120
+    },
+    {
+      "epoch": 0.3734567901234568,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9778947368421056e-05,
+      "loss": 1.5023,
+      "step": 121
+    },
+    {
+      "epoch": 0.3765432098765432,
+      "grad_norm": 9.25,
+      "learning_rate": 1.976842105263158e-05,
+      "loss": 1.4621,
+      "step": 122
+    },
+    {
+      "epoch": 0.37962962962962965,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9757894736842105e-05,
+      "loss": 1.3987,
+      "step": 123
+    },
+    {
+      "epoch": 0.38271604938271603,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.9747368421052633e-05,
+      "loss": 1.4243,
+      "step": 124
+    },
+    {
+      "epoch": 0.38580246913580246,
+      "grad_norm": 9.625,
+      "learning_rate": 1.9736842105263158e-05,
+      "loss": 1.4831,
+      "step": 125
+    },
+    {
+      "epoch": 0.3888888888888889,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.9726315789473686e-05,
+      "loss": 1.4254,
+      "step": 126
+    },
+    {
+      "epoch": 0.39197530864197533,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.9715789473684214e-05,
+      "loss": 1.3682,
+      "step": 127
+    },
+    {
+      "epoch": 0.3950617283950617,
+      "grad_norm": 10.0,
+      "learning_rate": 1.970526315789474e-05,
+      "loss": 1.4598,
+      "step": 128
+    },
+    {
+      "epoch": 0.39814814814814814,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9694736842105263e-05,
+      "loss": 1.4724,
+      "step": 129
+    },
+    {
+      "epoch": 0.4012345679012346,
+      "grad_norm": 8.375,
+      "learning_rate": 1.968421052631579e-05,
+      "loss": 1.3694,
+      "step": 130
+    },
+    {
+      "epoch": 0.404320987654321,
+      "grad_norm": 16.0,
+      "learning_rate": 1.967368421052632e-05,
+      "loss": 1.355,
+      "step": 131
+    },
+    {
+      "epoch": 0.4074074074074074,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.9663157894736844e-05,
+      "loss": 1.4245,
+      "step": 132
+    },
+    {
+      "epoch": 0.4104938271604938,
+      "grad_norm": 12.75,
+      "learning_rate": 1.965263157894737e-05,
+      "loss": 1.4356,
+      "step": 133
+    },
+    {
+      "epoch": 0.41358024691358025,
+      "grad_norm": 10.125,
+      "learning_rate": 1.9642105263157897e-05,
+      "loss": 1.4018,
+      "step": 134
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9631578947368425e-05,
+      "loss": 1.3364,
+      "step": 135
+    },
+    {
+      "epoch": 0.41975308641975306,
+      "grad_norm": 8.25,
+      "learning_rate": 1.962105263157895e-05,
+      "loss": 1.4035,
+      "step": 136
+    },
+    {
+      "epoch": 0.4228395061728395,
+      "grad_norm": 10.0,
+      "learning_rate": 1.9610526315789474e-05,
+      "loss": 1.3433,
+      "step": 137
+    },
+    {
+      "epoch": 0.42592592592592593,
+      "grad_norm": 9.25,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.4123,
+      "step": 138
+    },
+    {
+      "epoch": 0.42901234567901236,
+      "grad_norm": 9.125,
+      "learning_rate": 1.9589473684210527e-05,
+      "loss": 1.4396,
+      "step": 139
+    },
+    {
+      "epoch": 0.43209876543209874,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.9578947368421055e-05,
+      "loss": 1.3973,
+      "step": 140
+    },
+    {
+      "epoch": 0.4351851851851852,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.956842105263158e-05,
+      "loss": 1.3291,
+      "step": 141
+    },
+    {
+      "epoch": 0.4382716049382716,
+      "grad_norm": 8.875,
+      "learning_rate": 1.9557894736842107e-05,
+      "loss": 1.3654,
+      "step": 142
+    },
+    {
+      "epoch": 0.44135802469135804,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9547368421052632e-05,
+      "loss": 1.3382,
+      "step": 143
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 18.125,
+      "learning_rate": 1.953684210526316e-05,
+      "loss": 1.3081,
+      "step": 144
+    },
+    {
+      "epoch": 0.44753086419753085,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9526315789473688e-05,
+      "loss": 1.3502,
+      "step": 145
+    },
+    {
+      "epoch": 0.4506172839506173,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9515789473684213e-05,
+      "loss": 1.2978,
+      "step": 146
+    },
+    {
+      "epoch": 0.4537037037037037,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9505263157894737e-05,
+      "loss": 1.3586,
+      "step": 147
+    },
+    {
+      "epoch": 0.4567901234567901,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9494736842105265e-05,
+      "loss": 1.2815,
+      "step": 148
+    },
+    {
+      "epoch": 0.45987654320987653,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.9484210526315793e-05,
+      "loss": 1.3506,
+      "step": 149
+    },
+    {
+      "epoch": 0.46296296296296297,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.9473684210526318e-05,
+      "loss": 1.3059,
+      "step": 150
+    },
+    {
+      "epoch": 0.4660493827160494,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.9463157894736843e-05,
+      "loss": 1.2972,
+      "step": 151
+    },
+    {
+      "epoch": 0.4691358024691358,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.945263157894737e-05,
+      "loss": 1.2842,
+      "step": 152
+    },
+    {
+      "epoch": 0.4722222222222222,
+      "grad_norm": 7.625,
+      "learning_rate": 1.9442105263157895e-05,
+      "loss": 1.2824,
+      "step": 153
+    },
+    {
+      "epoch": 0.47530864197530864,
+      "grad_norm": 9.375,
+      "learning_rate": 1.9431578947368423e-05,
+      "loss": 1.2562,
+      "step": 154
+    },
+    {
+      "epoch": 0.4783950617283951,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.9421052631578948e-05,
+      "loss": 1.2718,
+      "step": 155
+    },
+    {
+      "epoch": 0.48148148148148145,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.9410526315789476e-05,
+      "loss": 1.2873,
+      "step": 156
+    },
+    {
+      "epoch": 0.4845679012345679,
+      "grad_norm": 11.0,
+      "learning_rate": 1.94e-05,
+      "loss": 1.1872,
+      "step": 157
+    },
+    {
+      "epoch": 0.4876543209876543,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.9389473684210525e-05,
+      "loss": 1.2303,
+      "step": 158
+    },
+    {
+      "epoch": 0.49074074074074076,
+      "grad_norm": 8.25,
+      "learning_rate": 1.9378947368421053e-05,
+      "loss": 1.302,
+      "step": 159
+    },
+    {
+      "epoch": 0.49382716049382713,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.936842105263158e-05,
+      "loss": 1.2989,
+      "step": 160
+    },
+    {
+      "epoch": 0.49691358024691357,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.9357894736842106e-05,
+      "loss": 1.2285,
+      "step": 161
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.9347368421052634e-05,
+      "loss": 1.2159,
+      "step": 162
+    },
+    {
+      "epoch": 0.5030864197530864,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.9336842105263162e-05,
+      "loss": 1.2723,
+      "step": 163
+    },
+    {
+      "epoch": 0.5061728395061729,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9326315789473687e-05,
+      "loss": 1.2254,
+      "step": 164
+    },
+    {
+      "epoch": 0.5092592592592593,
+      "grad_norm": 10.5,
+      "learning_rate": 1.931578947368421e-05,
+      "loss": 1.2352,
+      "step": 165
+    },
+    {
+      "epoch": 0.5123456790123457,
+      "grad_norm": 14.0,
+      "learning_rate": 1.930526315789474e-05,
+      "loss": 1.3237,
+      "step": 166
+    },
+    {
+      "epoch": 0.5154320987654321,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.9294736842105264e-05,
+      "loss": 1.1514,
+      "step": 167
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9284210526315792e-05,
+      "loss": 1.278,
+      "step": 168
+    },
+    {
+      "epoch": 0.5216049382716049,
+      "grad_norm": 9.5,
+      "learning_rate": 1.9273684210526317e-05,
+      "loss": 1.3033,
+      "step": 169
+    },
+    {
+      "epoch": 0.5246913580246914,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9263157894736845e-05,
+      "loss": 1.2591,
+      "step": 170
+    },
+    {
+      "epoch": 0.5277777777777778,
+      "grad_norm": 9.0,
+      "learning_rate": 1.925263157894737e-05,
+      "loss": 1.1424,
+      "step": 171
+    },
+    {
+      "epoch": 0.5308641975308642,
+      "grad_norm": 21.75,
+      "learning_rate": 1.9242105263157894e-05,
+      "loss": 1.2627,
+      "step": 172
+    },
+    {
+      "epoch": 0.5339506172839507,
+      "grad_norm": 10.625,
+      "learning_rate": 1.9231578947368422e-05,
+      "loss": 1.2372,
+      "step": 173
+    },
+    {
+      "epoch": 0.5370370370370371,
+      "grad_norm": 9.375,
+      "learning_rate": 1.922105263157895e-05,
+      "loss": 1.2553,
+      "step": 174
+    },
+    {
+      "epoch": 0.5401234567901234,
+      "grad_norm": 52.75,
+      "learning_rate": 1.9210526315789474e-05,
+      "loss": 1.2207,
+      "step": 175
+    },
+    {
+      "epoch": 0.5432098765432098,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.2446,
+      "step": 176
+    },
+    {
+      "epoch": 0.5462962962962963,
+      "grad_norm": 10.25,
+      "learning_rate": 1.918947368421053e-05,
+      "loss": 1.188,
+      "step": 177
+    },
+    {
+      "epoch": 0.5493827160493827,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.9178947368421055e-05,
+      "loss": 1.1732,
+      "step": 178
+    },
+    {
+      "epoch": 0.5524691358024691,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.916842105263158e-05,
+      "loss": 1.2389,
+      "step": 179
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 10.125,
+      "learning_rate": 1.9157894736842108e-05,
+      "loss": 1.246,
+      "step": 180
+    },
+    {
+      "epoch": 0.558641975308642,
+      "grad_norm": 10.125,
+      "learning_rate": 1.9147368421052632e-05,
+      "loss": 1.2276,
+      "step": 181
+    },
+    {
+      "epoch": 0.5617283950617284,
+      "grad_norm": 9.375,
+      "learning_rate": 1.913684210526316e-05,
+      "loss": 1.2173,
+      "step": 182
+    },
+    {
+      "epoch": 0.5648148148148148,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.9126315789473685e-05,
+      "loss": 1.1552,
+      "step": 183
+    },
+    {
+      "epoch": 0.5679012345679012,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.9115789473684213e-05,
+      "loss": 1.1582,
+      "step": 184
+    },
+    {
+      "epoch": 0.5709876543209876,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.9105263157894738e-05,
+      "loss": 1.1191,
+      "step": 185
+    },
+    {
+      "epoch": 0.5740740740740741,
+      "grad_norm": 12.0,
+      "learning_rate": 1.9094736842105262e-05,
+      "loss": 1.1694,
+      "step": 186
+    },
+    {
+      "epoch": 0.5771604938271605,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.908421052631579e-05,
+      "loss": 1.1225,
+      "step": 187
+    },
+    {
+      "epoch": 0.5802469135802469,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.907368421052632e-05,
+      "loss": 1.1049,
+      "step": 188
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.9063157894736843e-05,
+      "loss": 1.1485,
+      "step": 189
+    },
+    {
+      "epoch": 0.5864197530864198,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.9052631578947368e-05,
+      "loss": 1.1966,
+      "step": 190
+    },
+    {
+      "epoch": 0.5895061728395061,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.9042105263157896e-05,
+      "loss": 1.1766,
+      "step": 191
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9031578947368424e-05,
+      "loss": 1.1985,
+      "step": 192
+    },
+    {
+      "epoch": 0.595679012345679,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.902105263157895e-05,
+      "loss": 1.2548,
+      "step": 193
+    },
+    {
+      "epoch": 0.5987654320987654,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.9010526315789476e-05,
+      "loss": 1.177,
+      "step": 194
+    },
+    {
+      "epoch": 0.6018518518518519,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.9e-05,
+      "loss": 1.0891,
+      "step": 195
+    },
+    {
+      "epoch": 0.6049382716049383,
+      "grad_norm": 28.625,
+      "learning_rate": 1.898947368421053e-05,
+      "loss": 1.1796,
+      "step": 196
+    },
+    {
+      "epoch": 0.6080246913580247,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.8978947368421054e-05,
+      "loss": 1.2246,
+      "step": 197
+    },
+    {
+      "epoch": 0.6111111111111112,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.8968421052631582e-05,
+      "loss": 1.1831,
+      "step": 198
+    },
+    {
+      "epoch": 0.6141975308641975,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.8957894736842106e-05,
+      "loss": 1.1508,
+      "step": 199
+    },
+    {
+      "epoch": 0.6172839506172839,
+      "grad_norm": 8.125,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 1.1847,
+      "step": 200
+    },
+    {
+      "epoch": 0.6203703703703703,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.893684210526316e-05,
+      "loss": 1.1961,
+      "step": 201
+    },
+    {
+      "epoch": 0.6234567901234568,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.8926315789473687e-05,
+      "loss": 1.1355,
+      "step": 202
+    },
+    {
+      "epoch": 0.6265432098765432,
+      "grad_norm": 7.5,
+      "learning_rate": 1.891578947368421e-05,
+      "loss": 1.1927,
+      "step": 203
+    },
+    {
+      "epoch": 0.6296296296296297,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.8905263157894736e-05,
+      "loss": 1.1381,
+      "step": 204
+    },
+    {
+      "epoch": 0.6327160493827161,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.8894736842105264e-05,
+      "loss": 1.1865,
+      "step": 205
+    },
+    {
+      "epoch": 0.6358024691358025,
+      "grad_norm": 9.375,
+      "learning_rate": 1.8884210526315792e-05,
+      "loss": 1.2046,
+      "step": 206
+    },
+    {
+      "epoch": 0.6388888888888888,
+      "grad_norm": 8.375,
+      "learning_rate": 1.8873684210526317e-05,
+      "loss": 1.1855,
+      "step": 207
+    },
+    {
+      "epoch": 0.6419753086419753,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.886315789473684e-05,
+      "loss": 1.1474,
+      "step": 208
+    },
+    {
+      "epoch": 0.6450617283950617,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.885263157894737e-05,
+      "loss": 1.1346,
+      "step": 209
+    },
+    {
+      "epoch": 0.6481481481481481,
+      "grad_norm": 9.125,
+      "learning_rate": 1.8842105263157898e-05,
+      "loss": 1.1207,
+      "step": 210
+    },
+    {
+      "epoch": 0.6512345679012346,
+      "grad_norm": 10.25,
+      "learning_rate": 1.8831578947368422e-05,
+      "loss": 1.1536,
+      "step": 211
+    },
+    {
+      "epoch": 0.654320987654321,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.882105263157895e-05,
+      "loss": 1.1538,
+      "step": 212
+    },
+    {
+      "epoch": 0.6574074074074074,
+      "grad_norm": 11.0,
+      "learning_rate": 1.8810526315789475e-05,
+      "loss": 1.1464,
+      "step": 213
+    },
+    {
+      "epoch": 0.6604938271604939,
+      "grad_norm": 8.125,
+      "learning_rate": 1.88e-05,
+      "loss": 1.1042,
+      "step": 214
+    },
+    {
+      "epoch": 0.6635802469135802,
+      "grad_norm": 12.125,
+      "learning_rate": 1.8789473684210528e-05,
+      "loss": 1.105,
+      "step": 215
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8778947368421056e-05,
+      "loss": 1.1394,
+      "step": 216
+    },
+    {
+      "epoch": 0.6697530864197531,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.876842105263158e-05,
+      "loss": 1.0836,
+      "step": 217
+    },
+    {
+      "epoch": 0.6728395061728395,
+      "grad_norm": 8.75,
+      "learning_rate": 1.8757894736842105e-05,
+      "loss": 1.1169,
+      "step": 218
+    },
+    {
+      "epoch": 0.6759259259259259,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.8747368421052633e-05,
+      "loss": 1.0947,
+      "step": 219
+    },
+    {
+      "epoch": 0.6790123456790124,
+      "grad_norm": 9.25,
+      "learning_rate": 1.873684210526316e-05,
+      "loss": 1.126,
+      "step": 220
+    },
+    {
+      "epoch": 0.6820987654320988,
+      "grad_norm": 7.0,
+      "learning_rate": 1.8726315789473686e-05,
+      "loss": 1.0663,
+      "step": 221
+    },
+    {
+      "epoch": 0.6851851851851852,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.871578947368421e-05,
+      "loss": 1.1493,
+      "step": 222
+    },
+    {
+      "epoch": 0.6882716049382716,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8705263157894738e-05,
+      "loss": 1.0604,
+      "step": 223
+    },
+    {
+      "epoch": 0.691358024691358,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.8694736842105266e-05,
+      "loss": 1.1673,
+      "step": 224
+    },
+    {
+      "epoch": 0.6944444444444444,
+      "grad_norm": 7.875,
+      "learning_rate": 1.868421052631579e-05,
+      "loss": 1.1192,
+      "step": 225
+    },
+    {
+      "epoch": 0.6975308641975309,
+      "grad_norm": 8.125,
+      "learning_rate": 1.8673684210526316e-05,
+      "loss": 1.155,
+      "step": 226
+    },
+    {
+      "epoch": 0.7006172839506173,
+      "grad_norm": 11.25,
+      "learning_rate": 1.8663157894736844e-05,
+      "loss": 1.1356,
+      "step": 227
+    },
+    {
+      "epoch": 0.7037037037037037,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.8652631578947368e-05,
+      "loss": 1.0957,
+      "step": 228
+    },
+    {
+      "epoch": 0.7067901234567902,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.8642105263157896e-05,
+      "loss": 1.0404,
+      "step": 229
+    },
+    {
+      "epoch": 0.7098765432098766,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.8631578947368424e-05,
+      "loss": 1.1099,
+      "step": 230
+    },
+    {
+      "epoch": 0.7129629629629629,
+      "grad_norm": 8.5,
+      "learning_rate": 1.862105263157895e-05,
+      "loss": 1.0641,
+      "step": 231
+    },
+    {
+      "epoch": 0.7160493827160493,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.8610526315789473e-05,
+      "loss": 1.0798,
+      "step": 232
+    },
+    {
+      "epoch": 0.7191358024691358,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.86e-05,
+      "loss": 1.014,
+      "step": 233
+    },
+    {
+      "epoch": 0.7222222222222222,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.858947368421053e-05,
+      "loss": 1.1003,
+      "step": 234
+    },
+    {
+      "epoch": 0.7253086419753086,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.8578947368421054e-05,
+      "loss": 1.0618,
+      "step": 235
+    },
+    {
+      "epoch": 0.7283950617283951,
+      "grad_norm": 8.5,
+      "learning_rate": 1.856842105263158e-05,
+      "loss": 1.062,
+      "step": 236
+    },
+    {
+      "epoch": 0.7314814814814815,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.8557894736842107e-05,
+      "loss": 1.1445,
+      "step": 237
+    },
+    {
+      "epoch": 0.7345679012345679,
+      "grad_norm": 8.125,
+      "learning_rate": 1.8547368421052635e-05,
+      "loss": 1.0864,
+      "step": 238
+    },
+    {
+      "epoch": 0.7376543209876543,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.853684210526316e-05,
+      "loss": 1.0275,
+      "step": 239
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 1.0576,
+      "step": 240
+    },
+    {
+      "epoch": 0.7438271604938271,
+      "grad_norm": 8.0,
+      "learning_rate": 1.8515789473684212e-05,
+      "loss": 1.0991,
+      "step": 241
+    },
+    {
+      "epoch": 0.7469135802469136,
+      "grad_norm": 8.375,
+      "learning_rate": 1.8505263157894737e-05,
+      "loss": 1.0157,
+      "step": 242
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.8494736842105265e-05,
+      "loss": 1.0565,
+      "step": 243
+    },
+    {
+      "epoch": 0.7530864197530864,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.8484210526315793e-05,
+      "loss": 1.084,
+      "step": 244
+    },
+    {
+      "epoch": 0.7561728395061729,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.8473684210526317e-05,
+      "loss": 1.0797,
+      "step": 245
+    },
+    {
+      "epoch": 0.7592592592592593,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.8463157894736842e-05,
+      "loss": 1.1177,
+      "step": 246
+    },
+    {
+      "epoch": 0.7623456790123457,
+      "grad_norm": 6.53125,
+      "learning_rate": 1.845263157894737e-05,
+      "loss": 1.0993,
+      "step": 247
+    },
+    {
+      "epoch": 0.7654320987654321,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.8442105263157898e-05,
+      "loss": 1.1864,
+      "step": 248
+    },
+    {
+      "epoch": 0.7685185185185185,
+      "grad_norm": 8.0,
+      "learning_rate": 1.8431578947368423e-05,
+      "loss": 1.0729,
+      "step": 249
+    },
+    {
+      "epoch": 0.7716049382716049,
+      "grad_norm": 11.0,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 1.1264,
+      "step": 250
+    },
+    {
+      "epoch": 0.7746913580246914,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.8410526315789475e-05,
+      "loss": 1.0553,
+      "step": 251
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.095,
+      "step": 252
+    },
+    {
+      "epoch": 0.7808641975308642,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.8389473684210528e-05,
+      "loss": 1.0542,
+      "step": 253
+    },
+    {
+      "epoch": 0.7839506172839507,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.8378947368421053e-05,
+      "loss": 1.0694,
+      "step": 254
+    },
+    {
+      "epoch": 0.7870370370370371,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.836842105263158e-05,
+      "loss": 0.9647,
+      "step": 255
+    },
+    {
+      "epoch": 0.7901234567901234,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8357894736842105e-05,
+      "loss": 1.0074,
+      "step": 256
+    },
+    {
+      "epoch": 0.7932098765432098,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.8347368421052633e-05,
+      "loss": 1.1112,
+      "step": 257
+    },
+    {
+      "epoch": 0.7962962962962963,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.8336842105263158e-05,
+      "loss": 1.0298,
+      "step": 258
+    },
+    {
+      "epoch": 0.7993827160493827,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.8326315789473686e-05,
+      "loss": 1.0465,
+      "step": 259
+    },
+    {
+      "epoch": 0.8024691358024691,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.831578947368421e-05,
+      "loss": 1.0651,
+      "step": 260
+    },
+    {
+      "epoch": 0.8055555555555556,
+      "grad_norm": 7.625,
+      "learning_rate": 1.830526315789474e-05,
+      "loss": 1.1134,
+      "step": 261
+    },
+    {
+      "epoch": 0.808641975308642,
+      "grad_norm": 7.5,
+      "learning_rate": 1.8294736842105267e-05,
+      "loss": 1.0439,
+      "step": 262
+    },
+    {
+      "epoch": 0.8117283950617284,
+      "grad_norm": 6.5,
+      "learning_rate": 1.828421052631579e-05,
+      "loss": 0.9585,
+      "step": 263
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.8273684210526316e-05,
+      "loss": 1.1146,
+      "step": 264
+    },
+    {
+      "epoch": 0.8179012345679012,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8263157894736844e-05,
+      "loss": 1.0394,
+      "step": 265
+    },
+    {
+      "epoch": 0.8209876543209876,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.8252631578947372e-05,
+      "loss": 1.0097,
+      "step": 266
+    },
+    {
+      "epoch": 0.8240740740740741,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.8242105263157897e-05,
+      "loss": 1.0522,
+      "step": 267
+    },
+    {
+      "epoch": 0.8271604938271605,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.823157894736842e-05,
+      "loss": 1.0389,
+      "step": 268
+    },
+    {
+      "epoch": 0.8302469135802469,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.822105263157895e-05,
+      "loss": 1.0292,
+      "step": 269
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.8210526315789477e-05,
+      "loss": 1.033,
+      "step": 270
+    },
+    {
+      "epoch": 0.8364197530864198,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 1.0497,
+      "step": 271
+    },
+    {
+      "epoch": 0.8395061728395061,
+      "grad_norm": 9.0,
+      "learning_rate": 1.8189473684210527e-05,
+      "loss": 1.0667,
+      "step": 272
+    },
+    {
+      "epoch": 0.8425925925925926,
+      "grad_norm": 37.0,
+      "learning_rate": 1.8178947368421055e-05,
+      "loss": 1.122,
+      "step": 273
+    },
+    {
+      "epoch": 0.845679012345679,
+      "grad_norm": 13.875,
+      "learning_rate": 1.816842105263158e-05,
+      "loss": 1.0092,
+      "step": 274
+    },
+    {
+      "epoch": 0.8487654320987654,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.8157894736842107e-05,
+      "loss": 1.0074,
+      "step": 275
+    },
+    {
+      "epoch": 0.8518518518518519,
+      "grad_norm": 10.0,
+      "learning_rate": 1.8147368421052632e-05,
+      "loss": 1.0229,
+      "step": 276
+    },
+    {
+      "epoch": 0.8549382716049383,
+      "grad_norm": 8.25,
+      "learning_rate": 1.813684210526316e-05,
+      "loss": 0.9985,
+      "step": 277
+    },
+    {
+      "epoch": 0.8580246913580247,
+      "grad_norm": 9.875,
+      "learning_rate": 1.8126315789473685e-05,
+      "loss": 1.0034,
+      "step": 278
+    },
+    {
+      "epoch": 0.8611111111111112,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.8115789473684213e-05,
+      "loss": 1.0862,
+      "step": 279
+    },
+    {
+      "epoch": 0.8641975308641975,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.810526315789474e-05,
+      "loss": 1.0244,
+      "step": 280
+    },
+    {
+      "epoch": 0.8672839506172839,
+      "grad_norm": 7.25,
+      "learning_rate": 1.8094736842105265e-05,
+      "loss": 0.9906,
+      "step": 281
+    },
+    {
+      "epoch": 0.8703703703703703,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.808421052631579e-05,
+      "loss": 1.0507,
+      "step": 282
+    },
+    {
+      "epoch": 0.8734567901234568,
+      "grad_norm": 8.125,
+      "learning_rate": 1.8073684210526318e-05,
+      "loss": 1.0804,
+      "step": 283
+    },
+    {
+      "epoch": 0.8765432098765432,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.8063157894736846e-05,
+      "loss": 0.9389,
+      "step": 284
+    },
+    {
+      "epoch": 0.8796296296296297,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.805263157894737e-05,
+      "loss": 1.0067,
+      "step": 285
+    },
+    {
+      "epoch": 0.8827160493827161,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.8042105263157895e-05,
+      "loss": 1.0586,
+      "step": 286
+    },
+    {
+      "epoch": 0.8858024691358025,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.8031578947368423e-05,
+      "loss": 1.07,
+      "step": 287
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.8021052631578948e-05,
+      "loss": 1.0034,
+      "step": 288
+    },
+    {
+      "epoch": 0.8919753086419753,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.8010526315789476e-05,
+      "loss": 1.0518,
+      "step": 289
+    },
+    {
+      "epoch": 0.8950617283950617,
+      "grad_norm": 9.5,
+      "learning_rate": 1.8e-05,
+      "loss": 0.991,
+      "step": 290
+    },
+    {
+      "epoch": 0.8981481481481481,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.798947368421053e-05,
+      "loss": 1.0426,
+      "step": 291
+    },
+    {
+      "epoch": 0.9012345679012346,
+      "grad_norm": 10.0,
+      "learning_rate": 1.7978947368421053e-05,
+      "loss": 1.002,
+      "step": 292
+    },
+    {
+      "epoch": 0.904320987654321,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.7968421052631578e-05,
+      "loss": 0.9581,
+      "step": 293
+    },
+    {
+      "epoch": 0.9074074074074074,
+      "grad_norm": 17.375,
+      "learning_rate": 1.795789473684211e-05,
+      "loss": 1.0678,
+      "step": 294
+    },
+    {
+      "epoch": 0.9104938271604939,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.7947368421052634e-05,
+      "loss": 1.0071,
+      "step": 295
+    },
+    {
+      "epoch": 0.9135802469135802,
+      "grad_norm": 6.375,
+      "learning_rate": 1.793684210526316e-05,
+      "loss": 0.976,
+      "step": 296
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7926315789473686e-05,
+      "loss": 1.0011,
+      "step": 297
+    },
+    {
+      "epoch": 0.9197530864197531,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.7915789473684214e-05,
+      "loss": 1.0273,
+      "step": 298
+    },
+    {
+      "epoch": 0.9228395061728395,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.790526315789474e-05,
+      "loss": 1.0466,
+      "step": 299
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 1.036,
+      "step": 300
+    },
+    {
+      "epoch": 0.9290123456790124,
+      "grad_norm": 10.0,
+      "learning_rate": 1.7884210526315792e-05,
+      "loss": 0.9688,
+      "step": 301
+    },
+    {
+      "epoch": 0.9320987654320988,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.7873684210526316e-05,
+      "loss": 1.008,
+      "step": 302
+    },
+    {
+      "epoch": 0.9351851851851852,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.7863157894736844e-05,
+      "loss": 1.0051,
+      "step": 303
+    },
+    {
+      "epoch": 0.9382716049382716,
+      "grad_norm": 9.625,
+      "learning_rate": 1.785263157894737e-05,
+      "loss": 1.0797,
+      "step": 304
+    },
+    {
+      "epoch": 0.941358024691358,
+      "grad_norm": 9.875,
+      "learning_rate": 1.7842105263157897e-05,
+      "loss": 1.0509,
+      "step": 305
+    },
+    {
+      "epoch": 0.9444444444444444,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.7831578947368422e-05,
+      "loss": 0.987,
+      "step": 306
+    },
+    {
+      "epoch": 0.9475308641975309,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.7821052631578946e-05,
+      "loss": 0.9543,
+      "step": 307
+    },
+    {
+      "epoch": 0.9506172839506173,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.7810526315789474e-05,
+      "loss": 1.0548,
+      "step": 308
+    },
+    {
+      "epoch": 0.9537037037037037,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 1.0103,
+      "step": 309
+    },
+    {
+      "epoch": 0.9567901234567902,
+      "grad_norm": 7.75,
+      "learning_rate": 1.7789473684210527e-05,
+      "loss": 1.0858,
+      "step": 310
+    },
+    {
+      "epoch": 0.9598765432098766,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.7778947368421055e-05,
+      "loss": 1.0234,
+      "step": 311
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.7768421052631583e-05,
+      "loss": 0.9663,
+      "step": 312
+    },
+    {
+      "epoch": 0.9660493827160493,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.7757894736842108e-05,
+      "loss": 1.0015,
+      "step": 313
+    },
+    {
+      "epoch": 0.9691358024691358,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.7747368421052632e-05,
+      "loss": 1.0274,
+      "step": 314
+    },
+    {
+      "epoch": 0.9722222222222222,
+      "grad_norm": 11.875,
+      "learning_rate": 1.773684210526316e-05,
+      "loss": 1.0738,
+      "step": 315
+    },
+    {
+      "epoch": 0.9753086419753086,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.7726315789473685e-05,
+      "loss": 1.0306,
+      "step": 316
+    },
+    {
+      "epoch": 0.9783950617283951,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.7715789473684213e-05,
+      "loss": 0.9824,
+      "step": 317
+    },
+    {
+      "epoch": 0.9814814814814815,
+      "grad_norm": 14.375,
+      "learning_rate": 1.7705263157894738e-05,
+      "loss": 0.9647,
+      "step": 318
+    },
+    {
+      "epoch": 0.9845679012345679,
+      "grad_norm": 7.75,
+      "learning_rate": 1.7694736842105266e-05,
+      "loss": 1.0058,
+      "step": 319
+    },
+    {
+      "epoch": 0.9876543209876543,
+      "grad_norm": 17.875,
+      "learning_rate": 1.768421052631579e-05,
+      "loss": 1.0261,
+      "step": 320
+    },
+    {
+      "epoch": 0.9907407407407407,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.7673684210526315e-05,
+      "loss": 0.9741,
+      "step": 321
+    },
+    {
+      "epoch": 0.9938271604938271,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.7663157894736843e-05,
+      "loss": 1.0248,
+      "step": 322
+    },
+    {
+      "epoch": 0.9969135802469136,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.765263157894737e-05,
+      "loss": 0.9356,
+      "step": 323
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.7642105263157896e-05,
+      "loss": 1.0272,
+      "step": 324
+    },
+    {
+      "epoch": 1.0030864197530864,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.763157894736842e-05,
+      "loss": 0.9209,
+      "step": 325
+    },
+    {
+      "epoch": 1.0061728395061729,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.7621052631578948e-05,
+      "loss": 0.9895,
+      "step": 326
+    },
+    {
+      "epoch": 1.0092592592592593,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.7610526315789476e-05,
+      "loss": 0.9197,
+      "step": 327
+    },
+    {
+      "epoch": 1.0123456790123457,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.76e-05,
+      "loss": 0.9328,
+      "step": 328
+    },
+    {
+      "epoch": 1.0154320987654322,
+      "grad_norm": 7.375,
+      "learning_rate": 1.758947368421053e-05,
+      "loss": 0.9537,
+      "step": 329
+    },
+    {
+      "epoch": 1.0185185185185186,
+      "grad_norm": 8.375,
+      "learning_rate": 1.7578947368421054e-05,
+      "loss": 0.9403,
+      "step": 330
+    },
+    {
+      "epoch": 1.021604938271605,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.756842105263158e-05,
+      "loss": 0.9236,
+      "step": 331
+    },
+    {
+      "epoch": 1.0246913580246915,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.7557894736842106e-05,
+      "loss": 1.0124,
+      "step": 332
+    },
+    {
+      "epoch": 1.0277777777777777,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.7547368421052634e-05,
+      "loss": 1.0201,
+      "step": 333
+    },
+    {
+      "epoch": 1.0308641975308641,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.753684210526316e-05,
+      "loss": 0.9601,
+      "step": 334
+    },
+    {
+      "epoch": 1.0339506172839505,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7526315789473683e-05,
+      "loss": 0.9651,
+      "step": 335
+    },
+    {
+      "epoch": 1.037037037037037,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.751578947368421e-05,
+      "loss": 0.9505,
+      "step": 336
+    },
+    {
+      "epoch": 1.0401234567901234,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.750526315789474e-05,
+      "loss": 0.9157,
+      "step": 337
+    },
+    {
+      "epoch": 1.0432098765432098,
+      "grad_norm": 9.125,
+      "learning_rate": 1.7494736842105264e-05,
+      "loss": 0.9798,
+      "step": 338
+    },
+    {
+      "epoch": 1.0462962962962963,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.748421052631579e-05,
+      "loss": 0.9974,
+      "step": 339
+    },
+    {
+      "epoch": 1.0493827160493827,
+      "grad_norm": 8.0,
+      "learning_rate": 1.7473684210526317e-05,
+      "loss": 0.9415,
+      "step": 340
+    },
+    {
+      "epoch": 1.0524691358024691,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.7463157894736845e-05,
+      "loss": 0.9578,
+      "step": 341
+    },
+    {
+      "epoch": 1.0555555555555556,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.745263157894737e-05,
+      "loss": 0.9542,
+      "step": 342
+    },
+    {
+      "epoch": 1.058641975308642,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.7442105263157894e-05,
+      "loss": 1.0192,
+      "step": 343
+    },
+    {
+      "epoch": 1.0617283950617284,
+      "grad_norm": 16.25,
+      "learning_rate": 1.7431578947368422e-05,
+      "loss": 0.9551,
+      "step": 344
+    },
+    {
+      "epoch": 1.0648148148148149,
+      "grad_norm": 8.375,
+      "learning_rate": 1.742105263157895e-05,
+      "loss": 0.9024,
+      "step": 345
+    },
+    {
+      "epoch": 1.0679012345679013,
+      "grad_norm": 24.875,
+      "learning_rate": 1.7410526315789475e-05,
+      "loss": 0.9191,
+      "step": 346
+    },
+    {
+      "epoch": 1.0709876543209877,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 1.0022,
+      "step": 347
+    },
+    {
+      "epoch": 1.074074074074074,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.7389473684210527e-05,
+      "loss": 0.9967,
+      "step": 348
+    },
+    {
+      "epoch": 1.0771604938271604,
+      "grad_norm": 29.0,
+      "learning_rate": 1.7378947368421052e-05,
+      "loss": 0.9846,
+      "step": 349
+    },
+    {
+      "epoch": 1.0802469135802468,
+      "grad_norm": 8.0,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 0.9769,
+      "step": 350
+    },
+    {
+      "epoch": 1.0833333333333333,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.7357894736842108e-05,
+      "loss": 0.9495,
+      "step": 351
+    },
+    {
+      "epoch": 1.0864197530864197,
+      "grad_norm": 8.5,
+      "learning_rate": 1.7347368421052633e-05,
+      "loss": 0.96,
+      "step": 352
+    },
+    {
+      "epoch": 1.0895061728395061,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.7336842105263157e-05,
+      "loss": 0.9319,
+      "step": 353
+    },
+    {
+      "epoch": 1.0925925925925926,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.7326315789473685e-05,
+      "loss": 0.9503,
+      "step": 354
+    },
+    {
+      "epoch": 1.095679012345679,
+      "grad_norm": 12.125,
+      "learning_rate": 1.7315789473684213e-05,
+      "loss": 0.964,
+      "step": 355
+    },
+    {
+      "epoch": 1.0987654320987654,
+      "grad_norm": 8.0,
+      "learning_rate": 1.7305263157894738e-05,
+      "loss": 0.9299,
+      "step": 356
+    },
+    {
+      "epoch": 1.1018518518518519,
+      "grad_norm": 10.0,
+      "learning_rate": 1.7294736842105263e-05,
+      "loss": 0.9948,
+      "step": 357
+    },
+    {
+      "epoch": 1.1049382716049383,
+      "grad_norm": 6.375,
+      "learning_rate": 1.728421052631579e-05,
+      "loss": 0.9422,
+      "step": 358
+    },
+    {
+      "epoch": 1.1080246913580247,
+      "grad_norm": 6.625,
+      "learning_rate": 1.727368421052632e-05,
+      "loss": 0.9123,
+      "step": 359
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.7263157894736843e-05,
+      "loss": 0.9404,
+      "step": 360
+    },
+    {
+      "epoch": 1.1141975308641976,
+      "grad_norm": 20.25,
+      "learning_rate": 1.725263157894737e-05,
+      "loss": 0.9059,
+      "step": 361
+    },
+    {
+      "epoch": 1.117283950617284,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.7242105263157896e-05,
+      "loss": 0.9893,
+      "step": 362
+    },
+    {
+      "epoch": 1.1203703703703705,
+      "grad_norm": 8.75,
+      "learning_rate": 1.723157894736842e-05,
+      "loss": 0.9402,
+      "step": 363
+    },
+    {
+      "epoch": 1.123456790123457,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.722105263157895e-05,
+      "loss": 0.951,
+      "step": 364
+    },
+    {
+      "epoch": 1.126543209876543,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.7210526315789477e-05,
+      "loss": 0.8921,
+      "step": 365
+    },
+    {
+      "epoch": 1.1296296296296295,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.72e-05,
+      "loss": 0.8948,
+      "step": 366
+    },
+    {
+      "epoch": 1.132716049382716,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7189473684210526e-05,
+      "loss": 0.9533,
+      "step": 367
+    },
+    {
+      "epoch": 1.1358024691358024,
+      "grad_norm": 9.875,
+      "learning_rate": 1.7178947368421054e-05,
+      "loss": 0.8237,
+      "step": 368
+    },
+    {
+      "epoch": 1.1388888888888888,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.7168421052631582e-05,
+      "loss": 0.977,
+      "step": 369
+    },
+    {
+      "epoch": 1.1419753086419753,
+      "grad_norm": 8.5,
+      "learning_rate": 1.7157894736842107e-05,
+      "loss": 0.8955,
+      "step": 370
+    },
+    {
+      "epoch": 1.1450617283950617,
+      "grad_norm": 21.125,
+      "learning_rate": 1.714736842105263e-05,
+      "loss": 0.9086,
+      "step": 371
+    },
+    {
+      "epoch": 1.1481481481481481,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.713684210526316e-05,
+      "loss": 0.969,
+      "step": 372
+    },
+    {
+      "epoch": 1.1512345679012346,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.7126315789473687e-05,
+      "loss": 0.9897,
+      "step": 373
+    },
+    {
+      "epoch": 1.154320987654321,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7115789473684212e-05,
+      "loss": 0.9847,
+      "step": 374
+    },
+    {
+      "epoch": 1.1574074074074074,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.7105263157894737e-05,
+      "loss": 0.9894,
+      "step": 375
+    },
+    {
+      "epoch": 1.1604938271604939,
+      "grad_norm": 7.25,
+      "learning_rate": 1.7094736842105265e-05,
+      "loss": 0.9059,
+      "step": 376
+    },
+    {
+      "epoch": 1.1635802469135803,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.708421052631579e-05,
+      "loss": 0.9398,
+      "step": 377
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.7073684210526317e-05,
+      "loss": 0.8872,
+      "step": 378
+    },
+    {
+      "epoch": 1.1697530864197532,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.7063157894736845e-05,
+      "loss": 0.8904,
+      "step": 379
+    },
+    {
+      "epoch": 1.1728395061728394,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 0.9352,
+      "step": 380
+    },
+    {
+      "epoch": 1.175925925925926,
+      "grad_norm": 7.25,
+      "learning_rate": 1.7042105263157895e-05,
+      "loss": 0.9192,
+      "step": 381
+    },
+    {
+      "epoch": 1.1790123456790123,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.7031578947368423e-05,
+      "loss": 0.8636,
+      "step": 382
+    },
+    {
+      "epoch": 1.1820987654320987,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.702105263157895e-05,
+      "loss": 0.8618,
+      "step": 383
+    },
+    {
+      "epoch": 1.1851851851851851,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.7010526315789475e-05,
+      "loss": 0.872,
+      "step": 384
+    },
+    {
+      "epoch": 1.1882716049382716,
+      "grad_norm": 10.875,
+      "learning_rate": 1.7e-05,
+      "loss": 0.9046,
+      "step": 385
+    },
+    {
+      "epoch": 1.191358024691358,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.6989473684210528e-05,
+      "loss": 0.9099,
+      "step": 386
+    },
+    {
+      "epoch": 1.1944444444444444,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.6978947368421056e-05,
+      "loss": 0.9021,
+      "step": 387
+    },
+    {
+      "epoch": 1.1975308641975309,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.696842105263158e-05,
+      "loss": 0.9132,
+      "step": 388
+    },
+    {
+      "epoch": 1.2006172839506173,
+      "grad_norm": 6.75,
+      "learning_rate": 1.6957894736842105e-05,
+      "loss": 0.9235,
+      "step": 389
+    },
+    {
+      "epoch": 1.2037037037037037,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.6947368421052633e-05,
+      "loss": 0.8587,
+      "step": 390
+    },
+    {
+      "epoch": 1.2067901234567902,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.6936842105263158e-05,
+      "loss": 0.9701,
+      "step": 391
+    },
+    {
+      "epoch": 1.2098765432098766,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.6926315789473686e-05,
+      "loss": 0.9205,
+      "step": 392
+    },
+    {
+      "epoch": 1.212962962962963,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.691578947368421e-05,
+      "loss": 0.8929,
+      "step": 393
+    },
+    {
+      "epoch": 1.2160493827160495,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.690526315789474e-05,
+      "loss": 0.9513,
+      "step": 394
+    },
+    {
+      "epoch": 1.2191358024691359,
+      "grad_norm": 8.0,
+      "learning_rate": 1.6894736842105263e-05,
+      "loss": 0.9499,
+      "step": 395
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.688421052631579e-05,
+      "loss": 0.9235,
+      "step": 396
+    },
+    {
+      "epoch": 1.2253086419753085,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.687368421052632e-05,
+      "loss": 0.974,
+      "step": 397
+    },
+    {
+      "epoch": 1.228395061728395,
+      "grad_norm": 6.875,
+      "learning_rate": 1.6863157894736844e-05,
+      "loss": 0.9445,
+      "step": 398
+    },
+    {
+      "epoch": 1.2314814814814814,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.685263157894737e-05,
+      "loss": 0.9269,
+      "step": 399
+    },
+    {
+      "epoch": 1.2345679012345678,
+      "grad_norm": 8.0,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.9272,
+      "step": 400
+    },
+    {
+      "epoch": 1.2376543209876543,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.6831578947368424e-05,
+      "loss": 0.8816,
+      "step": 401
+    },
+    {
+      "epoch": 1.2407407407407407,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.682105263157895e-05,
+      "loss": 0.8917,
+      "step": 402
+    },
+    {
+      "epoch": 1.2438271604938271,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.6810526315789474e-05,
+      "loss": 0.9393,
+      "step": 403
+    },
+    {
+      "epoch": 1.2469135802469136,
+      "grad_norm": 7.125,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.9029,
+      "step": 404
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 6.625,
+      "learning_rate": 1.6789473684210526e-05,
+      "loss": 0.9184,
+      "step": 405
+    },
+    {
+      "epoch": 1.2530864197530864,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.6778947368421054e-05,
+      "loss": 0.9023,
+      "step": 406
+    },
+    {
+      "epoch": 1.2561728395061729,
+      "grad_norm": 9.125,
+      "learning_rate": 1.676842105263158e-05,
+      "loss": 0.8674,
+      "step": 407
+    },
+    {
+      "epoch": 1.2592592592592593,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.6757894736842107e-05,
+      "loss": 0.8568,
+      "step": 408
+    },
+    {
+      "epoch": 1.2623456790123457,
+      "grad_norm": 8.25,
+      "learning_rate": 1.6747368421052632e-05,
+      "loss": 0.8676,
+      "step": 409
+    },
+    {
+      "epoch": 1.2654320987654322,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.673684210526316e-05,
+      "loss": 0.8836,
+      "step": 410
+    },
+    {
+      "epoch": 1.2685185185185186,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.6726315789473684e-05,
+      "loss": 0.8844,
+      "step": 411
+    },
+    {
+      "epoch": 1.2716049382716048,
+      "grad_norm": 8.625,
+      "learning_rate": 1.6715789473684212e-05,
+      "loss": 0.9277,
+      "step": 412
+    },
+    {
+      "epoch": 1.2746913580246915,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.6705263157894737e-05,
+      "loss": 0.8896,
+      "step": 413
+    },
+    {
+      "epoch": 1.2777777777777777,
+      "grad_norm": 8.0,
+      "learning_rate": 1.6694736842105265e-05,
+      "loss": 0.8524,
+      "step": 414
+    },
+    {
+      "epoch": 1.2808641975308643,
+      "grad_norm": 9.25,
+      "learning_rate": 1.6684210526315793e-05,
+      "loss": 0.9114,
+      "step": 415
+    },
+    {
+      "epoch": 1.2839506172839505,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.6673684210526318e-05,
+      "loss": 0.8956,
+      "step": 416
+    },
+    {
+      "epoch": 1.287037037037037,
+      "grad_norm": 6.75,
+      "learning_rate": 1.6663157894736842e-05,
+      "loss": 0.895,
+      "step": 417
+    },
+    {
+      "epoch": 1.2901234567901234,
+      "grad_norm": 7.625,
+      "learning_rate": 1.665263157894737e-05,
+      "loss": 0.8595,
+      "step": 418
+    },
+    {
+      "epoch": 1.2932098765432098,
+      "grad_norm": 7.5,
+      "learning_rate": 1.66421052631579e-05,
+      "loss": 0.9662,
+      "step": 419
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.6631578947368423e-05,
+      "loss": 0.864,
+      "step": 420
+    },
+    {
+      "epoch": 1.2993827160493827,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.6621052631578948e-05,
+      "loss": 0.8945,
+      "step": 421
+    },
+    {
+      "epoch": 1.3024691358024691,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.6610526315789476e-05,
+      "loss": 0.8424,
+      "step": 422
+    },
+    {
+      "epoch": 1.3055555555555556,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.66e-05,
+      "loss": 0.8827,
+      "step": 423
+    },
+    {
+      "epoch": 1.308641975308642,
+      "grad_norm": 8.125,
+      "learning_rate": 1.658947368421053e-05,
+      "loss": 0.9313,
+      "step": 424
+    },
+    {
+      "epoch": 1.3117283950617284,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.6578947368421053e-05,
+      "loss": 0.9034,
+      "step": 425
+    },
+    {
+      "epoch": 1.3148148148148149,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.656842105263158e-05,
+      "loss": 0.8423,
+      "step": 426
+    },
+    {
+      "epoch": 1.3179012345679013,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.6557894736842106e-05,
+      "loss": 0.881,
+      "step": 427
+    },
+    {
+      "epoch": 1.3209876543209877,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.6547368421052634e-05,
+      "loss": 0.8803,
+      "step": 428
+    },
+    {
+      "epoch": 1.324074074074074,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.653684210526316e-05,
+      "loss": 0.8824,
+      "step": 429
+    },
+    {
+      "epoch": 1.3271604938271606,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.6526315789473686e-05,
+      "loss": 0.9174,
+      "step": 430
+    },
+    {
+      "epoch": 1.3302469135802468,
+      "grad_norm": 8.75,
+      "learning_rate": 1.651578947368421e-05,
+      "loss": 0.9193,
+      "step": 431
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 7.875,
+      "learning_rate": 1.650526315789474e-05,
+      "loss": 0.8533,
+      "step": 432
+    },
+    {
+      "epoch": 1.3364197530864197,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.6494736842105267e-05,
+      "loss": 0.8703,
+      "step": 433
+    },
+    {
+      "epoch": 1.3395061728395061,
+      "grad_norm": 9.0,
+      "learning_rate": 1.648421052631579e-05,
+      "loss": 0.9284,
+      "step": 434
+    },
+    {
+      "epoch": 1.3425925925925926,
+      "grad_norm": 9.5,
+      "learning_rate": 1.6473684210526316e-05,
+      "loss": 0.903,
+      "step": 435
+    },
+    {
+      "epoch": 1.345679012345679,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.6463157894736844e-05,
+      "loss": 0.9006,
+      "step": 436
+    },
+    {
+      "epoch": 1.3487654320987654,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.645263157894737e-05,
+      "loss": 0.9046,
+      "step": 437
+    },
+    {
+      "epoch": 1.3518518518518519,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.6442105263157897e-05,
+      "loss": 0.8803,
+      "step": 438
+    },
+    {
+      "epoch": 1.3549382716049383,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.643157894736842e-05,
+      "loss": 0.9882,
+      "step": 439
+    },
+    {
+      "epoch": 1.3580246913580247,
+      "grad_norm": 8.375,
+      "learning_rate": 1.642105263157895e-05,
+      "loss": 0.8634,
+      "step": 440
+    },
+    {
+      "epoch": 1.3611111111111112,
+      "grad_norm": 11.125,
+      "learning_rate": 1.6410526315789474e-05,
+      "loss": 0.9041,
+      "step": 441
+    },
+    {
+      "epoch": 1.3641975308641976,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.64e-05,
+      "loss": 0.8601,
+      "step": 442
+    },
+    {
+      "epoch": 1.367283950617284,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.6389473684210527e-05,
+      "loss": 0.9005,
+      "step": 443
+    },
+    {
+      "epoch": 1.3703703703703702,
+      "grad_norm": 6.875,
+      "learning_rate": 1.6378947368421055e-05,
+      "loss": 0.9455,
+      "step": 444
+    },
+    {
+      "epoch": 1.373456790123457,
+      "grad_norm": 9.375,
+      "learning_rate": 1.636842105263158e-05,
+      "loss": 0.9527,
+      "step": 445
+    },
+    {
+      "epoch": 1.376543209876543,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.6357894736842108e-05,
+      "loss": 0.9098,
+      "step": 446
+    },
+    {
+      "epoch": 1.3796296296296298,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.6347368421052636e-05,
+      "loss": 0.8834,
+      "step": 447
+    },
+    {
+      "epoch": 1.382716049382716,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.633684210526316e-05,
+      "loss": 0.9359,
+      "step": 448
+    },
+    {
+      "epoch": 1.3858024691358024,
+      "grad_norm": 7.375,
+      "learning_rate": 1.6326315789473685e-05,
+      "loss": 0.9518,
+      "step": 449
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 8.75,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 0.9009,
+      "step": 450
+    },
+    {
+      "epoch": 1.3919753086419753,
+      "grad_norm": 13.875,
+      "learning_rate": 1.6305263157894737e-05,
+      "loss": 0.8877,
+      "step": 451
+    },
+    {
+      "epoch": 1.3950617283950617,
+      "grad_norm": 7.125,
+      "learning_rate": 1.6294736842105265e-05,
+      "loss": 0.8625,
+      "step": 452
+    },
+    {
+      "epoch": 1.3981481481481481,
+      "grad_norm": 9.5,
+      "learning_rate": 1.628421052631579e-05,
+      "loss": 0.9144,
+      "step": 453
+    },
+    {
+      "epoch": 1.4012345679012346,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.6273684210526318e-05,
+      "loss": 0.8676,
+      "step": 454
+    },
+    {
+      "epoch": 1.404320987654321,
+      "grad_norm": 11.0,
+      "learning_rate": 1.6263157894736843e-05,
+      "loss": 0.9233,
+      "step": 455
+    },
+    {
+      "epoch": 1.4074074074074074,
+      "grad_norm": 17.75,
+      "learning_rate": 1.6252631578947367e-05,
+      "loss": 0.8357,
+      "step": 456
+    },
+    {
+      "epoch": 1.4104938271604939,
+      "grad_norm": 8.375,
+      "learning_rate": 1.6242105263157895e-05,
+      "loss": 0.8528,
+      "step": 457
+    },
+    {
+      "epoch": 1.4135802469135803,
+      "grad_norm": 8.25,
+      "learning_rate": 1.6231578947368423e-05,
+      "loss": 0.9339,
+      "step": 458
+    },
+    {
+      "epoch": 1.4166666666666667,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.6221052631578948e-05,
+      "loss": 0.9179,
+      "step": 459
+    },
+    {
+      "epoch": 1.4197530864197532,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 0.9156,
+      "step": 460
+    },
+    {
+      "epoch": 1.4228395061728394,
+      "grad_norm": 10.875,
+      "learning_rate": 1.62e-05,
+      "loss": 0.9491,
+      "step": 461
+    },
+    {
+      "epoch": 1.425925925925926,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.618947368421053e-05,
+      "loss": 0.8316,
+      "step": 462
+    },
+    {
+      "epoch": 1.4290123456790123,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.6178947368421053e-05,
+      "loss": 0.8693,
+      "step": 463
+    },
+    {
+      "epoch": 1.4320987654320987,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.616842105263158e-05,
+      "loss": 0.8742,
+      "step": 464
+    },
+    {
+      "epoch": 1.4351851851851851,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.6157894736842106e-05,
+      "loss": 0.8725,
+      "step": 465
+    },
+    {
+      "epoch": 1.4382716049382716,
+      "grad_norm": 6.5625,
+      "learning_rate": 1.6147368421052634e-05,
+      "loss": 0.8756,
+      "step": 466
+    },
+    {
+      "epoch": 1.441358024691358,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.613684210526316e-05,
+      "loss": 0.866,
+      "step": 467
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.6126315789473687e-05,
+      "loss": 0.9003,
+      "step": 468
+    },
+    {
+      "epoch": 1.4475308641975309,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.611578947368421e-05,
+      "loss": 0.8605,
+      "step": 469
+    },
+    {
+      "epoch": 1.4506172839506173,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.6105263157894736e-05,
+      "loss": 0.9157,
+      "step": 470
+    },
+    {
+      "epoch": 1.4537037037037037,
+      "grad_norm": 8.625,
+      "learning_rate": 1.6094736842105264e-05,
+      "loss": 0.9089,
+      "step": 471
+    },
+    {
+      "epoch": 1.4567901234567902,
+      "grad_norm": 8.125,
+      "learning_rate": 1.6084210526315792e-05,
+      "loss": 0.9078,
+      "step": 472
+    },
+    {
+      "epoch": 1.4598765432098766,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.6073684210526317e-05,
+      "loss": 0.9367,
+      "step": 473
+    },
+    {
+      "epoch": 1.462962962962963,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.606315789473684e-05,
+      "loss": 0.8707,
+      "step": 474
+    },
+    {
+      "epoch": 1.4660493827160495,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.605263157894737e-05,
+      "loss": 0.8787,
+      "step": 475
+    },
+    {
+      "epoch": 1.4691358024691357,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.6042105263157897e-05,
+      "loss": 0.8755,
+      "step": 476
+    },
+    {
+      "epoch": 1.4722222222222223,
+      "grad_norm": 7.375,
+      "learning_rate": 1.6031578947368422e-05,
+      "loss": 0.9681,
+      "step": 477
+    },
+    {
+      "epoch": 1.4753086419753085,
+      "grad_norm": 7.75,
+      "learning_rate": 1.6021052631578947e-05,
+      "loss": 0.8579,
+      "step": 478
+    },
+    {
+      "epoch": 1.4783950617283952,
+      "grad_norm": 6.875,
+      "learning_rate": 1.6010526315789475e-05,
+      "loss": 0.8914,
+      "step": 479
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.8911,
+      "step": 480
+    },
+    {
+      "epoch": 1.4845679012345678,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.5989473684210527e-05,
+      "loss": 0.869,
+      "step": 481
+    },
+    {
+      "epoch": 1.4876543209876543,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.5978947368421055e-05,
+      "loss": 0.9006,
+      "step": 482
+    },
+    {
+      "epoch": 1.4907407407407407,
+      "grad_norm": 7.125,
+      "learning_rate": 1.596842105263158e-05,
+      "loss": 0.8532,
+      "step": 483
+    },
+    {
+      "epoch": 1.4938271604938271,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.5957894736842105e-05,
+      "loss": 0.8859,
+      "step": 484
+    },
+    {
+      "epoch": 1.4969135802469136,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.5947368421052633e-05,
+      "loss": 0.8714,
+      "step": 485
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 6.75,
+      "learning_rate": 1.593684210526316e-05,
+      "loss": 0.8659,
+      "step": 486
+    },
+    {
+      "epoch": 1.5030864197530864,
+      "grad_norm": 7.625,
+      "learning_rate": 1.5926315789473685e-05,
+      "loss": 0.8689,
+      "step": 487
+    },
+    {
+      "epoch": 1.5061728395061729,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.591578947368421e-05,
+      "loss": 0.8845,
+      "step": 488
+    },
+    {
+      "epoch": 1.5092592592592593,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.5905263157894738e-05,
+      "loss": 0.8656,
+      "step": 489
+    },
+    {
+      "epoch": 1.5123456790123457,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.5894736842105266e-05,
+      "loss": 0.8989,
+      "step": 490
+    },
+    {
+      "epoch": 1.515432098765432,
+      "grad_norm": 8.75,
+      "learning_rate": 1.588421052631579e-05,
+      "loss": 0.9197,
+      "step": 491
+    },
+    {
+      "epoch": 1.5185185185185186,
+      "grad_norm": 7.75,
+      "learning_rate": 1.5873684210526315e-05,
+      "loss": 0.8972,
+      "step": 492
+    },
+    {
+      "epoch": 1.5216049382716048,
+      "grad_norm": 9.0,
+      "learning_rate": 1.5863157894736843e-05,
+      "loss": 0.8536,
+      "step": 493
+    },
+    {
+      "epoch": 1.5246913580246915,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.585263157894737e-05,
+      "loss": 0.8999,
+      "step": 494
+    },
+    {
+      "epoch": 1.5277777777777777,
+      "grad_norm": 9.625,
+      "learning_rate": 1.5842105263157896e-05,
+      "loss": 0.9513,
+      "step": 495
+    },
+    {
+      "epoch": 1.5308641975308643,
+      "grad_norm": 8.0,
+      "learning_rate": 1.5831578947368424e-05,
+      "loss": 0.931,
+      "step": 496
+    },
+    {
+      "epoch": 1.5339506172839505,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.582105263157895e-05,
+      "loss": 0.8779,
+      "step": 497
+    },
+    {
+      "epoch": 1.5370370370370372,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.5810526315789473e-05,
+      "loss": 0.8619,
+      "step": 498
+    },
+    {
+      "epoch": 1.5401234567901234,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.58e-05,
+      "loss": 0.8874,
+      "step": 499
+    },
+    {
+      "epoch": 1.5432098765432098,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.9158,
+      "step": 500
+    },
+    {
+      "epoch": 1.5462962962962963,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.5778947368421054e-05,
+      "loss": 0.8992,
+      "step": 501
+    },
+    {
+      "epoch": 1.5493827160493827,
+      "grad_norm": 10.0,
+      "learning_rate": 1.576842105263158e-05,
+      "loss": 0.8492,
+      "step": 502
+    },
+    {
+      "epoch": 1.5524691358024691,
+      "grad_norm": 8.5,
+      "learning_rate": 1.5757894736842107e-05,
+      "loss": 0.894,
+      "step": 503
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.5747368421052635e-05,
+      "loss": 0.8663,
+      "step": 504
+    },
+    {
+      "epoch": 1.558641975308642,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.573684210526316e-05,
+      "loss": 0.8489,
+      "step": 505
+    },
+    {
+      "epoch": 1.5617283950617284,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.5726315789473684e-05,
+      "loss": 0.8539,
+      "step": 506
+    },
+    {
+      "epoch": 1.5648148148148149,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.5715789473684212e-05,
+      "loss": 0.8705,
+      "step": 507
+    },
+    {
+      "epoch": 1.567901234567901,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.570526315789474e-05,
+      "loss": 0.8939,
+      "step": 508
+    },
+    {
+      "epoch": 1.5709876543209877,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.5694736842105264e-05,
+      "loss": 0.8757,
+      "step": 509
+    },
+    {
+      "epoch": 1.574074074074074,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.568421052631579e-05,
+      "loss": 0.8527,
+      "step": 510
+    },
+    {
+      "epoch": 1.5771604938271606,
+      "grad_norm": 10.625,
+      "learning_rate": 1.5673684210526317e-05,
+      "loss": 0.8325,
+      "step": 511
+    },
+    {
+      "epoch": 1.5802469135802468,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.5663157894736842e-05,
+      "loss": 0.9399,
+      "step": 512
+    },
+    {
+      "epoch": 1.5833333333333335,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.565263157894737e-05,
+      "loss": 0.8427,
+      "step": 513
+    },
+    {
+      "epoch": 1.5864197530864197,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.5642105263157898e-05,
+      "loss": 0.821,
+      "step": 514
+    },
+    {
+      "epoch": 1.5895061728395061,
+      "grad_norm": 7.0,
+      "learning_rate": 1.5631578947368422e-05,
+      "loss": 0.8746,
+      "step": 515
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.5621052631578947e-05,
+      "loss": 0.8205,
+      "step": 516
+    },
+    {
+      "epoch": 1.595679012345679,
+      "grad_norm": 9.125,
+      "learning_rate": 1.5610526315789475e-05,
+      "loss": 0.8179,
+      "step": 517
+    },
+    {
+      "epoch": 1.5987654320987654,
+      "grad_norm": 7.5,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.8613,
+      "step": 518
+    },
+    {
+      "epoch": 1.6018518518518519,
+      "grad_norm": 6.28125,
+      "learning_rate": 1.5589473684210528e-05,
+      "loss": 0.8469,
+      "step": 519
+    },
+    {
+      "epoch": 1.6049382716049383,
+      "grad_norm": 7.25,
+      "learning_rate": 1.5578947368421052e-05,
+      "loss": 0.8788,
+      "step": 520
+    },
+    {
+      "epoch": 1.6080246913580247,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.556842105263158e-05,
+      "loss": 0.8593,
+      "step": 521
+    },
+    {
+      "epoch": 1.6111111111111112,
+      "grad_norm": 8.375,
+      "learning_rate": 1.555789473684211e-05,
+      "loss": 0.8711,
+      "step": 522
+    },
+    {
+      "epoch": 1.6141975308641974,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.5547368421052633e-05,
+      "loss": 0.8189,
+      "step": 523
+    },
+    {
+      "epoch": 1.617283950617284,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.5536842105263158e-05,
+      "loss": 0.8366,
+      "step": 524
+    },
+    {
+      "epoch": 1.6203703703703702,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.5526315789473686e-05,
+      "loss": 0.8692,
+      "step": 525
+    },
+    {
+      "epoch": 1.623456790123457,
+      "grad_norm": 8.625,
+      "learning_rate": 1.551578947368421e-05,
+      "loss": 0.8365,
+      "step": 526
+    },
+    {
+      "epoch": 1.626543209876543,
+      "grad_norm": 8.25,
+      "learning_rate": 1.550526315789474e-05,
+      "loss": 0.8677,
+      "step": 527
+    },
+    {
+      "epoch": 1.6296296296296298,
+      "grad_norm": 6.75,
+      "learning_rate": 1.5494736842105263e-05,
+      "loss": 0.8187,
+      "step": 528
+    },
+    {
+      "epoch": 1.632716049382716,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.548421052631579e-05,
+      "loss": 0.8749,
+      "step": 529
+    },
+    {
+      "epoch": 1.6358024691358026,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.5473684210526316e-05,
+      "loss": 0.8947,
+      "step": 530
+    },
+    {
+      "epoch": 1.6388888888888888,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.5463157894736844e-05,
+      "loss": 0.8451,
+      "step": 531
+    },
+    {
+      "epoch": 1.6419753086419753,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.545263157894737e-05,
+      "loss": 0.8214,
+      "step": 532
+    },
+    {
+      "epoch": 1.6450617283950617,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.5442105263157896e-05,
+      "loss": 0.8369,
+      "step": 533
+    },
+    {
+      "epoch": 1.6481481481481481,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.543157894736842e-05,
+      "loss": 0.886,
+      "step": 534
+    },
+    {
+      "epoch": 1.6512345679012346,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.542105263157895e-05,
+      "loss": 0.8158,
+      "step": 535
+    },
+    {
+      "epoch": 1.654320987654321,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.5410526315789477e-05,
+      "loss": 0.8193,
+      "step": 536
+    },
+    {
+      "epoch": 1.6574074074074074,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.54e-05,
+      "loss": 0.8573,
+      "step": 537
+    },
+    {
+      "epoch": 1.6604938271604939,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.5389473684210526e-05,
+      "loss": 0.8796,
+      "step": 538
+    },
+    {
+      "epoch": 1.6635802469135803,
+      "grad_norm": 8.125,
+      "learning_rate": 1.5378947368421054e-05,
+      "loss": 0.8673,
+      "step": 539
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 6.34375,
+      "learning_rate": 1.536842105263158e-05,
+      "loss": 0.8867,
+      "step": 540
+    },
+    {
+      "epoch": 1.6697530864197532,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.5357894736842107e-05,
+      "loss": 0.8724,
+      "step": 541
+    },
+    {
+      "epoch": 1.6728395061728394,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.534736842105263e-05,
+      "loss": 0.8598,
+      "step": 542
+    },
+    {
+      "epoch": 1.675925925925926,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.533684210526316e-05,
+      "loss": 0.8418,
+      "step": 543
+    },
+    {
+      "epoch": 1.6790123456790123,
+      "grad_norm": 8.375,
+      "learning_rate": 1.5326315789473684e-05,
+      "loss": 0.8757,
+      "step": 544
+    },
+    {
+      "epoch": 1.682098765432099,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.5315789473684212e-05,
+      "loss": 0.8106,
+      "step": 545
+    },
+    {
+      "epoch": 1.6851851851851851,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.530526315789474e-05,
+      "loss": 0.9382,
+      "step": 546
+    },
+    {
+      "epoch": 1.6882716049382716,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.5294736842105265e-05,
+      "loss": 0.8419,
+      "step": 547
+    },
+    {
+      "epoch": 1.691358024691358,
+      "grad_norm": 8.75,
+      "learning_rate": 1.528421052631579e-05,
+      "loss": 0.8749,
+      "step": 548
+    },
+    {
+      "epoch": 1.6944444444444444,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.5273684210526318e-05,
+      "loss": 0.8468,
+      "step": 549
+    },
+    {
+      "epoch": 1.6975308641975309,
+      "grad_norm": 8.75,
+      "learning_rate": 1.5263157894736846e-05,
+      "loss": 0.8624,
+      "step": 550
+    },
+    {
+      "epoch": 1.7006172839506173,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.525263157894737e-05,
+      "loss": 0.8594,
+      "step": 551
+    },
+    {
+      "epoch": 1.7037037037037037,
+      "grad_norm": 9.875,
+      "learning_rate": 1.5242105263157897e-05,
+      "loss": 0.8676,
+      "step": 552
+    },
+    {
+      "epoch": 1.7067901234567902,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.5231578947368421e-05,
+      "loss": 0.8897,
+      "step": 553
+    },
+    {
+      "epoch": 1.7098765432098766,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.5221052631578948e-05,
+      "loss": 0.817,
+      "step": 554
+    },
+    {
+      "epoch": 1.7129629629629628,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.5210526315789476e-05,
+      "loss": 0.8579,
+      "step": 555
+    },
+    {
+      "epoch": 1.7160493827160495,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.8645,
+      "step": 556
+    },
+    {
+      "epoch": 1.7191358024691357,
+      "grad_norm": 9.25,
+      "learning_rate": 1.5189473684210526e-05,
+      "loss": 0.8662,
+      "step": 557
+    },
+    {
+      "epoch": 1.7222222222222223,
+      "grad_norm": 9.125,
+      "learning_rate": 1.5178947368421053e-05,
+      "loss": 0.8516,
+      "step": 558
+    },
+    {
+      "epoch": 1.7253086419753085,
+      "grad_norm": 7.75,
+      "learning_rate": 1.516842105263158e-05,
+      "loss": 0.8483,
+      "step": 559
+    },
+    {
+      "epoch": 1.7283950617283952,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.5157894736842107e-05,
+      "loss": 0.8253,
+      "step": 560
+    },
+    {
+      "epoch": 1.7314814814814814,
+      "grad_norm": 10.25,
+      "learning_rate": 1.5147368421052633e-05,
+      "loss": 0.8107,
+      "step": 561
+    },
+    {
+      "epoch": 1.734567901234568,
+      "grad_norm": 7.75,
+      "learning_rate": 1.5136842105263158e-05,
+      "loss": 0.8574,
+      "step": 562
+    },
+    {
+      "epoch": 1.7376543209876543,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5126315789473684e-05,
+      "loss": 0.8397,
+      "step": 563
+    },
+    {
+      "epoch": 1.7407407407407407,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.5115789473684212e-05,
+      "loss": 0.824,
+      "step": 564
+    },
+    {
+      "epoch": 1.7438271604938271,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.5105263157894739e-05,
+      "loss": 0.8392,
+      "step": 565
+    },
+    {
+      "epoch": 1.7469135802469136,
+      "grad_norm": 6.75,
+      "learning_rate": 1.5094736842105263e-05,
+      "loss": 0.8424,
+      "step": 566
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.508421052631579e-05,
+      "loss": 0.8415,
+      "step": 567
+    },
+    {
+      "epoch": 1.7530864197530864,
+      "grad_norm": 5.84375,
+      "learning_rate": 1.5073684210526316e-05,
+      "loss": 0.8239,
+      "step": 568
+    },
+    {
+      "epoch": 1.7561728395061729,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.5063157894736844e-05,
+      "loss": 0.895,
+      "step": 569
+    },
+    {
+      "epoch": 1.7592592592592593,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.505263157894737e-05,
+      "loss": 0.8774,
+      "step": 570
+    },
+    {
+      "epoch": 1.7623456790123457,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.5042105263157895e-05,
+      "loss": 0.8044,
+      "step": 571
+    },
+    {
+      "epoch": 1.765432098765432,
+      "grad_norm": 12.0,
+      "learning_rate": 1.5031578947368421e-05,
+      "loss": 0.9145,
+      "step": 572
+    },
+    {
+      "epoch": 1.7685185185185186,
+      "grad_norm": 7.375,
+      "learning_rate": 1.502105263157895e-05,
+      "loss": 0.7972,
+      "step": 573
+    },
+    {
+      "epoch": 1.7716049382716048,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.5010526315789476e-05,
+      "loss": 0.7876,
+      "step": 574
+    },
+    {
+      "epoch": 1.7746913580246915,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8747,
+      "step": 575
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.4989473684210527e-05,
+      "loss": 0.8623,
+      "step": 576
+    },
+    {
+      "epoch": 1.7808641975308643,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.4978947368421053e-05,
+      "loss": 0.8237,
+      "step": 577
+    },
+    {
+      "epoch": 1.7839506172839505,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.4968421052631581e-05,
+      "loss": 0.8878,
+      "step": 578
+    },
+    {
+      "epoch": 1.7870370370370372,
+      "grad_norm": 26.875,
+      "learning_rate": 1.4957894736842107e-05,
+      "loss": 0.862,
+      "step": 579
+    },
+    {
+      "epoch": 1.7901234567901234,
+      "grad_norm": 9.375,
+      "learning_rate": 1.4947368421052632e-05,
+      "loss": 0.8701,
+      "step": 580
+    },
+    {
+      "epoch": 1.7932098765432098,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.4936842105263158e-05,
+      "loss": 0.7985,
+      "step": 581
+    },
+    {
+      "epoch": 1.7962962962962963,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.4926315789473686e-05,
+      "loss": 0.8738,
+      "step": 582
+    },
+    {
+      "epoch": 1.7993827160493827,
+      "grad_norm": 7.0,
+      "learning_rate": 1.4915789473684213e-05,
+      "loss": 0.8828,
+      "step": 583
+    },
+    {
+      "epoch": 1.8024691358024691,
+      "grad_norm": 43.25,
+      "learning_rate": 1.4905263157894739e-05,
+      "loss": 0.8628,
+      "step": 584
+    },
+    {
+      "epoch": 1.8055555555555556,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.4894736842105264e-05,
+      "loss": 0.8598,
+      "step": 585
+    },
+    {
+      "epoch": 1.808641975308642,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.488421052631579e-05,
+      "loss": 0.8133,
+      "step": 586
+    },
+    {
+      "epoch": 1.8117283950617284,
+      "grad_norm": 8.125,
+      "learning_rate": 1.4873684210526318e-05,
+      "loss": 0.7877,
+      "step": 587
+    },
+    {
+      "epoch": 1.8148148148148149,
+      "grad_norm": 11.5,
+      "learning_rate": 1.4863157894736844e-05,
+      "loss": 0.8498,
+      "step": 588
+    },
+    {
+      "epoch": 1.817901234567901,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.4852631578947369e-05,
+      "loss": 0.8652,
+      "step": 589
+    },
+    {
+      "epoch": 1.8209876543209877,
+      "grad_norm": 6.875,
+      "learning_rate": 1.4842105263157895e-05,
+      "loss": 0.7987,
+      "step": 590
+    },
+    {
+      "epoch": 1.824074074074074,
+      "grad_norm": 7.875,
+      "learning_rate": 1.4831578947368422e-05,
+      "loss": 0.8074,
+      "step": 591
+    },
+    {
+      "epoch": 1.8271604938271606,
+      "grad_norm": 7.0,
+      "learning_rate": 1.482105263157895e-05,
+      "loss": 0.8298,
+      "step": 592
+    },
+    {
+      "epoch": 1.8302469135802468,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.4810526315789476e-05,
+      "loss": 0.8211,
+      "step": 593
+    },
+    {
+      "epoch": 1.8333333333333335,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.48e-05,
+      "loss": 0.7528,
+      "step": 594
+    },
+    {
+      "epoch": 1.8364197530864197,
+      "grad_norm": 6.875,
+      "learning_rate": 1.4789473684210527e-05,
+      "loss": 0.859,
+      "step": 595
+    },
+    {
+      "epoch": 1.8395061728395061,
+      "grad_norm": 11.125,
+      "learning_rate": 1.4778947368421055e-05,
+      "loss": 0.9496,
+      "step": 596
+    },
+    {
+      "epoch": 1.8425925925925926,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.4768421052631581e-05,
+      "loss": 0.8202,
+      "step": 597
+    },
+    {
+      "epoch": 1.845679012345679,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.4757894736842106e-05,
+      "loss": 0.8223,
+      "step": 598
+    },
+    {
+      "epoch": 1.8487654320987654,
+      "grad_norm": 8.75,
+      "learning_rate": 1.4747368421052632e-05,
+      "loss": 0.8144,
+      "step": 599
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.8449,
+      "step": 600
+    },
+    {
+      "epoch": 1.8549382716049383,
+      "grad_norm": 11.375,
+      "learning_rate": 1.4726315789473687e-05,
+      "loss": 0.8698,
+      "step": 601
+    },
+    {
+      "epoch": 1.8580246913580247,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.4715789473684213e-05,
+      "loss": 0.8276,
+      "step": 602
+    },
+    {
+      "epoch": 1.8611111111111112,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.4705263157894738e-05,
+      "loss": 0.8339,
+      "step": 603
+    },
+    {
+      "epoch": 1.8641975308641974,
+      "grad_norm": 11.625,
+      "learning_rate": 1.4694736842105264e-05,
+      "loss": 0.8364,
+      "step": 604
+    },
+    {
+      "epoch": 1.867283950617284,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.468421052631579e-05,
+      "loss": 0.8973,
+      "step": 605
+    },
+    {
+      "epoch": 1.8703703703703702,
+      "grad_norm": 11.5,
+      "learning_rate": 1.4673684210526318e-05,
+      "loss": 0.7722,
+      "step": 606
+    },
+    {
+      "epoch": 1.873456790123457,
+      "grad_norm": 7.5,
+      "learning_rate": 1.4663157894736843e-05,
+      "loss": 0.8528,
+      "step": 607
+    },
+    {
+      "epoch": 1.876543209876543,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.465263157894737e-05,
+      "loss": 0.843,
+      "step": 608
+    },
+    {
+      "epoch": 1.8796296296296298,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.4642105263157896e-05,
+      "loss": 0.8593,
+      "step": 609
+    },
+    {
+      "epoch": 1.882716049382716,
+      "grad_norm": 19.5,
+      "learning_rate": 1.4631578947368424e-05,
+      "loss": 0.8433,
+      "step": 610
+    },
+    {
+      "epoch": 1.8858024691358026,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.462105263157895e-05,
+      "loss": 0.8365,
+      "step": 611
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 7.75,
+      "learning_rate": 1.4610526315789474e-05,
+      "loss": 0.8366,
+      "step": 612
+    },
+    {
+      "epoch": 1.8919753086419753,
+      "grad_norm": 7.75,
+      "learning_rate": 1.46e-05,
+      "loss": 0.8609,
+      "step": 613
+    },
+    {
+      "epoch": 1.8950617283950617,
+      "grad_norm": 7.25,
+      "learning_rate": 1.4589473684210527e-05,
+      "loss": 0.7284,
+      "step": 614
+    },
+    {
+      "epoch": 1.8981481481481481,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.4578947368421055e-05,
+      "loss": 0.7984,
+      "step": 615
+    },
+    {
+      "epoch": 1.9012345679012346,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.456842105263158e-05,
+      "loss": 0.8086,
+      "step": 616
+    },
+    {
+      "epoch": 1.904320987654321,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.4557894736842106e-05,
+      "loss": 0.8288,
+      "step": 617
+    },
+    {
+      "epoch": 1.9074074074074074,
+      "grad_norm": 10.5,
+      "learning_rate": 1.4547368421052632e-05,
+      "loss": 0.8377,
+      "step": 618
+    },
+    {
+      "epoch": 1.9104938271604939,
+      "grad_norm": 10.875,
+      "learning_rate": 1.4536842105263159e-05,
+      "loss": 0.8495,
+      "step": 619
+    },
+    {
+      "epoch": 1.9135802469135803,
+      "grad_norm": 7.875,
+      "learning_rate": 1.4526315789473687e-05,
+      "loss": 0.8875,
+      "step": 620
+    },
+    {
+      "epoch": 1.9166666666666665,
+      "grad_norm": 9.375,
+      "learning_rate": 1.4515789473684211e-05,
+      "loss": 0.8598,
+      "step": 621
+    },
+    {
+      "epoch": 1.9197530864197532,
+      "grad_norm": 9.375,
+      "learning_rate": 1.4505263157894738e-05,
+      "loss": 0.8516,
+      "step": 622
+    },
+    {
+      "epoch": 1.9228395061728394,
+      "grad_norm": 11.875,
+      "learning_rate": 1.4494736842105264e-05,
+      "loss": 0.8137,
+      "step": 623
+    },
+    {
+      "epoch": 1.925925925925926,
+      "grad_norm": 6.625,
+      "learning_rate": 1.4484210526315792e-05,
+      "loss": 0.8004,
+      "step": 624
+    },
+    {
+      "epoch": 1.9290123456790123,
+      "grad_norm": 8.75,
+      "learning_rate": 1.4473684210526317e-05,
+      "loss": 0.8818,
+      "step": 625
+    },
+    {
+      "epoch": 1.932098765432099,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.4463157894736843e-05,
+      "loss": 0.8223,
+      "step": 626
+    },
+    {
+      "epoch": 1.9351851851851851,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.445263157894737e-05,
+      "loss": 0.7899,
+      "step": 627
+    },
+    {
+      "epoch": 1.9382716049382716,
+      "grad_norm": 7.875,
+      "learning_rate": 1.4442105263157896e-05,
+      "loss": 0.8252,
+      "step": 628
+    },
+    {
+      "epoch": 1.941358024691358,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.4431578947368424e-05,
+      "loss": 0.8692,
+      "step": 629
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.4421052631578948e-05,
+      "loss": 0.8224,
+      "step": 630
+    },
+    {
+      "epoch": 1.9475308641975309,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.4410526315789475e-05,
+      "loss": 0.803,
+      "step": 631
+    },
+    {
+      "epoch": 1.9506172839506173,
+      "grad_norm": 9.25,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.8271,
+      "step": 632
+    },
+    {
+      "epoch": 1.9537037037037037,
+      "grad_norm": 18.625,
+      "learning_rate": 1.4389473684210526e-05,
+      "loss": 0.809,
+      "step": 633
+    },
+    {
+      "epoch": 1.9567901234567902,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.4378947368421054e-05,
+      "loss": 0.8275,
+      "step": 634
+    },
+    {
+      "epoch": 1.9598765432098766,
+      "grad_norm": 8.375,
+      "learning_rate": 1.436842105263158e-05,
+      "loss": 0.8882,
+      "step": 635
+    },
+    {
+      "epoch": 1.9629629629629628,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.4357894736842106e-05,
+      "loss": 0.8135,
+      "step": 636
+    },
+    {
+      "epoch": 1.9660493827160495,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.4347368421052633e-05,
+      "loss": 0.8266,
+      "step": 637
+    },
+    {
+      "epoch": 1.9691358024691357,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.433684210526316e-05,
+      "loss": 0.7826,
+      "step": 638
+    },
+    {
+      "epoch": 1.9722222222222223,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.4326315789473685e-05,
+      "loss": 0.8297,
+      "step": 639
+    },
+    {
+      "epoch": 1.9753086419753085,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.4315789473684212e-05,
+      "loss": 0.7878,
+      "step": 640
+    },
+    {
+      "epoch": 1.9783950617283952,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.4305263157894738e-05,
+      "loss": 0.8487,
+      "step": 641
+    },
+    {
+      "epoch": 1.9814814814814814,
+      "grad_norm": 7.375,
+      "learning_rate": 1.4294736842105263e-05,
+      "loss": 0.8353,
+      "step": 642
+    },
+    {
+      "epoch": 1.984567901234568,
+      "grad_norm": 8.0,
+      "learning_rate": 1.4284210526315792e-05,
+      "loss": 0.8133,
+      "step": 643
+    },
+    {
+      "epoch": 1.9876543209876543,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.4273684210526317e-05,
+      "loss": 0.7983,
+      "step": 644
+    },
+    {
+      "epoch": 1.9907407407407407,
+      "grad_norm": 7.125,
+      "learning_rate": 1.4263157894736843e-05,
+      "loss": 0.8372,
+      "step": 645
+    },
+    {
+      "epoch": 1.9938271604938271,
+      "grad_norm": 9.75,
+      "learning_rate": 1.425263157894737e-05,
+      "loss": 0.7891,
+      "step": 646
+    },
+    {
+      "epoch": 1.9969135802469136,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4242105263157894e-05,
+      "loss": 0.791,
+      "step": 647
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.4231578947368422e-05,
+      "loss": 0.792,
+      "step": 648
+    },
+    {
+      "epoch": 2.003086419753086,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.4221052631578949e-05,
+      "loss": 0.7473,
+      "step": 649
+    },
+    {
+      "epoch": 2.006172839506173,
+      "grad_norm": 7.0,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 0.8113,
+      "step": 650
+    },
+    {
+      "epoch": 2.009259259259259,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 0.789,
+      "step": 651
+    },
+    {
+      "epoch": 2.0123456790123457,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.418947368421053e-05,
+      "loss": 0.7845,
+      "step": 652
+    },
+    {
+      "epoch": 2.015432098765432,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.4178947368421054e-05,
+      "loss": 0.8404,
+      "step": 653
+    },
+    {
+      "epoch": 2.0185185185185186,
+      "grad_norm": 6.09375,
+      "learning_rate": 1.416842105263158e-05,
+      "loss": 0.7845,
+      "step": 654
+    },
+    {
+      "epoch": 2.021604938271605,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.4157894736842107e-05,
+      "loss": 0.9012,
+      "step": 655
+    },
+    {
+      "epoch": 2.0246913580246915,
+      "grad_norm": 67.0,
+      "learning_rate": 1.4147368421052631e-05,
+      "loss": 0.8304,
+      "step": 656
+    },
+    {
+      "epoch": 2.0277777777777777,
+      "grad_norm": 6.53125,
+      "learning_rate": 1.413684210526316e-05,
+      "loss": 0.7734,
+      "step": 657
+    },
+    {
+      "epoch": 2.0308641975308643,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.4126315789473686e-05,
+      "loss": 0.8264,
+      "step": 658
+    },
+    {
+      "epoch": 2.0339506172839505,
+      "grad_norm": 9.875,
+      "learning_rate": 1.4115789473684212e-05,
+      "loss": 0.8028,
+      "step": 659
+    },
+    {
+      "epoch": 2.037037037037037,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.4105263157894738e-05,
+      "loss": 0.7605,
+      "step": 660
+    },
+    {
+      "epoch": 2.0401234567901234,
+      "grad_norm": 6.4375,
+      "learning_rate": 1.4094736842105263e-05,
+      "loss": 0.7706,
+      "step": 661
+    },
+    {
+      "epoch": 2.04320987654321,
+      "grad_norm": 8.25,
+      "learning_rate": 1.4084210526315791e-05,
+      "loss": 0.775,
+      "step": 662
+    },
+    {
+      "epoch": 2.0462962962962963,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.4073684210526317e-05,
+      "loss": 0.7911,
+      "step": 663
+    },
+    {
+      "epoch": 2.049382716049383,
+      "grad_norm": 6.1875,
+      "learning_rate": 1.4063157894736844e-05,
+      "loss": 0.8064,
+      "step": 664
+    },
+    {
+      "epoch": 2.052469135802469,
+      "grad_norm": 7.0,
+      "learning_rate": 1.4052631578947368e-05,
+      "loss": 0.7426,
+      "step": 665
+    },
+    {
+      "epoch": 2.0555555555555554,
+      "grad_norm": 12.0,
+      "learning_rate": 1.4042105263157896e-05,
+      "loss": 0.7576,
+      "step": 666
+    },
+    {
+      "epoch": 2.058641975308642,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.4031578947368423e-05,
+      "loss": 0.7908,
+      "step": 667
+    },
+    {
+      "epoch": 2.0617283950617282,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.4021052631578949e-05,
+      "loss": 0.8206,
+      "step": 668
+    },
+    {
+      "epoch": 2.064814814814815,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.4010526315789475e-05,
+      "loss": 0.7777,
+      "step": 669
+    },
+    {
+      "epoch": 2.067901234567901,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.4e-05,
+      "loss": 0.813,
+      "step": 670
+    },
+    {
+      "epoch": 2.0709876543209877,
+      "grad_norm": 8.625,
+      "learning_rate": 1.3989473684210528e-05,
+      "loss": 0.744,
+      "step": 671
+    },
+    {
+      "epoch": 2.074074074074074,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.3978947368421054e-05,
+      "loss": 0.7906,
+      "step": 672
+    },
+    {
+      "epoch": 2.0771604938271606,
+      "grad_norm": 6.1875,
+      "learning_rate": 1.396842105263158e-05,
+      "loss": 0.7944,
+      "step": 673
+    },
+    {
+      "epoch": 2.080246913580247,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.3957894736842105e-05,
+      "loss": 0.772,
+      "step": 674
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.3947368421052631e-05,
+      "loss": 0.7911,
+      "step": 675
+    },
+    {
+      "epoch": 2.0864197530864197,
+      "grad_norm": 6.15625,
+      "learning_rate": 1.393684210526316e-05,
+      "loss": 0.7074,
+      "step": 676
+    },
+    {
+      "epoch": 2.0895061728395063,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.3926315789473686e-05,
+      "loss": 0.7654,
+      "step": 677
+    },
+    {
+      "epoch": 2.0925925925925926,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.3915789473684212e-05,
+      "loss": 0.8346,
+      "step": 678
+    },
+    {
+      "epoch": 2.095679012345679,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.3905263157894737e-05,
+      "loss": 0.7139,
+      "step": 679
+    },
+    {
+      "epoch": 2.0987654320987654,
+      "grad_norm": 7.5,
+      "learning_rate": 1.3894736842105265e-05,
+      "loss": 0.8077,
+      "step": 680
+    },
+    {
+      "epoch": 2.1018518518518516,
+      "grad_norm": 11.0,
+      "learning_rate": 1.3884210526315791e-05,
+      "loss": 0.7321,
+      "step": 681
+    },
+    {
+      "epoch": 2.1049382716049383,
+      "grad_norm": 6.53125,
+      "learning_rate": 1.3873684210526317e-05,
+      "loss": 0.8022,
+      "step": 682
+    },
+    {
+      "epoch": 2.1080246913580245,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.3863157894736842e-05,
+      "loss": 0.7995,
+      "step": 683
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.3852631578947368e-05,
+      "loss": 0.739,
+      "step": 684
+    },
+    {
+      "epoch": 2.1141975308641974,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.3842105263157896e-05,
+      "loss": 0.7496,
+      "step": 685
+    },
+    {
+      "epoch": 2.117283950617284,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.3831578947368423e-05,
+      "loss": 0.8051,
+      "step": 686
+    },
+    {
+      "epoch": 2.1203703703703702,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.3821052631578949e-05,
+      "loss": 0.7323,
+      "step": 687
+    },
+    {
+      "epoch": 2.123456790123457,
+      "grad_norm": 8.375,
+      "learning_rate": 1.3810526315789474e-05,
+      "loss": 0.816,
+      "step": 688
+    },
+    {
+      "epoch": 2.126543209876543,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.38e-05,
+      "loss": 0.7615,
+      "step": 689
+    },
+    {
+      "epoch": 2.1296296296296298,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.3789473684210528e-05,
+      "loss": 0.7445,
+      "step": 690
+    },
+    {
+      "epoch": 2.132716049382716,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.3778947368421054e-05,
+      "loss": 0.7654,
+      "step": 691
+    },
+    {
+      "epoch": 2.1358024691358026,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.3768421052631579e-05,
+      "loss": 0.7743,
+      "step": 692
+    },
+    {
+      "epoch": 2.138888888888889,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.3757894736842105e-05,
+      "loss": 0.7794,
+      "step": 693
+    },
+    {
+      "epoch": 2.1419753086419755,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.3747368421052633e-05,
+      "loss": 0.7583,
+      "step": 694
+    },
+    {
+      "epoch": 2.1450617283950617,
+      "grad_norm": 51.75,
+      "learning_rate": 1.373684210526316e-05,
+      "loss": 0.7815,
+      "step": 695
+    },
+    {
+      "epoch": 2.148148148148148,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.3726315789473686e-05,
+      "loss": 0.7772,
+      "step": 696
+    },
+    {
+      "epoch": 2.1512345679012346,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.371578947368421e-05,
+      "loss": 0.7845,
+      "step": 697
+    },
+    {
+      "epoch": 2.154320987654321,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.3705263157894737e-05,
+      "loss": 0.8217,
+      "step": 698
+    },
+    {
+      "epoch": 2.1574074074074074,
+      "grad_norm": 8.75,
+      "learning_rate": 1.3694736842105265e-05,
+      "loss": 0.8132,
+      "step": 699
+    },
+    {
+      "epoch": 2.1604938271604937,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.7884,
+      "step": 700
+    },
+    {
+      "epoch": 2.1635802469135803,
+      "grad_norm": 9.875,
+      "learning_rate": 1.3673684210526316e-05,
+      "loss": 0.7816,
+      "step": 701
+    },
+    {
+      "epoch": 2.1666666666666665,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.3663157894736842e-05,
+      "loss": 0.7974,
+      "step": 702
+    },
+    {
+      "epoch": 2.169753086419753,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3652631578947369e-05,
+      "loss": 0.813,
+      "step": 703
+    },
+    {
+      "epoch": 2.1728395061728394,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.3642105263157897e-05,
+      "loss": 0.7645,
+      "step": 704
+    },
+    {
+      "epoch": 2.175925925925926,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.3631578947368423e-05,
+      "loss": 0.8052,
+      "step": 705
+    },
+    {
+      "epoch": 2.1790123456790123,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.3621052631578948e-05,
+      "loss": 0.7857,
+      "step": 706
+    },
+    {
+      "epoch": 2.182098765432099,
+      "grad_norm": 7.625,
+      "learning_rate": 1.3610526315789474e-05,
+      "loss": 0.767,
+      "step": 707
+    },
+    {
+      "epoch": 2.185185185185185,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.789,
+      "step": 708
+    },
+    {
+      "epoch": 2.1882716049382718,
+      "grad_norm": 34.75,
+      "learning_rate": 1.3589473684210528e-05,
+      "loss": 0.7917,
+      "step": 709
+    },
+    {
+      "epoch": 2.191358024691358,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.3578947368421055e-05,
+      "loss": 0.7435,
+      "step": 710
+    },
+    {
+      "epoch": 2.1944444444444446,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.356842105263158e-05,
+      "loss": 0.8828,
+      "step": 711
+    },
+    {
+      "epoch": 2.197530864197531,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.3557894736842106e-05,
+      "loss": 0.7288,
+      "step": 712
+    },
+    {
+      "epoch": 2.200617283950617,
+      "grad_norm": 8.5,
+      "learning_rate": 1.3547368421052634e-05,
+      "loss": 0.7967,
+      "step": 713
+    },
+    {
+      "epoch": 2.2037037037037037,
+      "grad_norm": 9.0,
+      "learning_rate": 1.353684210526316e-05,
+      "loss": 0.7439,
+      "step": 714
+    },
+    {
+      "epoch": 2.20679012345679,
+      "grad_norm": 7.25,
+      "learning_rate": 1.3526315789473685e-05,
+      "loss": 0.7824,
+      "step": 715
+    },
+    {
+      "epoch": 2.2098765432098766,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.3515789473684211e-05,
+      "loss": 0.7621,
+      "step": 716
+    },
+    {
+      "epoch": 2.212962962962963,
+      "grad_norm": 6.75,
+      "learning_rate": 1.3505263157894737e-05,
+      "loss": 0.7689,
+      "step": 717
+    },
+    {
+      "epoch": 2.2160493827160495,
+      "grad_norm": 11.75,
+      "learning_rate": 1.3494736842105265e-05,
+      "loss": 0.779,
+      "step": 718
+    },
+    {
+      "epoch": 2.2191358024691357,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.3484210526315792e-05,
+      "loss": 0.8421,
+      "step": 719
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.3473684210526316e-05,
+      "loss": 0.7658,
+      "step": 720
+    },
+    {
+      "epoch": 2.2253086419753085,
+      "grad_norm": 6.65625,
+      "learning_rate": 1.3463157894736842e-05,
+      "loss": 0.813,
+      "step": 721
+    },
+    {
+      "epoch": 2.228395061728395,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.345263157894737e-05,
+      "loss": 0.7758,
+      "step": 722
+    },
+    {
+      "epoch": 2.2314814814814814,
+      "grad_norm": 7.5,
+      "learning_rate": 1.3442105263157897e-05,
+      "loss": 0.7652,
+      "step": 723
+    },
+    {
+      "epoch": 2.234567901234568,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.3431578947368421e-05,
+      "loss": 0.7843,
+      "step": 724
+    },
+    {
+      "epoch": 2.2376543209876543,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.3421052631578948e-05,
+      "loss": 0.8272,
+      "step": 725
+    },
+    {
+      "epoch": 2.240740740740741,
+      "grad_norm": 8.875,
+      "learning_rate": 1.3410526315789474e-05,
+      "loss": 0.7867,
+      "step": 726
+    },
+    {
+      "epoch": 2.243827160493827,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.7767,
+      "step": 727
+    },
+    {
+      "epoch": 2.246913580246914,
+      "grad_norm": 6.5625,
+      "learning_rate": 1.3389473684210528e-05,
+      "loss": 0.7894,
+      "step": 728
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.3378947368421053e-05,
+      "loss": 0.8093,
+      "step": 729
+    },
+    {
+      "epoch": 2.253086419753086,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.336842105263158e-05,
+      "loss": 0.7674,
+      "step": 730
+    },
+    {
+      "epoch": 2.256172839506173,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.3357894736842106e-05,
+      "loss": 0.843,
+      "step": 731
+    },
+    {
+      "epoch": 2.259259259259259,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.3347368421052634e-05,
+      "loss": 0.7567,
+      "step": 732
+    },
+    {
+      "epoch": 2.2623456790123457,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.3336842105263158e-05,
+      "loss": 0.7961,
+      "step": 733
+    },
+    {
+      "epoch": 2.265432098765432,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.3326315789473685e-05,
+      "loss": 0.7982,
+      "step": 734
+    },
+    {
+      "epoch": 2.2685185185185186,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.3315789473684211e-05,
+      "loss": 0.801,
+      "step": 735
+    },
+    {
+      "epoch": 2.271604938271605,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.3305263157894739e-05,
+      "loss": 0.8037,
+      "step": 736
+    },
+    {
+      "epoch": 2.2746913580246915,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.3294736842105265e-05,
+      "loss": 0.7824,
+      "step": 737
+    },
+    {
+      "epoch": 2.2777777777777777,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.328421052631579e-05,
+      "loss": 0.769,
+      "step": 738
+    },
+    {
+      "epoch": 2.2808641975308643,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.3273684210526316e-05,
+      "loss": 0.7914,
+      "step": 739
+    },
+    {
+      "epoch": 2.2839506172839505,
+      "grad_norm": 8.0,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 0.7509,
+      "step": 740
+    },
+    {
+      "epoch": 2.287037037037037,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.325263157894737e-05,
+      "loss": 0.8352,
+      "step": 741
+    },
+    {
+      "epoch": 2.2901234567901234,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.3242105263157895e-05,
+      "loss": 0.8307,
+      "step": 742
+    },
+    {
+      "epoch": 2.29320987654321,
+      "grad_norm": 6.53125,
+      "learning_rate": 1.3231578947368422e-05,
+      "loss": 0.7596,
+      "step": 743
+    },
+    {
+      "epoch": 2.2962962962962963,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.3221052631578948e-05,
+      "loss": 0.7884,
+      "step": 744
+    },
+    {
+      "epoch": 2.299382716049383,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.3210526315789476e-05,
+      "loss": 0.749,
+      "step": 745
+    },
+    {
+      "epoch": 2.302469135802469,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.8032,
+      "step": 746
+    },
+    {
+      "epoch": 2.3055555555555554,
+      "grad_norm": 7.0,
+      "learning_rate": 1.3189473684210527e-05,
+      "loss": 0.7547,
+      "step": 747
+    },
+    {
+      "epoch": 2.308641975308642,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.3178947368421053e-05,
+      "loss": 0.7857,
+      "step": 748
+    },
+    {
+      "epoch": 2.3117283950617282,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.316842105263158e-05,
+      "loss": 0.7248,
+      "step": 749
+    },
+    {
+      "epoch": 2.314814814814815,
+      "grad_norm": 6.5625,
+      "learning_rate": 1.3157894736842108e-05,
+      "loss": 0.7102,
+      "step": 750
+    },
+    {
+      "epoch": 2.317901234567901,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.3147368421052632e-05,
+      "loss": 0.7861,
+      "step": 751
+    },
+    {
+      "epoch": 2.3209876543209877,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.3136842105263159e-05,
+      "loss": 0.7503,
+      "step": 752
+    },
+    {
+      "epoch": 2.324074074074074,
+      "grad_norm": 7.375,
+      "learning_rate": 1.3126315789473685e-05,
+      "loss": 0.769,
+      "step": 753
+    },
+    {
+      "epoch": 2.3271604938271606,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.3115789473684211e-05,
+      "loss": 0.803,
+      "step": 754
+    },
+    {
+      "epoch": 2.330246913580247,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.310526315789474e-05,
+      "loss": 0.7948,
+      "step": 755
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 7.625,
+      "learning_rate": 1.3094736842105264e-05,
+      "loss": 0.8061,
+      "step": 756
+    },
+    {
+      "epoch": 2.3364197530864197,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.308421052631579e-05,
+      "loss": 0.7388,
+      "step": 757
+    },
+    {
+      "epoch": 2.3395061728395063,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.3073684210526317e-05,
+      "loss": 0.7264,
+      "step": 758
+    },
+    {
+      "epoch": 2.3425925925925926,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.3063157894736845e-05,
+      "loss": 0.7752,
+      "step": 759
+    },
+    {
+      "epoch": 2.3456790123456788,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.305263157894737e-05,
+      "loss": 0.7889,
+      "step": 760
+    },
+    {
+      "epoch": 2.3487654320987654,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.3042105263157896e-05,
+      "loss": 0.7746,
+      "step": 761
+    },
+    {
+      "epoch": 2.351851851851852,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.3031578947368422e-05,
+      "loss": 0.8038,
+      "step": 762
+    },
+    {
+      "epoch": 2.3549382716049383,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.3021052631578948e-05,
+      "loss": 0.8054,
+      "step": 763
+    },
+    {
+      "epoch": 2.3580246913580245,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.3010526315789476e-05,
+      "loss": 0.78,
+      "step": 764
+    },
+    {
+      "epoch": 2.361111111111111,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.77,
+      "step": 765
+    },
+    {
+      "epoch": 2.3641975308641974,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.2989473684210527e-05,
+      "loss": 0.7674,
+      "step": 766
+    },
+    {
+      "epoch": 2.367283950617284,
+      "grad_norm": 6.125,
+      "learning_rate": 1.2978947368421054e-05,
+      "loss": 0.7228,
+      "step": 767
+    },
+    {
+      "epoch": 2.3703703703703702,
+      "grad_norm": 9.75,
+      "learning_rate": 1.2968421052631578e-05,
+      "loss": 0.7798,
+      "step": 768
+    },
+    {
+      "epoch": 2.373456790123457,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.2957894736842108e-05,
+      "loss": 0.7279,
+      "step": 769
+    },
+    {
+      "epoch": 2.376543209876543,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.2947368421052633e-05,
+      "loss": 0.7976,
+      "step": 770
+    },
+    {
+      "epoch": 2.3796296296296298,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.2936842105263159e-05,
+      "loss": 0.751,
+      "step": 771
+    },
+    {
+      "epoch": 2.382716049382716,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.2926315789473685e-05,
+      "loss": 0.7475,
+      "step": 772
+    },
+    {
+      "epoch": 2.3858024691358026,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.2915789473684213e-05,
+      "loss": 0.7235,
+      "step": 773
+    },
+    {
+      "epoch": 2.388888888888889,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.2905263157894738e-05,
+      "loss": 0.7871,
+      "step": 774
+    },
+    {
+      "epoch": 2.3919753086419755,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.2894736842105264e-05,
+      "loss": 0.7624,
+      "step": 775
+    },
+    {
+      "epoch": 2.3950617283950617,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.288421052631579e-05,
+      "loss": 0.6993,
+      "step": 776
+    },
+    {
+      "epoch": 2.398148148148148,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.2873684210526317e-05,
+      "loss": 0.7898,
+      "step": 777
+    },
+    {
+      "epoch": 2.4012345679012346,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.2863157894736845e-05,
+      "loss": 0.8086,
+      "step": 778
+    },
+    {
+      "epoch": 2.4043209876543212,
+      "grad_norm": 8.0,
+      "learning_rate": 1.285263157894737e-05,
+      "loss": 0.7243,
+      "step": 779
+    },
+    {
+      "epoch": 2.4074074074074074,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.2842105263157896e-05,
+      "loss": 0.8182,
+      "step": 780
+    },
+    {
+      "epoch": 2.4104938271604937,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.2831578947368422e-05,
+      "loss": 0.7627,
+      "step": 781
+    },
+    {
+      "epoch": 2.4135802469135803,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.2821052631578947e-05,
+      "loss": 0.7446,
+      "step": 782
+    },
+    {
+      "epoch": 2.4166666666666665,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.2810526315789475e-05,
+      "loss": 0.7567,
+      "step": 783
+    },
+    {
+      "epoch": 2.419753086419753,
+      "grad_norm": 6.5,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.7301,
+      "step": 784
+    },
+    {
+      "epoch": 2.4228395061728394,
+      "grad_norm": 8.25,
+      "learning_rate": 1.2789473684210527e-05,
+      "loss": 0.7687,
+      "step": 785
+    },
+    {
+      "epoch": 2.425925925925926,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.2778947368421054e-05,
+      "loss": 0.6791,
+      "step": 786
+    },
+    {
+      "epoch": 2.4290123456790123,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.2768421052631582e-05,
+      "loss": 0.763,
+      "step": 787
+    },
+    {
+      "epoch": 2.432098765432099,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.2757894736842106e-05,
+      "loss": 0.7835,
+      "step": 788
+    },
+    {
+      "epoch": 2.435185185185185,
+      "grad_norm": 9.0,
+      "learning_rate": 1.2747368421052633e-05,
+      "loss": 0.7721,
+      "step": 789
+    },
+    {
+      "epoch": 2.4382716049382718,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.2736842105263159e-05,
+      "loss": 0.7564,
+      "step": 790
+    },
+    {
+      "epoch": 2.441358024691358,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.2726315789473684e-05,
+      "loss": 0.7643,
+      "step": 791
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.2715789473684212e-05,
+      "loss": 0.7853,
+      "step": 792
+    },
+    {
+      "epoch": 2.447530864197531,
+      "grad_norm": 14.25,
+      "learning_rate": 1.2705263157894738e-05,
+      "loss": 0.8289,
+      "step": 793
+    },
+    {
+      "epoch": 2.450617283950617,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.2694736842105264e-05,
+      "loss": 0.8073,
+      "step": 794
+    },
+    {
+      "epoch": 2.4537037037037037,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.268421052631579e-05,
+      "loss": 0.7998,
+      "step": 795
+    },
+    {
+      "epoch": 2.45679012345679,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.2673684210526315e-05,
+      "loss": 0.7962,
+      "step": 796
+    },
+    {
+      "epoch": 2.4598765432098766,
+      "grad_norm": 9.0,
+      "learning_rate": 1.2663157894736843e-05,
+      "loss": 0.7456,
+      "step": 797
+    },
+    {
+      "epoch": 2.462962962962963,
+      "grad_norm": 7.125,
+      "learning_rate": 1.265263157894737e-05,
+      "loss": 0.8182,
+      "step": 798
+    },
+    {
+      "epoch": 2.4660493827160495,
+      "grad_norm": 10.5,
+      "learning_rate": 1.2642105263157896e-05,
+      "loss": 0.7312,
+      "step": 799
+    },
+    {
+      "epoch": 2.4691358024691357,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.7529,
+      "step": 800
+    },
+    {
+      "epoch": 2.4722222222222223,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.2621052631578949e-05,
+      "loss": 0.7914,
+      "step": 801
+    },
+    {
+      "epoch": 2.4753086419753085,
+      "grad_norm": 9.125,
+      "learning_rate": 1.2610526315789475e-05,
+      "loss": 0.7535,
+      "step": 802
+    },
+    {
+      "epoch": 2.478395061728395,
+      "grad_norm": 8.625,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.7866,
+      "step": 803
+    },
+    {
+      "epoch": 2.4814814814814814,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.2589473684210528e-05,
+      "loss": 0.7654,
+      "step": 804
+    },
+    {
+      "epoch": 2.484567901234568,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.2578947368421052e-05,
+      "loss": 0.7523,
+      "step": 805
+    },
+    {
+      "epoch": 2.4876543209876543,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.256842105263158e-05,
+      "loss": 0.7451,
+      "step": 806
+    },
+    {
+      "epoch": 2.490740740740741,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.2557894736842107e-05,
+      "loss": 0.7639,
+      "step": 807
+    },
+    {
+      "epoch": 2.493827160493827,
+      "grad_norm": 6.875,
+      "learning_rate": 1.2547368421052633e-05,
+      "loss": 0.7893,
+      "step": 808
+    },
+    {
+      "epoch": 2.496913580246914,
+      "grad_norm": 6.875,
+      "learning_rate": 1.2536842105263158e-05,
+      "loss": 0.7336,
+      "step": 809
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 8.625,
+      "learning_rate": 1.2526315789473684e-05,
+      "loss": 0.781,
+      "step": 810
+    },
+    {
+      "epoch": 2.503086419753086,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.2515789473684212e-05,
+      "loss": 0.78,
+      "step": 811
+    },
+    {
+      "epoch": 2.506172839506173,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.2505263157894738e-05,
+      "loss": 0.7177,
+      "step": 812
+    },
+    {
+      "epoch": 2.5092592592592595,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.2494736842105265e-05,
+      "loss": 0.7943,
+      "step": 813
+    },
+    {
+      "epoch": 2.5123456790123457,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.248421052631579e-05,
+      "loss": 0.7835,
+      "step": 814
+    },
+    {
+      "epoch": 2.515432098765432,
+      "grad_norm": 7.0,
+      "learning_rate": 1.2473684210526317e-05,
+      "loss": 0.815,
+      "step": 815
+    },
+    {
+      "epoch": 2.5185185185185186,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.2463157894736844e-05,
+      "loss": 0.7649,
+      "step": 816
+    },
+    {
+      "epoch": 2.521604938271605,
+      "grad_norm": 8.375,
+      "learning_rate": 1.245263157894737e-05,
+      "loss": 0.7041,
+      "step": 817
+    },
+    {
+      "epoch": 2.5246913580246915,
+      "grad_norm": 7.875,
+      "learning_rate": 1.2442105263157895e-05,
+      "loss": 0.7468,
+      "step": 818
+    },
+    {
+      "epoch": 2.5277777777777777,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.2431578947368421e-05,
+      "loss": 0.788,
+      "step": 819
+    },
+    {
+      "epoch": 2.5308641975308643,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.2421052631578949e-05,
+      "loss": 0.7901,
+      "step": 820
+    },
+    {
+      "epoch": 2.5339506172839505,
+      "grad_norm": 9.875,
+      "learning_rate": 1.2410526315789475e-05,
+      "loss": 0.7545,
+      "step": 821
+    },
+    {
+      "epoch": 2.537037037037037,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.814,
+      "step": 822
+    },
+    {
+      "epoch": 2.5401234567901234,
+      "grad_norm": 7.375,
+      "learning_rate": 1.2389473684210526e-05,
+      "loss": 0.7547,
+      "step": 823
+    },
+    {
+      "epoch": 2.5432098765432096,
+      "grad_norm": 7.875,
+      "learning_rate": 1.2378947368421053e-05,
+      "loss": 0.7728,
+      "step": 824
+    },
+    {
+      "epoch": 2.5462962962962963,
+      "grad_norm": 7.0,
+      "learning_rate": 1.236842105263158e-05,
+      "loss": 0.7868,
+      "step": 825
+    },
+    {
+      "epoch": 2.549382716049383,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.2357894736842107e-05,
+      "loss": 0.7429,
+      "step": 826
+    },
+    {
+      "epoch": 2.552469135802469,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.2347368421052631e-05,
+      "loss": 0.7826,
+      "step": 827
+    },
+    {
+      "epoch": 2.5555555555555554,
+      "grad_norm": 6.34375,
+      "learning_rate": 1.2336842105263158e-05,
+      "loss": 0.7355,
+      "step": 828
+    },
+    {
+      "epoch": 2.558641975308642,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.2326315789473686e-05,
+      "loss": 0.7817,
+      "step": 829
+    },
+    {
+      "epoch": 2.5617283950617287,
+      "grad_norm": 8.75,
+      "learning_rate": 1.2315789473684212e-05,
+      "loss": 0.8024,
+      "step": 830
+    },
+    {
+      "epoch": 2.564814814814815,
+      "grad_norm": 6.4375,
+      "learning_rate": 1.2305263157894739e-05,
+      "loss": 0.7594,
+      "step": 831
+    },
+    {
+      "epoch": 2.567901234567901,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.2294736842105263e-05,
+      "loss": 0.7857,
+      "step": 832
+    },
+    {
+      "epoch": 2.5709876543209877,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.228421052631579e-05,
+      "loss": 0.738,
+      "step": 833
+    },
+    {
+      "epoch": 2.574074074074074,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.2273684210526317e-05,
+      "loss": 0.7497,
+      "step": 834
+    },
+    {
+      "epoch": 2.5771604938271606,
+      "grad_norm": 9.25,
+      "learning_rate": 1.2263157894736844e-05,
+      "loss": 0.7954,
+      "step": 835
+    },
+    {
+      "epoch": 2.580246913580247,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.225263157894737e-05,
+      "loss": 0.8367,
+      "step": 836
+    },
+    {
+      "epoch": 2.5833333333333335,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.2242105263157895e-05,
+      "loss": 0.7075,
+      "step": 837
+    },
+    {
+      "epoch": 2.5864197530864197,
+      "grad_norm": 8.375,
+      "learning_rate": 1.2231578947368421e-05,
+      "loss": 0.7439,
+      "step": 838
+    },
+    {
+      "epoch": 2.5895061728395063,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.2221052631578949e-05,
+      "loss": 0.7518,
+      "step": 839
+    },
+    {
+      "epoch": 2.5925925925925926,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.2210526315789475e-05,
+      "loss": 0.7885,
+      "step": 840
+    },
+    {
+      "epoch": 2.5956790123456788,
+      "grad_norm": 8.875,
+      "learning_rate": 1.22e-05,
+      "loss": 0.7756,
+      "step": 841
+    },
+    {
+      "epoch": 2.5987654320987654,
+      "grad_norm": 6.75,
+      "learning_rate": 1.2189473684210526e-05,
+      "loss": 0.7428,
+      "step": 842
+    },
+    {
+      "epoch": 2.601851851851852,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.2178947368421054e-05,
+      "loss": 0.7043,
+      "step": 843
+    },
+    {
+      "epoch": 2.6049382716049383,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.216842105263158e-05,
+      "loss": 0.7708,
+      "step": 844
+    },
+    {
+      "epoch": 2.6080246913580245,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.2157894736842107e-05,
+      "loss": 0.767,
+      "step": 845
+    },
+    {
+      "epoch": 2.611111111111111,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.2147368421052632e-05,
+      "loss": 0.7446,
+      "step": 846
+    },
+    {
+      "epoch": 2.6141975308641974,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.2136842105263158e-05,
+      "loss": 0.7031,
+      "step": 847
+    },
+    {
+      "epoch": 2.617283950617284,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.2126315789473686e-05,
+      "loss": 0.7926,
+      "step": 848
+    },
+    {
+      "epoch": 2.6203703703703702,
+      "grad_norm": 6.875,
+      "learning_rate": 1.2115789473684212e-05,
+      "loss": 0.752,
+      "step": 849
+    },
+    {
+      "epoch": 2.623456790123457,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 0.7589,
+      "step": 850
+    },
+    {
+      "epoch": 2.626543209876543,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.2094736842105263e-05,
+      "loss": 0.744,
+      "step": 851
+    },
+    {
+      "epoch": 2.6296296296296298,
+      "grad_norm": 12.375,
+      "learning_rate": 1.208421052631579e-05,
+      "loss": 0.7844,
+      "step": 852
+    },
+    {
+      "epoch": 2.632716049382716,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.2073684210526318e-05,
+      "loss": 0.8026,
+      "step": 853
+    },
+    {
+      "epoch": 2.6358024691358026,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.2063157894736844e-05,
+      "loss": 0.7273,
+      "step": 854
+    },
+    {
+      "epoch": 2.638888888888889,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.2052631578947369e-05,
+      "loss": 0.7864,
+      "step": 855
+    },
+    {
+      "epoch": 2.6419753086419755,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.2042105263157895e-05,
+      "loss": 0.7777,
+      "step": 856
+    },
+    {
+      "epoch": 2.6450617283950617,
+      "grad_norm": 6.625,
+      "learning_rate": 1.2031578947368423e-05,
+      "loss": 0.7086,
+      "step": 857
+    },
+    {
+      "epoch": 2.648148148148148,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.202105263157895e-05,
+      "loss": 0.7441,
+      "step": 858
+    },
+    {
+      "epoch": 2.6512345679012346,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.2010526315789474e-05,
+      "loss": 0.7585,
+      "step": 859
+    },
+    {
+      "epoch": 2.6543209876543212,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.2e-05,
+      "loss": 0.7608,
+      "step": 860
+    },
+    {
+      "epoch": 2.6574074074074074,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.1989473684210527e-05,
+      "loss": 0.7929,
+      "step": 861
+    },
+    {
+      "epoch": 2.6604938271604937,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.1978947368421055e-05,
+      "loss": 0.7843,
+      "step": 862
+    },
+    {
+      "epoch": 2.6635802469135803,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.1968421052631581e-05,
+      "loss": 0.7153,
+      "step": 863
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.1957894736842106e-05,
+      "loss": 0.742,
+      "step": 864
+    },
+    {
+      "epoch": 2.669753086419753,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.1947368421052632e-05,
+      "loss": 0.7286,
+      "step": 865
+    },
+    {
+      "epoch": 2.6728395061728394,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.1936842105263158e-05,
+      "loss": 0.7724,
+      "step": 866
+    },
+    {
+      "epoch": 2.675925925925926,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.1926315789473686e-05,
+      "loss": 0.7633,
+      "step": 867
+    },
+    {
+      "epoch": 2.6790123456790123,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.1915789473684211e-05,
+      "loss": 0.7396,
+      "step": 868
+    },
+    {
+      "epoch": 2.682098765432099,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.1905263157894737e-05,
+      "loss": 0.7473,
+      "step": 869
+    },
+    {
+      "epoch": 2.685185185185185,
+      "grad_norm": 8.5,
+      "learning_rate": 1.1894736842105264e-05,
+      "loss": 0.8197,
+      "step": 870
+    },
+    {
+      "epoch": 2.6882716049382713,
+      "grad_norm": 6.4375,
+      "learning_rate": 1.1884210526315792e-05,
+      "loss": 0.7024,
+      "step": 871
+    },
+    {
+      "epoch": 2.691358024691358,
+      "grad_norm": 6.625,
+      "learning_rate": 1.1873684210526318e-05,
+      "loss": 0.7737,
+      "step": 872
+    },
+    {
+      "epoch": 2.6944444444444446,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.1863157894736843e-05,
+      "loss": 0.7495,
+      "step": 873
+    },
+    {
+      "epoch": 2.697530864197531,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.1852631578947369e-05,
+      "loss": 0.8306,
+      "step": 874
+    },
+    {
+      "epoch": 2.700617283950617,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.1842105263157895e-05,
+      "loss": 0.7304,
+      "step": 875
+    },
+    {
+      "epoch": 2.7037037037037037,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.1831578947368423e-05,
+      "loss": 0.6939,
+      "step": 876
+    },
+    {
+      "epoch": 2.7067901234567904,
+      "grad_norm": 8.25,
+      "learning_rate": 1.1821052631578948e-05,
+      "loss": 0.7529,
+      "step": 877
+    },
+    {
+      "epoch": 2.7098765432098766,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.1810526315789474e-05,
+      "loss": 0.7384,
+      "step": 878
+    },
+    {
+      "epoch": 2.712962962962963,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.18e-05,
+      "loss": 0.72,
+      "step": 879
+    },
+    {
+      "epoch": 2.7160493827160495,
+      "grad_norm": 6.625,
+      "learning_rate": 1.1789473684210527e-05,
+      "loss": 0.7208,
+      "step": 880
+    },
+    {
+      "epoch": 2.7191358024691357,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.1778947368421055e-05,
+      "loss": 0.7057,
+      "step": 881
+    },
+    {
+      "epoch": 2.7222222222222223,
+      "grad_norm": 7.375,
+      "learning_rate": 1.176842105263158e-05,
+      "loss": 0.7543,
+      "step": 882
+    },
+    {
+      "epoch": 2.7253086419753085,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.1757894736842106e-05,
+      "loss": 0.6954,
+      "step": 883
+    },
+    {
+      "epoch": 2.728395061728395,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.1747368421052632e-05,
+      "loss": 0.6984,
+      "step": 884
+    },
+    {
+      "epoch": 2.7314814814814814,
+      "grad_norm": 9.25,
+      "learning_rate": 1.173684210526316e-05,
+      "loss": 0.7593,
+      "step": 885
+    },
+    {
+      "epoch": 2.734567901234568,
+      "grad_norm": 8.25,
+      "learning_rate": 1.1726315789473685e-05,
+      "loss": 0.7593,
+      "step": 886
+    },
+    {
+      "epoch": 2.7376543209876543,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.1715789473684211e-05,
+      "loss": 0.7411,
+      "step": 887
+    },
+    {
+      "epoch": 2.7407407407407405,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.1705263157894737e-05,
+      "loss": 0.7464,
+      "step": 888
+    },
+    {
+      "epoch": 2.743827160493827,
+      "grad_norm": 11.25,
+      "learning_rate": 1.1694736842105264e-05,
+      "loss": 0.752,
+      "step": 889
+    },
+    {
+      "epoch": 2.746913580246914,
+      "grad_norm": 10.25,
+      "learning_rate": 1.1684210526315792e-05,
+      "loss": 0.7443,
+      "step": 890
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 7.0,
+      "learning_rate": 1.1673684210526316e-05,
+      "loss": 0.7341,
+      "step": 891
+    },
+    {
+      "epoch": 2.753086419753086,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.1663157894736843e-05,
+      "loss": 0.7671,
+      "step": 892
+    },
+    {
+      "epoch": 2.756172839506173,
+      "grad_norm": 7.375,
+      "learning_rate": 1.1652631578947369e-05,
+      "loss": 0.6923,
+      "step": 893
+    },
+    {
+      "epoch": 2.7592592592592595,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.1642105263157897e-05,
+      "loss": 0.7374,
+      "step": 894
+    },
+    {
+      "epoch": 2.7623456790123457,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.1631578947368423e-05,
+      "loss": 0.7051,
+      "step": 895
+    },
+    {
+      "epoch": 2.765432098765432,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.1621052631578948e-05,
+      "loss": 0.7625,
+      "step": 896
+    },
+    {
+      "epoch": 2.7685185185185186,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.1610526315789474e-05,
+      "loss": 0.7364,
+      "step": 897
+    },
+    {
+      "epoch": 2.771604938271605,
+      "grad_norm": 7.5,
+      "learning_rate": 1.16e-05,
+      "loss": 0.6986,
+      "step": 898
+    },
+    {
+      "epoch": 2.7746913580246915,
+      "grad_norm": 8.125,
+      "learning_rate": 1.1589473684210529e-05,
+      "loss": 0.7037,
+      "step": 899
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.8273,
+      "step": 900
+    },
+    {
+      "epoch": 2.7808641975308643,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.156842105263158e-05,
+      "loss": 0.7849,
+      "step": 901
+    },
+    {
+      "epoch": 2.7839506172839505,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.1557894736842106e-05,
+      "loss": 0.7205,
+      "step": 902
+    },
+    {
+      "epoch": 2.787037037037037,
+      "grad_norm": 13.0,
+      "learning_rate": 1.1547368421052632e-05,
+      "loss": 0.754,
+      "step": 903
+    },
+    {
+      "epoch": 2.7901234567901234,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.153684210526316e-05,
+      "loss": 0.7908,
+      "step": 904
+    },
+    {
+      "epoch": 2.7932098765432096,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.1526315789473685e-05,
+      "loss": 0.7709,
+      "step": 905
+    },
+    {
+      "epoch": 2.7962962962962963,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.1515789473684211e-05,
+      "loss": 0.7769,
+      "step": 906
+    },
+    {
+      "epoch": 2.799382716049383,
+      "grad_norm": 9.0,
+      "learning_rate": 1.1505263157894738e-05,
+      "loss": 0.8472,
+      "step": 907
+    },
+    {
+      "epoch": 2.802469135802469,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.1494736842105266e-05,
+      "loss": 0.755,
+      "step": 908
+    },
+    {
+      "epoch": 2.8055555555555554,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.148421052631579e-05,
+      "loss": 0.7823,
+      "step": 909
+    },
+    {
+      "epoch": 2.808641975308642,
+      "grad_norm": 9.625,
+      "learning_rate": 1.1473684210526317e-05,
+      "loss": 0.7585,
+      "step": 910
+    },
+    {
+      "epoch": 2.8117283950617287,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.1463157894736843e-05,
+      "loss": 0.7545,
+      "step": 911
+    },
+    {
+      "epoch": 2.814814814814815,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.145263157894737e-05,
+      "loss": 0.6885,
+      "step": 912
+    },
+    {
+      "epoch": 2.817901234567901,
+      "grad_norm": 6.375,
+      "learning_rate": 1.1442105263157897e-05,
+      "loss": 0.7247,
+      "step": 913
+    },
+    {
+      "epoch": 2.8209876543209877,
+      "grad_norm": 8.25,
+      "learning_rate": 1.1431578947368422e-05,
+      "loss": 0.7886,
+      "step": 914
+    },
+    {
+      "epoch": 2.824074074074074,
+      "grad_norm": 7.625,
+      "learning_rate": 1.1421052631578948e-05,
+      "loss": 0.6996,
+      "step": 915
+    },
+    {
+      "epoch": 2.8271604938271606,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.1410526315789475e-05,
+      "loss": 0.7617,
+      "step": 916
+    },
+    {
+      "epoch": 2.830246913580247,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.14e-05,
+      "loss": 0.6916,
+      "step": 917
+    },
+    {
+      "epoch": 2.8333333333333335,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1389473684210527e-05,
+      "loss": 0.7514,
+      "step": 918
+    },
+    {
+      "epoch": 2.8364197530864197,
+      "grad_norm": 8.875,
+      "learning_rate": 1.1378947368421054e-05,
+      "loss": 0.801,
+      "step": 919
+    },
+    {
+      "epoch": 2.8395061728395063,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 0.8006,
+      "step": 920
+    },
+    {
+      "epoch": 2.8425925925925926,
+      "grad_norm": 6.625,
+      "learning_rate": 1.1357894736842106e-05,
+      "loss": 0.771,
+      "step": 921
+    },
+    {
+      "epoch": 2.8456790123456788,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.1347368421052634e-05,
+      "loss": 0.7508,
+      "step": 922
+    },
+    {
+      "epoch": 2.8487654320987654,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.1336842105263159e-05,
+      "loss": 0.7466,
+      "step": 923
+    },
+    {
+      "epoch": 2.851851851851852,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.1326315789473685e-05,
+      "loss": 0.7557,
+      "step": 924
+    },
+    {
+      "epoch": 2.8549382716049383,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.1315789473684212e-05,
+      "loss": 0.7397,
+      "step": 925
+    },
+    {
+      "epoch": 2.8580246913580245,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.1305263157894736e-05,
+      "loss": 0.7169,
+      "step": 926
+    },
+    {
+      "epoch": 2.861111111111111,
+      "grad_norm": 6.875,
+      "learning_rate": 1.1294736842105264e-05,
+      "loss": 0.7441,
+      "step": 927
+    },
+    {
+      "epoch": 2.8641975308641974,
+      "grad_norm": 9.125,
+      "learning_rate": 1.128421052631579e-05,
+      "loss": 0.7418,
+      "step": 928
+    },
+    {
+      "epoch": 2.867283950617284,
+      "grad_norm": 7.625,
+      "learning_rate": 1.1273684210526317e-05,
+      "loss": 0.7805,
+      "step": 929
+    },
+    {
+      "epoch": 2.8703703703703702,
+      "grad_norm": 7.625,
+      "learning_rate": 1.1263157894736843e-05,
+      "loss": 0.754,
+      "step": 930
+    },
+    {
+      "epoch": 2.873456790123457,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.1252631578947368e-05,
+      "loss": 0.7643,
+      "step": 931
+    },
+    {
+      "epoch": 2.876543209876543,
+      "grad_norm": 8.0,
+      "learning_rate": 1.1242105263157896e-05,
+      "loss": 0.7507,
+      "step": 932
+    },
+    {
+      "epoch": 2.8796296296296298,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.1231578947368422e-05,
+      "loss": 0.7924,
+      "step": 933
+    },
+    {
+      "epoch": 2.882716049382716,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.1221052631578949e-05,
+      "loss": 0.737,
+      "step": 934
+    },
+    {
+      "epoch": 2.8858024691358026,
+      "grad_norm": 69.0,
+      "learning_rate": 1.1210526315789473e-05,
+      "loss": 0.7353,
+      "step": 935
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.7761,
+      "step": 936
+    },
+    {
+      "epoch": 2.8919753086419755,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.1189473684210528e-05,
+      "loss": 0.7427,
+      "step": 937
+    },
+    {
+      "epoch": 2.8950617283950617,
+      "grad_norm": 8.125,
+      "learning_rate": 1.1178947368421054e-05,
+      "loss": 0.7443,
+      "step": 938
+    },
+    {
+      "epoch": 2.898148148148148,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.116842105263158e-05,
+      "loss": 0.768,
+      "step": 939
+    },
+    {
+      "epoch": 2.9012345679012346,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 0.7248,
+      "step": 940
+    },
+    {
+      "epoch": 2.9043209876543212,
+      "grad_norm": 7.875,
+      "learning_rate": 1.1147368421052633e-05,
+      "loss": 0.7426,
+      "step": 941
+    },
+    {
+      "epoch": 2.9074074074074074,
+      "grad_norm": 8.625,
+      "learning_rate": 1.1136842105263159e-05,
+      "loss": 0.7526,
+      "step": 942
+    },
+    {
+      "epoch": 2.9104938271604937,
+      "grad_norm": 8.25,
+      "learning_rate": 1.1126315789473685e-05,
+      "loss": 0.7615,
+      "step": 943
+    },
+    {
+      "epoch": 2.9135802469135803,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.111578947368421e-05,
+      "loss": 0.6982,
+      "step": 944
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.1105263157894736e-05,
+      "loss": 0.7399,
+      "step": 945
+    },
+    {
+      "epoch": 2.919753086419753,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.1094736842105264e-05,
+      "loss": 0.7344,
+      "step": 946
+    },
+    {
+      "epoch": 2.9228395061728394,
+      "grad_norm": 26.25,
+      "learning_rate": 1.108421052631579e-05,
+      "loss": 0.8505,
+      "step": 947
+    },
+    {
+      "epoch": 2.925925925925926,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.1073684210526317e-05,
+      "loss": 0.7986,
+      "step": 948
+    },
+    {
+      "epoch": 2.9290123456790123,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.1063157894736842e-05,
+      "loss": 0.7263,
+      "step": 949
+    },
+    {
+      "epoch": 2.932098765432099,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.105263157894737e-05,
+      "loss": 0.7041,
+      "step": 950
+    },
+    {
+      "epoch": 2.935185185185185,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.1042105263157896e-05,
+      "loss": 0.7821,
+      "step": 951
+    },
+    {
+      "epoch": 2.9382716049382713,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.1031578947368422e-05,
+      "loss": 0.8303,
+      "step": 952
+    },
+    {
+      "epoch": 2.941358024691358,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.1021052631578947e-05,
+      "loss": 0.7841,
+      "step": 953
+    },
+    {
+      "epoch": 2.9444444444444446,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.1010526315789473e-05,
+      "loss": 0.7336,
+      "step": 954
+    },
+    {
+      "epoch": 2.947530864197531,
+      "grad_norm": 8.625,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7521,
+      "step": 955
+    },
+    {
+      "epoch": 2.950617283950617,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.0989473684210528e-05,
+      "loss": 0.7358,
+      "step": 956
+    },
+    {
+      "epoch": 2.9537037037037037,
+      "grad_norm": 7.375,
+      "learning_rate": 1.0978947368421054e-05,
+      "loss": 0.7195,
+      "step": 957
+    },
+    {
+      "epoch": 2.9567901234567904,
+      "grad_norm": 10.125,
+      "learning_rate": 1.0968421052631579e-05,
+      "loss": 0.7679,
+      "step": 958
+    },
+    {
+      "epoch": 2.9598765432098766,
+      "grad_norm": 8.125,
+      "learning_rate": 1.0957894736842105e-05,
+      "loss": 0.7229,
+      "step": 959
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.0947368421052633e-05,
+      "loss": 0.7054,
+      "step": 960
+    },
+    {
+      "epoch": 2.9660493827160495,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.093684210526316e-05,
+      "loss": 0.7349,
+      "step": 961
+    },
+    {
+      "epoch": 2.9691358024691357,
+      "grad_norm": 8.125,
+      "learning_rate": 1.0926315789473686e-05,
+      "loss": 0.6981,
+      "step": 962
+    },
+    {
+      "epoch": 2.9722222222222223,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.091578947368421e-05,
+      "loss": 0.7151,
+      "step": 963
+    },
+    {
+      "epoch": 2.9753086419753085,
+      "grad_norm": 8.375,
+      "learning_rate": 1.0905263157894738e-05,
+      "loss": 0.7618,
+      "step": 964
+    },
+    {
+      "epoch": 2.978395061728395,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.0894736842105265e-05,
+      "loss": 0.7193,
+      "step": 965
+    },
+    {
+      "epoch": 2.9814814814814814,
+      "grad_norm": 6.1875,
+      "learning_rate": 1.0884210526315791e-05,
+      "loss": 0.7222,
+      "step": 966
+    },
+    {
+      "epoch": 2.984567901234568,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.0873684210526316e-05,
+      "loss": 0.7617,
+      "step": 967
+    },
+    {
+      "epoch": 2.9876543209876543,
+      "grad_norm": 7.625,
+      "learning_rate": 1.0863157894736842e-05,
+      "loss": 0.7197,
+      "step": 968
+    },
+    {
+      "epoch": 2.9907407407407405,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.085263157894737e-05,
+      "loss": 0.7842,
+      "step": 969
+    },
+    {
+      "epoch": 2.993827160493827,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.0842105263157896e-05,
+      "loss": 0.802,
+      "step": 970
+    },
+    {
+      "epoch": 2.996913580246914,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.0831578947368423e-05,
+      "loss": 0.7569,
+      "step": 971
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 7.625,
+      "learning_rate": 1.0821052631578947e-05,
+      "loss": 0.6535,
+      "step": 972
+    },
+    {
+      "epoch": 3.003086419753086,
+      "grad_norm": 9.125,
+      "learning_rate": 1.0810526315789474e-05,
+      "loss": 0.6935,
+      "step": 973
+    },
+    {
+      "epoch": 3.006172839506173,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.7212,
+      "step": 974
+    },
+    {
+      "epoch": 3.009259259259259,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.0789473684210528e-05,
+      "loss": 0.73,
+      "step": 975
+    },
+    {
+      "epoch": 3.0123456790123457,
+      "grad_norm": 7.375,
+      "learning_rate": 1.0778947368421053e-05,
+      "loss": 0.742,
+      "step": 976
+    },
+    {
+      "epoch": 3.015432098765432,
+      "grad_norm": 6.15625,
+      "learning_rate": 1.0768421052631579e-05,
+      "loss": 0.6877,
+      "step": 977
+    },
+    {
+      "epoch": 3.0185185185185186,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.0757894736842107e-05,
+      "loss": 0.7165,
+      "step": 978
+    },
+    {
+      "epoch": 3.021604938271605,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.0747368421052633e-05,
+      "loss": 0.7308,
+      "step": 979
+    },
+    {
+      "epoch": 3.0246913580246915,
+      "grad_norm": 7.25,
+      "learning_rate": 1.073684210526316e-05,
+      "loss": 0.7548,
+      "step": 980
+    },
+    {
+      "epoch": 3.0277777777777777,
+      "grad_norm": 6.75,
+      "learning_rate": 1.0726315789473684e-05,
+      "loss": 0.7699,
+      "step": 981
+    },
+    {
+      "epoch": 3.0308641975308643,
+      "grad_norm": 7.25,
+      "learning_rate": 1.071578947368421e-05,
+      "loss": 0.749,
+      "step": 982
+    },
+    {
+      "epoch": 3.0339506172839505,
+      "grad_norm": 6.5,
+      "learning_rate": 1.0705263157894739e-05,
+      "loss": 0.7434,
+      "step": 983
+    },
+    {
+      "epoch": 3.037037037037037,
+      "grad_norm": 13.125,
+      "learning_rate": 1.0694736842105265e-05,
+      "loss": 0.7266,
+      "step": 984
+    },
+    {
+      "epoch": 3.0401234567901234,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.068421052631579e-05,
+      "loss": 0.7443,
+      "step": 985
+    },
+    {
+      "epoch": 3.04320987654321,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.0673684210526316e-05,
+      "loss": 0.7166,
+      "step": 986
+    },
+    {
+      "epoch": 3.0462962962962963,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.0663157894736842e-05,
+      "loss": 0.7329,
+      "step": 987
+    },
+    {
+      "epoch": 3.049382716049383,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.065263157894737e-05,
+      "loss": 0.716,
+      "step": 988
+    },
+    {
+      "epoch": 3.052469135802469,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.0642105263157897e-05,
+      "loss": 0.7075,
+      "step": 989
+    },
+    {
+      "epoch": 3.0555555555555554,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.0631578947368421e-05,
+      "loss": 0.7371,
+      "step": 990
+    },
+    {
+      "epoch": 3.058641975308642,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.0621052631578948e-05,
+      "loss": 0.7537,
+      "step": 991
+    },
+    {
+      "epoch": 3.0617283950617282,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.0610526315789476e-05,
+      "loss": 0.7207,
+      "step": 992
+    },
+    {
+      "epoch": 3.064814814814815,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 0.7298,
+      "step": 993
+    },
+    {
+      "epoch": 3.067901234567901,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0589473684210526e-05,
+      "loss": 0.6623,
+      "step": 994
+    },
+    {
+      "epoch": 3.0709876543209877,
+      "grad_norm": 6.375,
+      "learning_rate": 1.0578947368421053e-05,
+      "loss": 0.7073,
+      "step": 995
+    },
+    {
+      "epoch": 3.074074074074074,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.0568421052631579e-05,
+      "loss": 0.7311,
+      "step": 996
+    },
+    {
+      "epoch": 3.0771604938271606,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.0557894736842107e-05,
+      "loss": 0.7899,
+      "step": 997
+    },
+    {
+      "epoch": 3.080246913580247,
+      "grad_norm": 7.875,
+      "learning_rate": 1.0547368421052633e-05,
+      "loss": 0.7201,
+      "step": 998
+    },
+    {
+      "epoch": 3.0833333333333335,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.0536842105263158e-05,
+      "loss": 0.7124,
+      "step": 999
+    },
+    {
+      "epoch": 3.0864197530864197,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.7173,
+      "step": 1000
+    },
+    {
+      "epoch": 3.0895061728395063,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.051578947368421e-05,
+      "loss": 0.7177,
+      "step": 1001
+    },
+    {
+      "epoch": 3.0925925925925926,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.0505263157894739e-05,
+      "loss": 0.6885,
+      "step": 1002
+    },
+    {
+      "epoch": 3.095679012345679,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.0494736842105263e-05,
+      "loss": 0.7459,
+      "step": 1003
+    },
+    {
+      "epoch": 3.0987654320987654,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.048421052631579e-05,
+      "loss": 0.7358,
+      "step": 1004
+    },
+    {
+      "epoch": 3.1018518518518516,
+      "grad_norm": 10.25,
+      "learning_rate": 1.0473684210526316e-05,
+      "loss": 0.7635,
+      "step": 1005
+    },
+    {
+      "epoch": 3.1049382716049383,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.0463157894736844e-05,
+      "loss": 0.7203,
+      "step": 1006
+    },
+    {
+      "epoch": 3.1080246913580245,
+      "grad_norm": 9.0,
+      "learning_rate": 1.045263157894737e-05,
+      "loss": 0.7472,
+      "step": 1007
+    },
+    {
+      "epoch": 3.111111111111111,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.0442105263157895e-05,
+      "loss": 0.68,
+      "step": 1008
+    },
+    {
+      "epoch": 3.1141975308641974,
+      "grad_norm": 6.65625,
+      "learning_rate": 1.0431578947368421e-05,
+      "loss": 0.6978,
+      "step": 1009
+    },
+    {
+      "epoch": 3.117283950617284,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.0421052631578948e-05,
+      "loss": 0.7551,
+      "step": 1010
+    },
+    {
+      "epoch": 3.1203703703703702,
+      "grad_norm": 7.25,
+      "learning_rate": 1.0410526315789476e-05,
+      "loss": 0.6931,
+      "step": 1011
+    },
+    {
+      "epoch": 3.123456790123457,
+      "grad_norm": 6.28125,
+      "learning_rate": 1.04e-05,
+      "loss": 0.7474,
+      "step": 1012
+    },
+    {
+      "epoch": 3.126543209876543,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.0389473684210527e-05,
+      "loss": 0.7201,
+      "step": 1013
+    },
+    {
+      "epoch": 3.1296296296296298,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.0378947368421053e-05,
+      "loss": 0.7206,
+      "step": 1014
+    },
+    {
+      "epoch": 3.132716049382716,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.036842105263158e-05,
+      "loss": 0.7309,
+      "step": 1015
+    },
+    {
+      "epoch": 3.1358024691358026,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.0357894736842107e-05,
+      "loss": 0.6859,
+      "step": 1016
+    },
+    {
+      "epoch": 3.138888888888889,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.0347368421052632e-05,
+      "loss": 0.7103,
+      "step": 1017
+    },
+    {
+      "epoch": 3.1419753086419755,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.0336842105263158e-05,
+      "loss": 0.8025,
+      "step": 1018
+    },
+    {
+      "epoch": 3.1450617283950617,
+      "grad_norm": 6.34375,
+      "learning_rate": 1.0326315789473685e-05,
+      "loss": 0.6941,
+      "step": 1019
+    },
+    {
+      "epoch": 3.148148148148148,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.0315789473684213e-05,
+      "loss": 0.6646,
+      "step": 1020
+    },
+    {
+      "epoch": 3.1512345679012346,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.0305263157894739e-05,
+      "loss": 0.6843,
+      "step": 1021
+    },
+    {
+      "epoch": 3.154320987654321,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.0294736842105264e-05,
+      "loss": 0.7803,
+      "step": 1022
+    },
+    {
+      "epoch": 3.1574074074074074,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.028421052631579e-05,
+      "loss": 0.7014,
+      "step": 1023
+    },
+    {
+      "epoch": 3.1604938271604937,
+      "grad_norm": 7.25,
+      "learning_rate": 1.0273684210526316e-05,
+      "loss": 0.6515,
+      "step": 1024
+    },
+    {
+      "epoch": 3.1635802469135803,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.0263157894736844e-05,
+      "loss": 0.7804,
+      "step": 1025
+    },
+    {
+      "epoch": 3.1666666666666665,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.0252631578947369e-05,
+      "loss": 0.7017,
+      "step": 1026
+    },
+    {
+      "epoch": 3.169753086419753,
+      "grad_norm": 8.25,
+      "learning_rate": 1.0242105263157895e-05,
+      "loss": 0.7616,
+      "step": 1027
+    },
+    {
+      "epoch": 3.1728395061728394,
+      "grad_norm": 8.0,
+      "learning_rate": 1.0231578947368422e-05,
+      "loss": 0.7366,
+      "step": 1028
+    },
+    {
+      "epoch": 3.175925925925926,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.0221052631578948e-05,
+      "loss": 0.7233,
+      "step": 1029
+    },
+    {
+      "epoch": 3.1790123456790123,
+      "grad_norm": 8.25,
+      "learning_rate": 1.0210526315789476e-05,
+      "loss": 0.7689,
+      "step": 1030
+    },
+    {
+      "epoch": 3.182098765432099,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.02e-05,
+      "loss": 0.6472,
+      "step": 1031
+    },
+    {
+      "epoch": 3.185185185185185,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.0189473684210527e-05,
+      "loss": 0.7301,
+      "step": 1032
+    },
+    {
+      "epoch": 3.1882716049382718,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.0178947368421053e-05,
+      "loss": 0.6875,
+      "step": 1033
+    },
+    {
+      "epoch": 3.191358024691358,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.0168421052631581e-05,
+      "loss": 0.6688,
+      "step": 1034
+    },
+    {
+      "epoch": 3.1944444444444446,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.0157894736842106e-05,
+      "loss": 0.7293,
+      "step": 1035
+    },
+    {
+      "epoch": 3.197530864197531,
+      "grad_norm": 6.625,
+      "learning_rate": 1.0147368421052632e-05,
+      "loss": 0.6749,
+      "step": 1036
+    },
+    {
+      "epoch": 3.200617283950617,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.0136842105263159e-05,
+      "loss": 0.716,
+      "step": 1037
+    },
+    {
+      "epoch": 3.2037037037037037,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.0126315789473685e-05,
+      "loss": 0.7348,
+      "step": 1038
+    },
+    {
+      "epoch": 3.20679012345679,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0115789473684213e-05,
+      "loss": 0.7011,
+      "step": 1039
+    },
+    {
+      "epoch": 3.2098765432098766,
+      "grad_norm": 6.15625,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 0.7047,
+      "step": 1040
+    },
+    {
+      "epoch": 3.212962962962963,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.0094736842105264e-05,
+      "loss": 0.7723,
+      "step": 1041
+    },
+    {
+      "epoch": 3.2160493827160495,
+      "grad_norm": 12.375,
+      "learning_rate": 1.008421052631579e-05,
+      "loss": 0.6801,
+      "step": 1042
+    },
+    {
+      "epoch": 3.2191358024691357,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.0073684210526315e-05,
+      "loss": 0.725,
+      "step": 1043
+    },
+    {
+      "epoch": 3.2222222222222223,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.0063157894736843e-05,
+      "loss": 0.7116,
+      "step": 1044
+    },
+    {
+      "epoch": 3.2253086419753085,
+      "grad_norm": 8.75,
+      "learning_rate": 1.005263157894737e-05,
+      "loss": 0.697,
+      "step": 1045
+    },
+    {
+      "epoch": 3.228395061728395,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.0042105263157896e-05,
+      "loss": 0.6643,
+      "step": 1046
+    },
+    {
+      "epoch": 3.2314814814814814,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.0031578947368422e-05,
+      "loss": 0.6884,
+      "step": 1047
+    },
+    {
+      "epoch": 3.234567901234568,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.002105263157895e-05,
+      "loss": 0.7483,
+      "step": 1048
+    },
+    {
+      "epoch": 3.2376543209876543,
+      "grad_norm": 7.875,
+      "learning_rate": 1.0010526315789474e-05,
+      "loss": 0.7782,
+      "step": 1049
+    },
+    {
+      "epoch": 3.240740740740741,
+      "grad_norm": 8.125,
+      "learning_rate": 1e-05,
+      "loss": 0.6839,
+      "step": 1050
+    },
+    {
+      "epoch": 3.243827160493827,
+      "grad_norm": 7.0625,
+      "learning_rate": 9.989473684210527e-06,
+      "loss": 0.7171,
+      "step": 1051
+    },
+    {
+      "epoch": 3.246913580246914,
+      "grad_norm": 9.625,
+      "learning_rate": 9.978947368421053e-06,
+      "loss": 0.6894,
+      "step": 1052
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 8.9375,
+      "learning_rate": 9.96842105263158e-06,
+      "loss": 0.7639,
+      "step": 1053
+    },
+    {
+      "epoch": 3.253086419753086,
+      "grad_norm": 7.5,
+      "learning_rate": 9.957894736842106e-06,
+      "loss": 0.7097,
+      "step": 1054
+    },
+    {
+      "epoch": 3.256172839506173,
+      "grad_norm": 6.59375,
+      "learning_rate": 9.947368421052632e-06,
+      "loss": 0.7479,
+      "step": 1055
+    },
+    {
+      "epoch": 3.259259259259259,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.936842105263159e-06,
+      "loss": 0.7019,
+      "step": 1056
+    },
+    {
+      "epoch": 3.2623456790123457,
+      "grad_norm": 9.5625,
+      "learning_rate": 9.926315789473685e-06,
+      "loss": 0.669,
+      "step": 1057
+    },
+    {
+      "epoch": 3.265432098765432,
+      "grad_norm": 7.3125,
+      "learning_rate": 9.915789473684211e-06,
+      "loss": 0.7544,
+      "step": 1058
+    },
+    {
+      "epoch": 3.2685185185185186,
+      "grad_norm": 13.25,
+      "learning_rate": 9.905263157894738e-06,
+      "loss": 0.7141,
+      "step": 1059
+    },
+    {
+      "epoch": 3.271604938271605,
+      "grad_norm": 16.75,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 0.7053,
+      "step": 1060
+    },
+    {
+      "epoch": 3.2746913580246915,
+      "grad_norm": 7.0,
+      "learning_rate": 9.88421052631579e-06,
+      "loss": 0.6498,
+      "step": 1061
+    },
+    {
+      "epoch": 3.2777777777777777,
+      "grad_norm": 6.84375,
+      "learning_rate": 9.873684210526317e-06,
+      "loss": 0.716,
+      "step": 1062
+    },
+    {
+      "epoch": 3.2808641975308643,
+      "grad_norm": 6.9375,
+      "learning_rate": 9.863157894736843e-06,
+      "loss": 0.7315,
+      "step": 1063
+    },
+    {
+      "epoch": 3.2839506172839505,
+      "grad_norm": 7.59375,
+      "learning_rate": 9.85263157894737e-06,
+      "loss": 0.6977,
+      "step": 1064
+    },
+    {
+      "epoch": 3.287037037037037,
+      "grad_norm": 8.0625,
+      "learning_rate": 9.842105263157896e-06,
+      "loss": 0.7265,
+      "step": 1065
+    },
+    {
+      "epoch": 3.2901234567901234,
+      "grad_norm": 7.125,
+      "learning_rate": 9.831578947368422e-06,
+      "loss": 0.7586,
+      "step": 1066
+    },
+    {
+      "epoch": 3.29320987654321,
+      "grad_norm": 7.5,
+      "learning_rate": 9.821052631578948e-06,
+      "loss": 0.6873,
+      "step": 1067
+    },
+    {
+      "epoch": 3.2962962962962963,
+      "grad_norm": 7.09375,
+      "learning_rate": 9.810526315789475e-06,
+      "loss": 0.6615,
+      "step": 1068
+    },
+    {
+      "epoch": 3.299382716049383,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.6678,
+      "step": 1069
+    },
+    {
+      "epoch": 3.302469135802469,
+      "grad_norm": 7.5,
+      "learning_rate": 9.789473684210527e-06,
+      "loss": 0.6877,
+      "step": 1070
+    },
+    {
+      "epoch": 3.3055555555555554,
+      "grad_norm": 6.75,
+      "learning_rate": 9.778947368421054e-06,
+      "loss": 0.6991,
+      "step": 1071
+    },
+    {
+      "epoch": 3.308641975308642,
+      "grad_norm": 7.46875,
+      "learning_rate": 9.76842105263158e-06,
+      "loss": 0.7197,
+      "step": 1072
+    },
+    {
+      "epoch": 3.3117283950617282,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.757894736842106e-06,
+      "loss": 0.7414,
+      "step": 1073
+    },
+    {
+      "epoch": 3.314814814814815,
+      "grad_norm": 8.9375,
+      "learning_rate": 9.747368421052633e-06,
+      "loss": 0.7213,
+      "step": 1074
+    },
+    {
+      "epoch": 3.317901234567901,
+      "grad_norm": 8.375,
+      "learning_rate": 9.736842105263159e-06,
+      "loss": 0.6993,
+      "step": 1075
+    },
+    {
+      "epoch": 3.3209876543209877,
+      "grad_norm": 7.53125,
+      "learning_rate": 9.726315789473685e-06,
+      "loss": 0.7429,
+      "step": 1076
+    },
+    {
+      "epoch": 3.324074074074074,
+      "grad_norm": 7.0625,
+      "learning_rate": 9.715789473684212e-06,
+      "loss": 0.753,
+      "step": 1077
+    },
+    {
+      "epoch": 3.3271604938271606,
+      "grad_norm": 8.125,
+      "learning_rate": 9.705263157894738e-06,
+      "loss": 0.669,
+      "step": 1078
+    },
+    {
+      "epoch": 3.330246913580247,
+      "grad_norm": 6.9375,
+      "learning_rate": 9.694736842105263e-06,
+      "loss": 0.6943,
+      "step": 1079
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 0.6512,
+      "step": 1080
+    },
+    {
+      "epoch": 3.3364197530864197,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.673684210526317e-06,
+      "loss": 0.7352,
+      "step": 1081
+    },
+    {
+      "epoch": 3.3395061728395063,
+      "grad_norm": 6.96875,
+      "learning_rate": 9.663157894736843e-06,
+      "loss": 0.6943,
+      "step": 1082
+    },
+    {
+      "epoch": 3.3425925925925926,
+      "grad_norm": 6.625,
+      "learning_rate": 9.65263157894737e-06,
+      "loss": 0.7372,
+      "step": 1083
+    },
+    {
+      "epoch": 3.3456790123456788,
+      "grad_norm": 10.25,
+      "learning_rate": 9.642105263157896e-06,
+      "loss": 0.6927,
+      "step": 1084
+    },
+    {
+      "epoch": 3.3487654320987654,
+      "grad_norm": 7.1875,
+      "learning_rate": 9.631578947368422e-06,
+      "loss": 0.7422,
+      "step": 1085
+    },
+    {
+      "epoch": 3.351851851851852,
+      "grad_norm": 10.9375,
+      "learning_rate": 9.621052631578947e-06,
+      "loss": 0.716,
+      "step": 1086
+    },
+    {
+      "epoch": 3.3549382716049383,
+      "grad_norm": 118.0,
+      "learning_rate": 9.610526315789475e-06,
+      "loss": 0.7298,
+      "step": 1087
+    },
+    {
+      "epoch": 3.3580246913580245,
+      "grad_norm": 8.3125,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.7297,
+      "step": 1088
+    },
+    {
+      "epoch": 3.361111111111111,
+      "grad_norm": 6.59375,
+      "learning_rate": 9.589473684210528e-06,
+      "loss": 0.6901,
+      "step": 1089
+    },
+    {
+      "epoch": 3.3641975308641974,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.578947368421054e-06,
+      "loss": 0.7199,
+      "step": 1090
+    },
+    {
+      "epoch": 3.367283950617284,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.56842105263158e-06,
+      "loss": 0.7512,
+      "step": 1091
+    },
+    {
+      "epoch": 3.3703703703703702,
+      "grad_norm": 8.75,
+      "learning_rate": 9.557894736842107e-06,
+      "loss": 0.7179,
+      "step": 1092
+    },
+    {
+      "epoch": 3.373456790123457,
+      "grad_norm": 7.625,
+      "learning_rate": 9.547368421052631e-06,
+      "loss": 0.724,
+      "step": 1093
+    },
+    {
+      "epoch": 3.376543209876543,
+      "grad_norm": 8.8125,
+      "learning_rate": 9.53684210526316e-06,
+      "loss": 0.7245,
+      "step": 1094
+    },
+    {
+      "epoch": 3.3796296296296298,
+      "grad_norm": 7.09375,
+      "learning_rate": 9.526315789473684e-06,
+      "loss": 0.731,
+      "step": 1095
+    },
+    {
+      "epoch": 3.382716049382716,
+      "grad_norm": 22.75,
+      "learning_rate": 9.515789473684212e-06,
+      "loss": 0.7083,
+      "step": 1096
+    },
+    {
+      "epoch": 3.3858024691358026,
+      "grad_norm": 7.4375,
+      "learning_rate": 9.505263157894738e-06,
+      "loss": 0.7367,
+      "step": 1097
+    },
+    {
+      "epoch": 3.388888888888889,
+      "grad_norm": 8.0625,
+      "learning_rate": 9.494736842105265e-06,
+      "loss": 0.7166,
+      "step": 1098
+    },
+    {
+      "epoch": 3.3919753086419755,
+      "grad_norm": 8.3125,
+      "learning_rate": 9.484210526315791e-06,
+      "loss": 0.7515,
+      "step": 1099
+    },
+    {
+      "epoch": 3.3950617283950617,
+      "grad_norm": 7.59375,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.6909,
+      "step": 1100
+    },
+    {
+      "epoch": 3.398148148148148,
+      "grad_norm": 9.9375,
+      "learning_rate": 9.463157894736844e-06,
+      "loss": 0.7411,
+      "step": 1101
+    },
+    {
+      "epoch": 3.4012345679012346,
+      "grad_norm": 6.5,
+      "learning_rate": 9.452631578947368e-06,
+      "loss": 0.6988,
+      "step": 1102
+    },
+    {
+      "epoch": 3.4043209876543212,
+      "grad_norm": 8.625,
+      "learning_rate": 9.442105263157896e-06,
+      "loss": 0.7254,
+      "step": 1103
+    },
+    {
+      "epoch": 3.4074074074074074,
+      "grad_norm": 7.25,
+      "learning_rate": 9.43157894736842e-06,
+      "loss": 0.6977,
+      "step": 1104
+    },
+    {
+      "epoch": 3.4104938271604937,
+      "grad_norm": 8.5625,
+      "learning_rate": 9.421052631578949e-06,
+      "loss": 0.7307,
+      "step": 1105
+    },
+    {
+      "epoch": 3.4135802469135803,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.410526315789475e-06,
+      "loss": 0.6783,
+      "step": 1106
+    },
+    {
+      "epoch": 3.4166666666666665,
+      "grad_norm": 10.4375,
+      "learning_rate": 9.4e-06,
+      "loss": 0.7169,
+      "step": 1107
+    },
+    {
+      "epoch": 3.419753086419753,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.389473684210528e-06,
+      "loss": 0.7179,
+      "step": 1108
+    },
+    {
+      "epoch": 3.4228395061728394,
+      "grad_norm": 8.125,
+      "learning_rate": 9.378947368421052e-06,
+      "loss": 0.7781,
+      "step": 1109
+    },
+    {
+      "epoch": 3.425925925925926,
+      "grad_norm": 8.125,
+      "learning_rate": 9.36842105263158e-06,
+      "loss": 0.7025,
+      "step": 1110
+    },
+    {
+      "epoch": 3.4290123456790123,
+      "grad_norm": 7.65625,
+      "learning_rate": 9.357894736842105e-06,
+      "loss": 0.7231,
+      "step": 1111
+    },
+    {
+      "epoch": 3.432098765432099,
+      "grad_norm": 6.21875,
+      "learning_rate": 9.347368421052633e-06,
+      "loss": 0.6908,
+      "step": 1112
+    },
+    {
+      "epoch": 3.435185185185185,
+      "grad_norm": 7.59375,
+      "learning_rate": 9.336842105263158e-06,
+      "loss": 0.6742,
+      "step": 1113
+    },
+    {
+      "epoch": 3.4382716049382718,
+      "grad_norm": 6.59375,
+      "learning_rate": 9.326315789473684e-06,
+      "loss": 0.7291,
+      "step": 1114
+    },
+    {
+      "epoch": 3.441358024691358,
+      "grad_norm": 10.125,
+      "learning_rate": 9.315789473684212e-06,
+      "loss": 0.7184,
+      "step": 1115
+    },
+    {
+      "epoch": 3.4444444444444446,
+      "grad_norm": 9.8125,
+      "learning_rate": 9.305263157894737e-06,
+      "loss": 0.702,
+      "step": 1116
+    },
+    {
+      "epoch": 3.447530864197531,
+      "grad_norm": 9.1875,
+      "learning_rate": 9.294736842105265e-06,
+      "loss": 0.7287,
+      "step": 1117
+    },
+    {
+      "epoch": 3.450617283950617,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.28421052631579e-06,
+      "loss": 0.6986,
+      "step": 1118
+    },
+    {
+      "epoch": 3.4537037037037037,
+      "grad_norm": 10.5625,
+      "learning_rate": 9.273684210526317e-06,
+      "loss": 0.6298,
+      "step": 1119
+    },
+    {
+      "epoch": 3.45679012345679,
+      "grad_norm": 8.1875,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.7519,
+      "step": 1120
+    },
+    {
+      "epoch": 3.4598765432098766,
+      "grad_norm": 7.75,
+      "learning_rate": 9.252631578947368e-06,
+      "loss": 0.7005,
+      "step": 1121
+    },
+    {
+      "epoch": 3.462962962962963,
+      "grad_norm": 7.375,
+      "learning_rate": 9.242105263157896e-06,
+      "loss": 0.6865,
+      "step": 1122
+    },
+    {
+      "epoch": 3.4660493827160495,
+      "grad_norm": 7.6875,
+      "learning_rate": 9.231578947368421e-06,
+      "loss": 0.7539,
+      "step": 1123
+    },
+    {
+      "epoch": 3.4691358024691357,
+      "grad_norm": 7.125,
+      "learning_rate": 9.221052631578949e-06,
+      "loss": 0.6785,
+      "step": 1124
+    },
+    {
+      "epoch": 3.4722222222222223,
+      "grad_norm": 7.71875,
+      "learning_rate": 9.210526315789474e-06,
+      "loss": 0.6888,
+      "step": 1125
+    },
+    {
+      "epoch": 3.4753086419753085,
+      "grad_norm": 6.625,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.6601,
+      "step": 1126
+    },
+    {
+      "epoch": 3.478395061728395,
+      "grad_norm": 9.0,
+      "learning_rate": 9.189473684210526e-06,
+      "loss": 0.7612,
+      "step": 1127
+    },
+    {
+      "epoch": 3.4814814814814814,
+      "grad_norm": 7.15625,
+      "learning_rate": 9.178947368421053e-06,
+      "loss": 0.7008,
+      "step": 1128
+    },
+    {
+      "epoch": 3.484567901234568,
+      "grad_norm": 7.09375,
+      "learning_rate": 9.168421052631579e-06,
+      "loss": 0.7153,
+      "step": 1129
+    },
+    {
+      "epoch": 3.4876543209876543,
+      "grad_norm": 8.375,
+      "learning_rate": 9.157894736842105e-06,
+      "loss": 0.7232,
+      "step": 1130
+    },
+    {
+      "epoch": 3.490740740740741,
+      "grad_norm": 9.9375,
+      "learning_rate": 9.147368421052633e-06,
+      "loss": 0.6668,
+      "step": 1131
+    },
+    {
+      "epoch": 3.493827160493827,
+      "grad_norm": 7.0,
+      "learning_rate": 9.136842105263158e-06,
+      "loss": 0.7361,
+      "step": 1132
+    },
+    {
+      "epoch": 3.496913580246914,
+      "grad_norm": 9.8125,
+      "learning_rate": 9.126315789473686e-06,
+      "loss": 0.6975,
+      "step": 1133
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 7.34375,
+      "learning_rate": 9.11578947368421e-06,
+      "loss": 0.7229,
+      "step": 1134
+    },
+    {
+      "epoch": 3.503086419753086,
+      "grad_norm": 7.96875,
+      "learning_rate": 9.105263157894739e-06,
+      "loss": 0.7324,
+      "step": 1135
+    },
+    {
+      "epoch": 3.506172839506173,
+      "grad_norm": 7.625,
+      "learning_rate": 9.094736842105263e-06,
+      "loss": 0.6965,
+      "step": 1136
+    },
+    {
+      "epoch": 3.5092592592592595,
+      "grad_norm": 7.625,
+      "learning_rate": 9.08421052631579e-06,
+      "loss": 0.7274,
+      "step": 1137
+    },
+    {
+      "epoch": 3.5123456790123457,
+      "grad_norm": 8.3125,
+      "learning_rate": 9.073684210526316e-06,
+      "loss": 0.6886,
+      "step": 1138
+    },
+    {
+      "epoch": 3.515432098765432,
+      "grad_norm": 7.625,
+      "learning_rate": 9.063157894736842e-06,
+      "loss": 0.7075,
+      "step": 1139
+    },
+    {
+      "epoch": 3.5185185185185186,
+      "grad_norm": 9.4375,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 0.6846,
+      "step": 1140
+    },
+    {
+      "epoch": 3.521604938271605,
+      "grad_norm": 7.84375,
+      "learning_rate": 9.042105263157895e-06,
+      "loss": 0.6805,
+      "step": 1141
+    },
+    {
+      "epoch": 3.5246913580246915,
+      "grad_norm": 9.0,
+      "learning_rate": 9.031578947368423e-06,
+      "loss": 0.7501,
+      "step": 1142
+    },
+    {
+      "epoch": 3.5277777777777777,
+      "grad_norm": 7.0625,
+      "learning_rate": 9.021052631578948e-06,
+      "loss": 0.713,
+      "step": 1143
+    },
+    {
+      "epoch": 3.5308641975308643,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.010526315789474e-06,
+      "loss": 0.686,
+      "step": 1144
+    },
+    {
+      "epoch": 3.5339506172839505,
+      "grad_norm": 7.0,
+      "learning_rate": 9e-06,
+      "loss": 0.7144,
+      "step": 1145
+    },
+    {
+      "epoch": 3.537037037037037,
+      "grad_norm": 7.28125,
+      "learning_rate": 8.989473684210527e-06,
+      "loss": 0.7481,
+      "step": 1146
+    },
+    {
+      "epoch": 3.5401234567901234,
+      "grad_norm": 7.40625,
+      "learning_rate": 8.978947368421055e-06,
+      "loss": 0.717,
+      "step": 1147
+    },
+    {
+      "epoch": 3.5432098765432096,
+      "grad_norm": 8.1875,
+      "learning_rate": 8.96842105263158e-06,
+      "loss": 0.7156,
+      "step": 1148
+    },
+    {
+      "epoch": 3.5462962962962963,
+      "grad_norm": 7.0625,
+      "learning_rate": 8.957894736842107e-06,
+      "loss": 0.6922,
+      "step": 1149
+    },
+    {
+      "epoch": 3.549382716049383,
+      "grad_norm": 6.8125,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.7329,
+      "step": 1150
+    },
+    {
+      "epoch": 3.552469135802469,
+      "grad_norm": 7.125,
+      "learning_rate": 8.936842105263158e-06,
+      "loss": 0.6754,
+      "step": 1151
+    },
+    {
+      "epoch": 3.5555555555555554,
+      "grad_norm": 6.96875,
+      "learning_rate": 8.926315789473685e-06,
+      "loss": 0.6523,
+      "step": 1152
+    },
+    {
+      "epoch": 3.558641975308642,
+      "grad_norm": 6.84375,
+      "learning_rate": 8.915789473684211e-06,
+      "loss": 0.6573,
+      "step": 1153
+    },
+    {
+      "epoch": 3.5617283950617287,
+      "grad_norm": 7.875,
+      "learning_rate": 8.905263157894737e-06,
+      "loss": 0.7719,
+      "step": 1154
+    },
+    {
+      "epoch": 3.564814814814815,
+      "grad_norm": 7.1875,
+      "learning_rate": 8.894736842105264e-06,
+      "loss": 0.7262,
+      "step": 1155
+    },
+    {
+      "epoch": 3.567901234567901,
+      "grad_norm": 6.78125,
+      "learning_rate": 8.884210526315792e-06,
+      "loss": 0.6707,
+      "step": 1156
+    },
+    {
+      "epoch": 3.5709876543209877,
+      "grad_norm": 8.75,
+      "learning_rate": 8.873684210526316e-06,
+      "loss": 0.6759,
+      "step": 1157
+    },
+    {
+      "epoch": 3.574074074074074,
+      "grad_norm": 8.4375,
+      "learning_rate": 8.863157894736842e-06,
+      "loss": 0.7266,
+      "step": 1158
+    },
+    {
+      "epoch": 3.5771604938271606,
+      "grad_norm": 8.5625,
+      "learning_rate": 8.852631578947369e-06,
+      "loss": 0.7344,
+      "step": 1159
+    },
+    {
+      "epoch": 3.580246913580247,
+      "grad_norm": 7.25,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 0.7794,
+      "step": 1160
+    },
+    {
+      "epoch": 3.5833333333333335,
+      "grad_norm": 6.3125,
+      "learning_rate": 8.831578947368421e-06,
+      "loss": 0.6879,
+      "step": 1161
+    },
+    {
+      "epoch": 3.5864197530864197,
+      "grad_norm": 6.34375,
+      "learning_rate": 8.821052631578948e-06,
+      "loss": 0.7177,
+      "step": 1162
+    },
+    {
+      "epoch": 3.5895061728395063,
+      "grad_norm": 14.875,
+      "learning_rate": 8.810526315789474e-06,
+      "loss": 0.7204,
+      "step": 1163
+    },
+    {
+      "epoch": 3.5925925925925926,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.8e-06,
+      "loss": 0.7408,
+      "step": 1164
+    },
+    {
+      "epoch": 3.5956790123456788,
+      "grad_norm": 6.625,
+      "learning_rate": 8.789473684210527e-06,
+      "loss": 0.6928,
+      "step": 1165
+    },
+    {
+      "epoch": 3.5987654320987654,
+      "grad_norm": 8.5,
+      "learning_rate": 8.778947368421053e-06,
+      "loss": 0.7745,
+      "step": 1166
+    },
+    {
+      "epoch": 3.601851851851852,
+      "grad_norm": 9.375,
+      "learning_rate": 8.76842105263158e-06,
+      "loss": 0.6935,
+      "step": 1167
+    },
+    {
+      "epoch": 3.6049382716049383,
+      "grad_norm": 7.78125,
+      "learning_rate": 8.757894736842106e-06,
+      "loss": 0.6565,
+      "step": 1168
+    },
+    {
+      "epoch": 3.6080246913580245,
+      "grad_norm": 6.625,
+      "learning_rate": 8.747368421052632e-06,
+      "loss": 0.7368,
+      "step": 1169
+    },
+    {
+      "epoch": 3.611111111111111,
+      "grad_norm": 8.0,
+      "learning_rate": 8.736842105263158e-06,
+      "loss": 0.7717,
+      "step": 1170
+    },
+    {
+      "epoch": 3.6141975308641974,
+      "grad_norm": 11.125,
+      "learning_rate": 8.726315789473685e-06,
+      "loss": 0.7153,
+      "step": 1171
+    },
+    {
+      "epoch": 3.617283950617284,
+      "grad_norm": 7.75,
+      "learning_rate": 8.715789473684211e-06,
+      "loss": 0.7361,
+      "step": 1172
+    },
+    {
+      "epoch": 3.6203703703703702,
+      "grad_norm": 10.0,
+      "learning_rate": 8.705263157894737e-06,
+      "loss": 0.6705,
+      "step": 1173
+    },
+    {
+      "epoch": 3.623456790123457,
+      "grad_norm": 6.25,
+      "learning_rate": 8.694736842105264e-06,
+      "loss": 0.6701,
+      "step": 1174
+    },
+    {
+      "epoch": 3.626543209876543,
+      "grad_norm": 7.0625,
+      "learning_rate": 8.68421052631579e-06,
+      "loss": 0.7038,
+      "step": 1175
+    },
+    {
+      "epoch": 3.6296296296296298,
+      "grad_norm": 29.75,
+      "learning_rate": 8.673684210526316e-06,
+      "loss": 0.6844,
+      "step": 1176
+    },
+    {
+      "epoch": 3.632716049382716,
+      "grad_norm": 6.96875,
+      "learning_rate": 8.663157894736843e-06,
+      "loss": 0.7003,
+      "step": 1177
+    },
+    {
+      "epoch": 3.6358024691358026,
+      "grad_norm": 7.15625,
+      "learning_rate": 8.652631578947369e-06,
+      "loss": 0.7084,
+      "step": 1178
+    },
+    {
+      "epoch": 3.638888888888889,
+      "grad_norm": 8.0,
+      "learning_rate": 8.642105263157895e-06,
+      "loss": 0.7541,
+      "step": 1179
+    },
+    {
+      "epoch": 3.6419753086419755,
+      "grad_norm": 7.375,
+      "learning_rate": 8.631578947368422e-06,
+      "loss": 0.7177,
+      "step": 1180
+    },
+    {
+      "epoch": 3.6450617283950617,
+      "grad_norm": 7.8125,
+      "learning_rate": 8.621052631578948e-06,
+      "loss": 0.7603,
+      "step": 1181
+    },
+    {
+      "epoch": 3.648148148148148,
+      "grad_norm": 7.40625,
+      "learning_rate": 8.610526315789474e-06,
+      "loss": 0.7189,
+      "step": 1182
+    },
+    {
+      "epoch": 3.6512345679012346,
+      "grad_norm": 7.375,
+      "learning_rate": 8.6e-06,
+      "loss": 0.696,
+      "step": 1183
+    },
+    {
+      "epoch": 3.6543209876543212,
+      "grad_norm": 8.0,
+      "learning_rate": 8.589473684210527e-06,
+      "loss": 0.7064,
+      "step": 1184
+    },
+    {
+      "epoch": 3.6574074074074074,
+      "grad_norm": 8.3125,
+      "learning_rate": 8.578947368421053e-06,
+      "loss": 0.7206,
+      "step": 1185
+    },
+    {
+      "epoch": 3.6604938271604937,
+      "grad_norm": 7.5625,
+      "learning_rate": 8.56842105263158e-06,
+      "loss": 0.6796,
+      "step": 1186
+    },
+    {
+      "epoch": 3.6635802469135803,
+      "grad_norm": 7.53125,
+      "learning_rate": 8.557894736842106e-06,
+      "loss": 0.6851,
+      "step": 1187
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 7.75,
+      "learning_rate": 8.547368421052632e-06,
+      "loss": 0.6677,
+      "step": 1188
+    },
+    {
+      "epoch": 3.669753086419753,
+      "grad_norm": 7.28125,
+      "learning_rate": 8.536842105263159e-06,
+      "loss": 0.6662,
+      "step": 1189
+    },
+    {
+      "epoch": 3.6728395061728394,
+      "grad_norm": 8.1875,
+      "learning_rate": 8.526315789473685e-06,
+      "loss": 0.7648,
+      "step": 1190
+    },
+    {
+      "epoch": 3.675925925925926,
+      "grad_norm": 10.125,
+      "learning_rate": 8.515789473684211e-06,
+      "loss": 0.6893,
+      "step": 1191
+    },
+    {
+      "epoch": 3.6790123456790123,
+      "grad_norm": 6.65625,
+      "learning_rate": 8.505263157894738e-06,
+      "loss": 0.6873,
+      "step": 1192
+    },
+    {
+      "epoch": 3.682098765432099,
+      "grad_norm": 6.375,
+      "learning_rate": 8.494736842105264e-06,
+      "loss": 0.655,
+      "step": 1193
+    },
+    {
+      "epoch": 3.685185185185185,
+      "grad_norm": 20.25,
+      "learning_rate": 8.48421052631579e-06,
+      "loss": 0.7376,
+      "step": 1194
+    },
+    {
+      "epoch": 3.6882716049382713,
+      "grad_norm": 10.25,
+      "learning_rate": 8.473684210526317e-06,
+      "loss": 0.7684,
+      "step": 1195
+    },
+    {
+      "epoch": 3.691358024691358,
+      "grad_norm": 10.625,
+      "learning_rate": 8.463157894736843e-06,
+      "loss": 0.6943,
+      "step": 1196
+    },
+    {
+      "epoch": 3.6944444444444446,
+      "grad_norm": 6.625,
+      "learning_rate": 8.45263157894737e-06,
+      "loss": 0.7398,
+      "step": 1197
+    },
+    {
+      "epoch": 3.697530864197531,
+      "grad_norm": 7.21875,
+      "learning_rate": 8.442105263157896e-06,
+      "loss": 0.7367,
+      "step": 1198
+    },
+    {
+      "epoch": 3.700617283950617,
+      "grad_norm": 8.0,
+      "learning_rate": 8.431578947368422e-06,
+      "loss": 0.673,
+      "step": 1199
+    },
+    {
+      "epoch": 3.7037037037037037,
+      "grad_norm": 6.875,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.6844,
+      "step": 1200
+    },
+    {
+      "epoch": 3.7067901234567904,
+      "grad_norm": 6.9375,
+      "learning_rate": 8.410526315789475e-06,
+      "loss": 0.7222,
+      "step": 1201
+    },
+    {
+      "epoch": 3.7098765432098766,
+      "grad_norm": 8.25,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.7509,
+      "step": 1202
+    },
+    {
+      "epoch": 3.712962962962963,
+      "grad_norm": 7.65625,
+      "learning_rate": 8.389473684210527e-06,
+      "loss": 0.7303,
+      "step": 1203
+    },
+    {
+      "epoch": 3.7160493827160495,
+      "grad_norm": 9.625,
+      "learning_rate": 8.378947368421054e-06,
+      "loss": 0.7166,
+      "step": 1204
+    },
+    {
+      "epoch": 3.7191358024691357,
+      "grad_norm": 11.0625,
+      "learning_rate": 8.36842105263158e-06,
+      "loss": 0.752,
+      "step": 1205
+    },
+    {
+      "epoch": 3.7222222222222223,
+      "grad_norm": 7.0,
+      "learning_rate": 8.357894736842106e-06,
+      "loss": 0.6385,
+      "step": 1206
+    },
+    {
+      "epoch": 3.7253086419753085,
+      "grad_norm": 11.875,
+      "learning_rate": 8.347368421052633e-06,
+      "loss": 0.6982,
+      "step": 1207
+    },
+    {
+      "epoch": 3.728395061728395,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.336842105263159e-06,
+      "loss": 0.7234,
+      "step": 1208
+    },
+    {
+      "epoch": 3.7314814814814814,
+      "grad_norm": 6.8125,
+      "learning_rate": 8.326315789473685e-06,
+      "loss": 0.6789,
+      "step": 1209
+    },
+    {
+      "epoch": 3.734567901234568,
+      "grad_norm": 9.4375,
+      "learning_rate": 8.315789473684212e-06,
+      "loss": 0.6857,
+      "step": 1210
+    },
+    {
+      "epoch": 3.7376543209876543,
+      "grad_norm": 7.90625,
+      "learning_rate": 8.305263157894738e-06,
+      "loss": 0.703,
+      "step": 1211
+    },
+    {
+      "epoch": 3.7407407407407405,
+      "grad_norm": 8.0,
+      "learning_rate": 8.294736842105264e-06,
+      "loss": 0.7361,
+      "step": 1212
+    },
+    {
+      "epoch": 3.743827160493827,
+      "grad_norm": 8.8125,
+      "learning_rate": 8.28421052631579e-06,
+      "loss": 0.75,
+      "step": 1213
+    },
+    {
+      "epoch": 3.746913580246914,
+      "grad_norm": 8.5625,
+      "learning_rate": 8.273684210526317e-06,
+      "loss": 0.7514,
+      "step": 1214
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 8.75,
+      "learning_rate": 8.263157894736843e-06,
+      "loss": 0.7187,
+      "step": 1215
+    },
+    {
+      "epoch": 3.753086419753086,
+      "grad_norm": 8.9375,
+      "learning_rate": 8.25263157894737e-06,
+      "loss": 0.7042,
+      "step": 1216
+    },
+    {
+      "epoch": 3.756172839506173,
+      "grad_norm": 11.625,
+      "learning_rate": 8.242105263157896e-06,
+      "loss": 0.7419,
+      "step": 1217
+    },
+    {
+      "epoch": 3.7592592592592595,
+      "grad_norm": 6.5625,
+      "learning_rate": 8.231578947368422e-06,
+      "loss": 0.7319,
+      "step": 1218
+    },
+    {
+      "epoch": 3.7623456790123457,
+      "grad_norm": 7.90625,
+      "learning_rate": 8.221052631578948e-06,
+      "loss": 0.7533,
+      "step": 1219
+    },
+    {
+      "epoch": 3.765432098765432,
+      "grad_norm": 11.0,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 0.7112,
+      "step": 1220
+    },
+    {
+      "epoch": 3.7685185185185186,
+      "grad_norm": 6.84375,
+      "learning_rate": 8.2e-06,
+      "loss": 0.6746,
+      "step": 1221
+    },
+    {
+      "epoch": 3.771604938271605,
+      "grad_norm": 8.5,
+      "learning_rate": 8.189473684210527e-06,
+      "loss": 0.7147,
+      "step": 1222
+    },
+    {
+      "epoch": 3.7746913580246915,
+      "grad_norm": 9.0,
+      "learning_rate": 8.178947368421054e-06,
+      "loss": 0.7212,
+      "step": 1223
+    },
+    {
+      "epoch": 3.7777777777777777,
+      "grad_norm": 7.75,
+      "learning_rate": 8.16842105263158e-06,
+      "loss": 0.7066,
+      "step": 1224
+    },
+    {
+      "epoch": 3.7808641975308643,
+      "grad_norm": 7.21875,
+      "learning_rate": 8.157894736842106e-06,
+      "loss": 0.6398,
+      "step": 1225
+    },
+    {
+      "epoch": 3.7839506172839505,
+      "grad_norm": 14.125,
+      "learning_rate": 8.147368421052633e-06,
+      "loss": 0.7254,
+      "step": 1226
+    },
+    {
+      "epoch": 3.787037037037037,
+      "grad_norm": 7.34375,
+      "learning_rate": 8.136842105263159e-06,
+      "loss": 0.6682,
+      "step": 1227
+    },
+    {
+      "epoch": 3.7901234567901234,
+      "grad_norm": 10.9375,
+      "learning_rate": 8.126315789473684e-06,
+      "loss": 0.7107,
+      "step": 1228
+    },
+    {
+      "epoch": 3.7932098765432096,
+      "grad_norm": 7.53125,
+      "learning_rate": 8.115789473684212e-06,
+      "loss": 0.688,
+      "step": 1229
+    },
+    {
+      "epoch": 3.7962962962962963,
+      "grad_norm": 8.0,
+      "learning_rate": 8.105263157894736e-06,
+      "loss": 0.6773,
+      "step": 1230
+    },
+    {
+      "epoch": 3.799382716049383,
+      "grad_norm": 6.46875,
+      "learning_rate": 8.094736842105264e-06,
+      "loss": 0.7217,
+      "step": 1231
+    },
+    {
+      "epoch": 3.802469135802469,
+      "grad_norm": 7.5625,
+      "learning_rate": 8.08421052631579e-06,
+      "loss": 0.6662,
+      "step": 1232
+    },
+    {
+      "epoch": 3.8055555555555554,
+      "grad_norm": 7.21875,
+      "learning_rate": 8.073684210526317e-06,
+      "loss": 0.6664,
+      "step": 1233
+    },
+    {
+      "epoch": 3.808641975308642,
+      "grad_norm": 6.40625,
+      "learning_rate": 8.063157894736843e-06,
+      "loss": 0.7067,
+      "step": 1234
+    },
+    {
+      "epoch": 3.8117283950617287,
+      "grad_norm": 27.375,
+      "learning_rate": 8.052631578947368e-06,
+      "loss": 0.7334,
+      "step": 1235
+    },
+    {
+      "epoch": 3.814814814814815,
+      "grad_norm": 7.15625,
+      "learning_rate": 8.042105263157896e-06,
+      "loss": 0.7417,
+      "step": 1236
+    },
+    {
+      "epoch": 3.817901234567901,
+      "grad_norm": 7.34375,
+      "learning_rate": 8.03157894736842e-06,
+      "loss": 0.6525,
+      "step": 1237
+    },
+    {
+      "epoch": 3.8209876543209877,
+      "grad_norm": 7.125,
+      "learning_rate": 8.021052631578949e-06,
+      "loss": 0.688,
+      "step": 1238
+    },
+    {
+      "epoch": 3.824074074074074,
+      "grad_norm": 9.125,
+      "learning_rate": 8.010526315789473e-06,
+      "loss": 0.7352,
+      "step": 1239
+    },
+    {
+      "epoch": 3.8271604938271606,
+      "grad_norm": 7.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7082,
+      "step": 1240
+    },
+    {
+      "epoch": 3.830246913580247,
+      "grad_norm": 7.75,
+      "learning_rate": 7.989473684210528e-06,
+      "loss": 0.7295,
+      "step": 1241
+    },
+    {
+      "epoch": 3.8333333333333335,
+      "grad_norm": 7.15625,
+      "learning_rate": 7.978947368421052e-06,
+      "loss": 0.7016,
+      "step": 1242
+    },
+    {
+      "epoch": 3.8364197530864197,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.96842105263158e-06,
+      "loss": 0.6882,
+      "step": 1243
+    },
+    {
+      "epoch": 3.8395061728395063,
+      "grad_norm": 7.78125,
+      "learning_rate": 7.957894736842105e-06,
+      "loss": 0.7348,
+      "step": 1244
+    },
+    {
+      "epoch": 3.8425925925925926,
+      "grad_norm": 7.40625,
+      "learning_rate": 7.947368421052633e-06,
+      "loss": 0.654,
+      "step": 1245
+    },
+    {
+      "epoch": 3.8456790123456788,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.936842105263158e-06,
+      "loss": 0.7605,
+      "step": 1246
+    },
+    {
+      "epoch": 3.8487654320987654,
+      "grad_norm": 6.75,
+      "learning_rate": 7.926315789473686e-06,
+      "loss": 0.6292,
+      "step": 1247
+    },
+    {
+      "epoch": 3.851851851851852,
+      "grad_norm": 8.8125,
+      "learning_rate": 7.915789473684212e-06,
+      "loss": 0.6885,
+      "step": 1248
+    },
+    {
+      "epoch": 3.8549382716049383,
+      "grad_norm": 19.75,
+      "learning_rate": 7.905263157894737e-06,
+      "loss": 0.6336,
+      "step": 1249
+    },
+    {
+      "epoch": 3.8580246913580245,
+      "grad_norm": 8.3125,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.6573,
+      "step": 1250
+    },
+    {
+      "epoch": 3.861111111111111,
+      "grad_norm": 6.5,
+      "learning_rate": 7.88421052631579e-06,
+      "loss": 0.6961,
+      "step": 1251
+    },
+    {
+      "epoch": 3.8641975308641974,
+      "grad_norm": 7.25,
+      "learning_rate": 7.873684210526317e-06,
+      "loss": 0.6762,
+      "step": 1252
+    },
+    {
+      "epoch": 3.867283950617284,
+      "grad_norm": 6.40625,
+      "learning_rate": 7.863157894736842e-06,
+      "loss": 0.5932,
+      "step": 1253
+    },
+    {
+      "epoch": 3.8703703703703702,
+      "grad_norm": 8.0625,
+      "learning_rate": 7.85263157894737e-06,
+      "loss": 0.7193,
+      "step": 1254
+    },
+    {
+      "epoch": 3.873456790123457,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.842105263157895e-06,
+      "loss": 0.7292,
+      "step": 1255
+    },
+    {
+      "epoch": 3.876543209876543,
+      "grad_norm": 7.28125,
+      "learning_rate": 7.831578947368421e-06,
+      "loss": 0.7309,
+      "step": 1256
+    },
+    {
+      "epoch": 3.8796296296296298,
+      "grad_norm": 8.1875,
+      "learning_rate": 7.821052631578949e-06,
+      "loss": 0.6804,
+      "step": 1257
+    },
+    {
+      "epoch": 3.882716049382716,
+      "grad_norm": 8.5625,
+      "learning_rate": 7.810526315789474e-06,
+      "loss": 0.7394,
+      "step": 1258
+    },
+    {
+      "epoch": 3.8858024691358026,
+      "grad_norm": 7.40625,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.6768,
+      "step": 1259
+    },
+    {
+      "epoch": 3.888888888888889,
+      "grad_norm": 8.4375,
+      "learning_rate": 7.789473684210526e-06,
+      "loss": 0.716,
+      "step": 1260
+    },
+    {
+      "epoch": 3.8919753086419755,
+      "grad_norm": 7.875,
+      "learning_rate": 7.778947368421054e-06,
+      "loss": 0.7454,
+      "step": 1261
+    },
+    {
+      "epoch": 3.8950617283950617,
+      "grad_norm": 6.75,
+      "learning_rate": 7.768421052631579e-06,
+      "loss": 0.6785,
+      "step": 1262
+    },
+    {
+      "epoch": 3.898148148148148,
+      "grad_norm": 6.9375,
+      "learning_rate": 7.757894736842105e-06,
+      "loss": 0.6845,
+      "step": 1263
+    },
+    {
+      "epoch": 3.9012345679012346,
+      "grad_norm": 8.0625,
+      "learning_rate": 7.747368421052631e-06,
+      "loss": 0.7325,
+      "step": 1264
+    },
+    {
+      "epoch": 3.9043209876543212,
+      "grad_norm": 6.8125,
+      "learning_rate": 7.736842105263158e-06,
+      "loss": 0.6814,
+      "step": 1265
+    },
+    {
+      "epoch": 3.9074074074074074,
+      "grad_norm": 7.625,
+      "learning_rate": 7.726315789473686e-06,
+      "loss": 0.7217,
+      "step": 1266
+    },
+    {
+      "epoch": 3.9104938271604937,
+      "grad_norm": 7.53125,
+      "learning_rate": 7.71578947368421e-06,
+      "loss": 0.7441,
+      "step": 1267
+    },
+    {
+      "epoch": 3.9135802469135803,
+      "grad_norm": 10.4375,
+      "learning_rate": 7.705263157894738e-06,
+      "loss": 0.7063,
+      "step": 1268
+    },
+    {
+      "epoch": 3.9166666666666665,
+      "grad_norm": 7.125,
+      "learning_rate": 7.694736842105263e-06,
+      "loss": 0.5993,
+      "step": 1269
+    },
+    {
+      "epoch": 3.919753086419753,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.68421052631579e-06,
+      "loss": 0.6893,
+      "step": 1270
+    },
+    {
+      "epoch": 3.9228395061728394,
+      "grad_norm": 6.84375,
+      "learning_rate": 7.673684210526316e-06,
+      "loss": 0.6904,
+      "step": 1271
+    },
+    {
+      "epoch": 3.925925925925926,
+      "grad_norm": 8.4375,
+      "learning_rate": 7.663157894736842e-06,
+      "loss": 0.7684,
+      "step": 1272
+    },
+    {
+      "epoch": 3.9290123456790123,
+      "grad_norm": 6.4375,
+      "learning_rate": 7.65263157894737e-06,
+      "loss": 0.7034,
+      "step": 1273
+    },
+    {
+      "epoch": 3.932098765432099,
+      "grad_norm": 9.25,
+      "learning_rate": 7.642105263157895e-06,
+      "loss": 0.7045,
+      "step": 1274
+    },
+    {
+      "epoch": 3.935185185185185,
+      "grad_norm": 8.0,
+      "learning_rate": 7.631578947368423e-06,
+      "loss": 0.6874,
+      "step": 1275
+    },
+    {
+      "epoch": 3.9382716049382713,
+      "grad_norm": 8.8125,
+      "learning_rate": 7.621052631578948e-06,
+      "loss": 0.7455,
+      "step": 1276
+    },
+    {
+      "epoch": 3.941358024691358,
+      "grad_norm": 7.28125,
+      "learning_rate": 7.610526315789474e-06,
+      "loss": 0.7021,
+      "step": 1277
+    },
+    {
+      "epoch": 3.9444444444444446,
+      "grad_norm": 7.34375,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.6762,
+      "step": 1278
+    },
+    {
+      "epoch": 3.947530864197531,
+      "grad_norm": 7.78125,
+      "learning_rate": 7.589473684210526e-06,
+      "loss": 0.7265,
+      "step": 1279
+    },
+    {
+      "epoch": 3.950617283950617,
+      "grad_norm": 8.125,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 0.7204,
+      "step": 1280
+    },
+    {
+      "epoch": 3.9537037037037037,
+      "grad_norm": 7.59375,
+      "learning_rate": 7.568421052631579e-06,
+      "loss": 0.7642,
+      "step": 1281
+    },
+    {
+      "epoch": 3.9567901234567904,
+      "grad_norm": 7.0625,
+      "learning_rate": 7.557894736842106e-06,
+      "loss": 0.6623,
+      "step": 1282
+    },
+    {
+      "epoch": 3.9598765432098766,
+      "grad_norm": 7.875,
+      "learning_rate": 7.547368421052632e-06,
+      "loss": 0.7037,
+      "step": 1283
+    },
+    {
+      "epoch": 3.962962962962963,
+      "grad_norm": 7.15625,
+      "learning_rate": 7.536842105263158e-06,
+      "loss": 0.7092,
+      "step": 1284
+    },
+    {
+      "epoch": 3.9660493827160495,
+      "grad_norm": 6.71875,
+      "learning_rate": 7.526315789473685e-06,
+      "loss": 0.7399,
+      "step": 1285
+    },
+    {
+      "epoch": 3.9691358024691357,
+      "grad_norm": 7.375,
+      "learning_rate": 7.515789473684211e-06,
+      "loss": 0.7256,
+      "step": 1286
+    },
+    {
+      "epoch": 3.9722222222222223,
+      "grad_norm": 6.84375,
+      "learning_rate": 7.505263157894738e-06,
+      "loss": 0.7226,
+      "step": 1287
+    },
+    {
+      "epoch": 3.9753086419753085,
+      "grad_norm": 6.53125,
+      "learning_rate": 7.494736842105263e-06,
+      "loss": 0.7395,
+      "step": 1288
+    },
+    {
+      "epoch": 3.978395061728395,
+      "grad_norm": 7.6875,
+      "learning_rate": 7.4842105263157905e-06,
+      "loss": 0.7806,
+      "step": 1289
+    },
+    {
+      "epoch": 3.9814814814814814,
+      "grad_norm": 9.25,
+      "learning_rate": 7.473684210526316e-06,
+      "loss": 0.8023,
+      "step": 1290
+    },
+    {
+      "epoch": 3.984567901234568,
+      "grad_norm": 7.0,
+      "learning_rate": 7.463157894736843e-06,
+      "loss": 0.7113,
+      "step": 1291
+    },
+    {
+      "epoch": 3.9876543209876543,
+      "grad_norm": 8.1875,
+      "learning_rate": 7.4526315789473695e-06,
+      "loss": 0.7586,
+      "step": 1292
+    },
+    {
+      "epoch": 3.9907407407407405,
+      "grad_norm": 7.1875,
+      "learning_rate": 7.442105263157895e-06,
+      "loss": 0.695,
+      "step": 1293
+    },
+    {
+      "epoch": 3.993827160493827,
+      "grad_norm": 6.59375,
+      "learning_rate": 7.431578947368422e-06,
+      "loss": 0.6791,
+      "step": 1294
+    },
+    {
+      "epoch": 3.996913580246914,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.421052631578948e-06,
+      "loss": 0.6991,
+      "step": 1295
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 16.0,
+      "learning_rate": 7.410526315789475e-06,
+      "loss": 0.6944,
+      "step": 1296
+    },
+    {
+      "epoch": 4.003086419753086,
+      "grad_norm": 8.8125,
+      "learning_rate": 7.4e-06,
+      "loss": 0.7132,
+      "step": 1297
+    },
+    {
+      "epoch": 4.006172839506172,
+      "grad_norm": 6.71875,
+      "learning_rate": 7.3894736842105275e-06,
+      "loss": 0.7338,
+      "step": 1298
+    },
+    {
+      "epoch": 4.0092592592592595,
+      "grad_norm": 7.75,
+      "learning_rate": 7.378947368421053e-06,
+      "loss": 0.716,
+      "step": 1299
+    },
+    {
+      "epoch": 4.012345679012346,
+      "grad_norm": 7.03125,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.6739,
+      "step": 1300
+    },
+    {
+      "epoch": 4.015432098765432,
+      "grad_norm": 6.875,
+      "learning_rate": 7.3578947368421065e-06,
+      "loss": 0.6844,
+      "step": 1301
+    },
+    {
+      "epoch": 4.018518518518518,
+      "grad_norm": 6.8125,
+      "learning_rate": 7.347368421052632e-06,
+      "loss": 0.6697,
+      "step": 1302
+    },
+    {
+      "epoch": 4.021604938271605,
+      "grad_norm": 6.3125,
+      "learning_rate": 7.336842105263159e-06,
+      "loss": 0.6529,
+      "step": 1303
+    },
+    {
+      "epoch": 4.0246913580246915,
+      "grad_norm": 8.9375,
+      "learning_rate": 7.326315789473685e-06,
+      "loss": 0.7015,
+      "step": 1304
+    },
+    {
+      "epoch": 4.027777777777778,
+      "grad_norm": 9.9375,
+      "learning_rate": 7.315789473684212e-06,
+      "loss": 0.712,
+      "step": 1305
+    },
+    {
+      "epoch": 4.030864197530864,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.305263157894737e-06,
+      "loss": 0.734,
+      "step": 1306
+    },
+    {
+      "epoch": 4.033950617283951,
+      "grad_norm": 7.53125,
+      "learning_rate": 7.2947368421052636e-06,
+      "loss": 0.6167,
+      "step": 1307
+    },
+    {
+      "epoch": 4.037037037037037,
+      "grad_norm": 10.1875,
+      "learning_rate": 7.28421052631579e-06,
+      "loss": 0.6942,
+      "step": 1308
+    },
+    {
+      "epoch": 4.040123456790123,
+      "grad_norm": 7.84375,
+      "learning_rate": 7.273684210526316e-06,
+      "loss": 0.6926,
+      "step": 1309
+    },
+    {
+      "epoch": 4.04320987654321,
+      "grad_norm": 7.75,
+      "learning_rate": 7.263157894736843e-06,
+      "loss": 0.7111,
+      "step": 1310
+    },
+    {
+      "epoch": 4.046296296296297,
+      "grad_norm": 10.5625,
+      "learning_rate": 7.252631578947369e-06,
+      "loss": 0.668,
+      "step": 1311
+    },
+    {
+      "epoch": 4.049382716049383,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.242105263157896e-06,
+      "loss": 0.7296,
+      "step": 1312
+    },
+    {
+      "epoch": 4.052469135802469,
+      "grad_norm": 16.875,
+      "learning_rate": 7.2315789473684215e-06,
+      "loss": 0.7438,
+      "step": 1313
+    },
+    {
+      "epoch": 4.055555555555555,
+      "grad_norm": 19.5,
+      "learning_rate": 7.221052631578948e-06,
+      "loss": 0.6394,
+      "step": 1314
+    },
+    {
+      "epoch": 4.058641975308642,
+      "grad_norm": 7.4375,
+      "learning_rate": 7.210526315789474e-06,
+      "loss": 0.7009,
+      "step": 1315
+    },
+    {
+      "epoch": 4.061728395061729,
+      "grad_norm": 7.28125,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.718,
+      "step": 1316
+    },
+    {
+      "epoch": 4.064814814814815,
+      "grad_norm": 8.0625,
+      "learning_rate": 7.189473684210527e-06,
+      "loss": 0.7183,
+      "step": 1317
+    },
+    {
+      "epoch": 4.067901234567901,
+      "grad_norm": 7.15625,
+      "learning_rate": 7.178947368421053e-06,
+      "loss": 0.6767,
+      "step": 1318
+    },
+    {
+      "epoch": 4.070987654320987,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.16842105263158e-06,
+      "loss": 0.6853,
+      "step": 1319
+    },
+    {
+      "epoch": 4.074074074074074,
+      "grad_norm": 7.21875,
+      "learning_rate": 7.157894736842106e-06,
+      "loss": 0.7006,
+      "step": 1320
+    },
+    {
+      "epoch": 4.077160493827161,
+      "grad_norm": 6.96875,
+      "learning_rate": 7.147368421052631e-06,
+      "loss": 0.7404,
+      "step": 1321
+    },
+    {
+      "epoch": 4.080246913580247,
+      "grad_norm": 17.375,
+      "learning_rate": 7.1368421052631585e-06,
+      "loss": 0.7252,
+      "step": 1322
+    },
+    {
+      "epoch": 4.083333333333333,
+      "grad_norm": 6.5,
+      "learning_rate": 7.126315789473685e-06,
+      "loss": 0.7009,
+      "step": 1323
+    },
+    {
+      "epoch": 4.08641975308642,
+      "grad_norm": 6.3125,
+      "learning_rate": 7.115789473684211e-06,
+      "loss": 0.6668,
+      "step": 1324
+    },
+    {
+      "epoch": 4.089506172839506,
+      "grad_norm": 8.0625,
+      "learning_rate": 7.1052631578947375e-06,
+      "loss": 0.7405,
+      "step": 1325
+    },
+    {
+      "epoch": 4.092592592592593,
+      "grad_norm": 7.8125,
+      "learning_rate": 7.094736842105265e-06,
+      "loss": 0.6393,
+      "step": 1326
+    },
+    {
+      "epoch": 4.095679012345679,
+      "grad_norm": 9.4375,
+      "learning_rate": 7.08421052631579e-06,
+      "loss": 0.6897,
+      "step": 1327
+    },
+    {
+      "epoch": 4.098765432098766,
+      "grad_norm": 9.3125,
+      "learning_rate": 7.073684210526316e-06,
+      "loss": 0.6921,
+      "step": 1328
+    },
+    {
+      "epoch": 4.101851851851852,
+      "grad_norm": 6.09375,
+      "learning_rate": 7.063157894736843e-06,
+      "loss": 0.6729,
+      "step": 1329
+    },
+    {
+      "epoch": 4.104938271604938,
+      "grad_norm": 7.40625,
+      "learning_rate": 7.052631578947369e-06,
+      "loss": 0.7038,
+      "step": 1330
+    },
+    {
+      "epoch": 4.1080246913580245,
+      "grad_norm": 6.625,
+      "learning_rate": 7.0421052631578954e-06,
+      "loss": 0.7465,
+      "step": 1331
+    },
+    {
+      "epoch": 4.111111111111111,
+      "grad_norm": 7.0625,
+      "learning_rate": 7.031578947368422e-06,
+      "loss": 0.6888,
+      "step": 1332
+    },
+    {
+      "epoch": 4.114197530864198,
+      "grad_norm": 6.5,
+      "learning_rate": 7.021052631578948e-06,
+      "loss": 0.7083,
+      "step": 1333
+    },
+    {
+      "epoch": 4.117283950617284,
+      "grad_norm": 12.75,
+      "learning_rate": 7.010526315789474e-06,
+      "loss": 0.6463,
+      "step": 1334
+    },
+    {
+      "epoch": 4.12037037037037,
+      "grad_norm": 7.0,
+      "learning_rate": 7e-06,
+      "loss": 0.6533,
+      "step": 1335
+    },
+    {
+      "epoch": 4.1234567901234565,
+      "grad_norm": 7.46875,
+      "learning_rate": 6.989473684210527e-06,
+      "loss": 0.7268,
+      "step": 1336
+    },
+    {
+      "epoch": 4.1265432098765435,
+      "grad_norm": 10.25,
+      "learning_rate": 6.9789473684210525e-06,
+      "loss": 0.635,
+      "step": 1337
+    },
+    {
+      "epoch": 4.12962962962963,
+      "grad_norm": 7.875,
+      "learning_rate": 6.96842105263158e-06,
+      "loss": 0.6933,
+      "step": 1338
+    },
+    {
+      "epoch": 4.132716049382716,
+      "grad_norm": 12.4375,
+      "learning_rate": 6.957894736842106e-06,
+      "loss": 0.6917,
+      "step": 1339
+    },
+    {
+      "epoch": 4.135802469135802,
+      "grad_norm": 7.53125,
+      "learning_rate": 6.947368421052632e-06,
+      "loss": 0.6529,
+      "step": 1340
+    },
+    {
+      "epoch": 4.138888888888889,
+      "grad_norm": 7.1875,
+      "learning_rate": 6.936842105263159e-06,
+      "loss": 0.685,
+      "step": 1341
+    },
+    {
+      "epoch": 4.1419753086419755,
+      "grad_norm": 7.3125,
+      "learning_rate": 6.926315789473684e-06,
+      "loss": 0.6715,
+      "step": 1342
+    },
+    {
+      "epoch": 4.145061728395062,
+      "grad_norm": 8.25,
+      "learning_rate": 6.915789473684211e-06,
+      "loss": 0.7607,
+      "step": 1343
+    },
+    {
+      "epoch": 4.148148148148148,
+      "grad_norm": 6.8125,
+      "learning_rate": 6.905263157894737e-06,
+      "loss": 0.6607,
+      "step": 1344
+    },
+    {
+      "epoch": 4.151234567901234,
+      "grad_norm": 8.25,
+      "learning_rate": 6.894736842105264e-06,
+      "loss": 0.6906,
+      "step": 1345
+    },
+    {
+      "epoch": 4.154320987654321,
+      "grad_norm": 6.34375,
+      "learning_rate": 6.8842105263157895e-06,
+      "loss": 0.6603,
+      "step": 1346
+    },
+    {
+      "epoch": 4.157407407407407,
+      "grad_norm": 7.4375,
+      "learning_rate": 6.873684210526317e-06,
+      "loss": 0.672,
+      "step": 1347
+    },
+    {
+      "epoch": 4.160493827160494,
+      "grad_norm": 8.8125,
+      "learning_rate": 6.863157894736843e-06,
+      "loss": 0.6942,
+      "step": 1348
+    },
+    {
+      "epoch": 4.16358024691358,
+      "grad_norm": 7.40625,
+      "learning_rate": 6.8526315789473685e-06,
+      "loss": 0.6901,
+      "step": 1349
+    },
+    {
+      "epoch": 4.166666666666667,
+      "grad_norm": 8.875,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.6988,
+      "step": 1350
+    },
+    {
+      "epoch": 4.169753086419753,
+      "grad_norm": 11.4375,
+      "learning_rate": 6.831578947368421e-06,
+      "loss": 0.6869,
+      "step": 1351
+    },
+    {
+      "epoch": 4.172839506172839,
+      "grad_norm": 6.9375,
+      "learning_rate": 6.821052631578948e-06,
+      "loss": 0.7208,
+      "step": 1352
+    },
+    {
+      "epoch": 4.175925925925926,
+      "grad_norm": 8.375,
+      "learning_rate": 6.810526315789474e-06,
+      "loss": 0.6955,
+      "step": 1353
+    },
+    {
+      "epoch": 4.179012345679013,
+      "grad_norm": 8.375,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.6701,
+      "step": 1354
+    },
+    {
+      "epoch": 4.182098765432099,
+      "grad_norm": 7.96875,
+      "learning_rate": 6.789473684210527e-06,
+      "loss": 0.7003,
+      "step": 1355
+    },
+    {
+      "epoch": 4.185185185185185,
+      "grad_norm": 7.34375,
+      "learning_rate": 6.778947368421053e-06,
+      "loss": 0.6827,
+      "step": 1356
+    },
+    {
+      "epoch": 4.188271604938271,
+      "grad_norm": 6.75,
+      "learning_rate": 6.76842105263158e-06,
+      "loss": 0.645,
+      "step": 1357
+    },
+    {
+      "epoch": 4.191358024691358,
+      "grad_norm": 6.59375,
+      "learning_rate": 6.7578947368421054e-06,
+      "loss": 0.7087,
+      "step": 1358
+    },
+    {
+      "epoch": 4.194444444444445,
+      "grad_norm": 8.75,
+      "learning_rate": 6.747368421052633e-06,
+      "loss": 0.7315,
+      "step": 1359
+    },
+    {
+      "epoch": 4.197530864197531,
+      "grad_norm": 7.625,
+      "learning_rate": 6.736842105263158e-06,
+      "loss": 0.6632,
+      "step": 1360
+    },
+    {
+      "epoch": 4.200617283950617,
+      "grad_norm": 7.1875,
+      "learning_rate": 6.726315789473685e-06,
+      "loss": 0.732,
+      "step": 1361
+    },
+    {
+      "epoch": 4.203703703703703,
+      "grad_norm": 10.375,
+      "learning_rate": 6.715789473684211e-06,
+      "loss": 0.7123,
+      "step": 1362
+    },
+    {
+      "epoch": 4.20679012345679,
+      "grad_norm": 8.5625,
+      "learning_rate": 6.705263157894737e-06,
+      "loss": 0.7325,
+      "step": 1363
+    },
+    {
+      "epoch": 4.209876543209877,
+      "grad_norm": 6.78125,
+      "learning_rate": 6.694736842105264e-06,
+      "loss": 0.7091,
+      "step": 1364
+    },
+    {
+      "epoch": 4.212962962962963,
+      "grad_norm": 7.5625,
+      "learning_rate": 6.68421052631579e-06,
+      "loss": 0.661,
+      "step": 1365
+    },
+    {
+      "epoch": 4.216049382716049,
+      "grad_norm": 8.0625,
+      "learning_rate": 6.673684210526317e-06,
+      "loss": 0.6648,
+      "step": 1366
+    },
+    {
+      "epoch": 4.219135802469136,
+      "grad_norm": 7.1875,
+      "learning_rate": 6.663157894736842e-06,
+      "loss": 0.7191,
+      "step": 1367
+    },
+    {
+      "epoch": 4.222222222222222,
+      "grad_norm": 6.875,
+      "learning_rate": 6.6526315789473695e-06,
+      "loss": 0.6623,
+      "step": 1368
+    },
+    {
+      "epoch": 4.2253086419753085,
+      "grad_norm": 12.25,
+      "learning_rate": 6.642105263157895e-06,
+      "loss": 0.6468,
+      "step": 1369
+    },
+    {
+      "epoch": 4.228395061728395,
+      "grad_norm": 8.875,
+      "learning_rate": 6.631578947368421e-06,
+      "loss": 0.6829,
+      "step": 1370
+    },
+    {
+      "epoch": 4.231481481481482,
+      "grad_norm": 7.3125,
+      "learning_rate": 6.621052631578948e-06,
+      "loss": 0.7408,
+      "step": 1371
+    },
+    {
+      "epoch": 4.234567901234568,
+      "grad_norm": 7.34375,
+      "learning_rate": 6.610526315789474e-06,
+      "loss": 0.7032,
+      "step": 1372
+    },
+    {
+      "epoch": 4.237654320987654,
+      "grad_norm": 7.5625,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.7092,
+      "step": 1373
+    },
+    {
+      "epoch": 4.2407407407407405,
+      "grad_norm": 8.5625,
+      "learning_rate": 6.589473684210527e-06,
+      "loss": 0.686,
+      "step": 1374
+    },
+    {
+      "epoch": 4.243827160493828,
+      "grad_norm": 6.53125,
+      "learning_rate": 6.578947368421054e-06,
+      "loss": 0.6537,
+      "step": 1375
+    },
+    {
+      "epoch": 4.246913580246914,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.568421052631579e-06,
+      "loss": 0.688,
+      "step": 1376
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 6.6875,
+      "learning_rate": 6.557894736842106e-06,
+      "loss": 0.7003,
+      "step": 1377
+    },
+    {
+      "epoch": 4.253086419753086,
+      "grad_norm": 6.78125,
+      "learning_rate": 6.547368421052632e-06,
+      "loss": 0.6965,
+      "step": 1378
+    },
+    {
+      "epoch": 4.256172839506172,
+      "grad_norm": 10.75,
+      "learning_rate": 6.536842105263158e-06,
+      "loss": 0.7202,
+      "step": 1379
+    },
+    {
+      "epoch": 4.2592592592592595,
+      "grad_norm": 12.375,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 0.7217,
+      "step": 1380
+    },
+    {
+      "epoch": 4.262345679012346,
+      "grad_norm": 14.5,
+      "learning_rate": 6.515789473684211e-06,
+      "loss": 0.7083,
+      "step": 1381
+    },
+    {
+      "epoch": 4.265432098765432,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.505263157894738e-06,
+      "loss": 0.6692,
+      "step": 1382
+    },
+    {
+      "epoch": 4.268518518518518,
+      "grad_norm": 8.0625,
+      "learning_rate": 6.494736842105264e-06,
+      "loss": 0.6868,
+      "step": 1383
+    },
+    {
+      "epoch": 4.271604938271605,
+      "grad_norm": 7.25,
+      "learning_rate": 6.484210526315789e-06,
+      "loss": 0.6762,
+      "step": 1384
+    },
+    {
+      "epoch": 4.2746913580246915,
+      "grad_norm": 7.28125,
+      "learning_rate": 6.473684210526316e-06,
+      "loss": 0.6414,
+      "step": 1385
+    },
+    {
+      "epoch": 4.277777777777778,
+      "grad_norm": 7.625,
+      "learning_rate": 6.463157894736843e-06,
+      "loss": 0.7353,
+      "step": 1386
+    },
+    {
+      "epoch": 4.280864197530864,
+      "grad_norm": 7.375,
+      "learning_rate": 6.452631578947369e-06,
+      "loss": 0.7178,
+      "step": 1387
+    },
+    {
+      "epoch": 4.283950617283951,
+      "grad_norm": 6.625,
+      "learning_rate": 6.442105263157895e-06,
+      "loss": 0.647,
+      "step": 1388
+    },
+    {
+      "epoch": 4.287037037037037,
+      "grad_norm": 6.8125,
+      "learning_rate": 6.431578947368422e-06,
+      "loss": 0.6701,
+      "step": 1389
+    },
+    {
+      "epoch": 4.290123456790123,
+      "grad_norm": 6.625,
+      "learning_rate": 6.421052631578948e-06,
+      "loss": 0.6703,
+      "step": 1390
+    },
+    {
+      "epoch": 4.29320987654321,
+      "grad_norm": 9.1875,
+      "learning_rate": 6.410526315789473e-06,
+      "loss": 0.7151,
+      "step": 1391
+    },
+    {
+      "epoch": 4.296296296296296,
+      "grad_norm": 8.25,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.704,
+      "step": 1392
+    },
+    {
+      "epoch": 4.299382716049383,
+      "grad_norm": 11.5,
+      "learning_rate": 6.389473684210527e-06,
+      "loss": 0.6972,
+      "step": 1393
+    },
+    {
+      "epoch": 4.302469135802469,
+      "grad_norm": 7.59375,
+      "learning_rate": 6.378947368421053e-06,
+      "loss": 0.6551,
+      "step": 1394
+    },
+    {
+      "epoch": 4.305555555555555,
+      "grad_norm": 6.59375,
+      "learning_rate": 6.3684210526315795e-06,
+      "loss": 0.6809,
+      "step": 1395
+    },
+    {
+      "epoch": 4.308641975308642,
+      "grad_norm": 7.4375,
+      "learning_rate": 6.357894736842106e-06,
+      "loss": 0.6418,
+      "step": 1396
+    },
+    {
+      "epoch": 4.311728395061729,
+      "grad_norm": 8.0625,
+      "learning_rate": 6.347368421052632e-06,
+      "loss": 0.6527,
+      "step": 1397
+    },
+    {
+      "epoch": 4.314814814814815,
+      "grad_norm": 8.125,
+      "learning_rate": 6.336842105263158e-06,
+      "loss": 0.6843,
+      "step": 1398
+    },
+    {
+      "epoch": 4.317901234567901,
+      "grad_norm": 6.21875,
+      "learning_rate": 6.326315789473685e-06,
+      "loss": 0.615,
+      "step": 1399
+    },
+    {
+      "epoch": 4.320987654320987,
+      "grad_norm": 7.1875,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.677,
+      "step": 1400
+    },
+    {
+      "epoch": 4.324074074074074,
+      "grad_norm": 7.375,
+      "learning_rate": 6.3052631578947375e-06,
+      "loss": 0.6744,
+      "step": 1401
+    },
+    {
+      "epoch": 4.327160493827161,
+      "grad_norm": 6.6875,
+      "learning_rate": 6.294736842105264e-06,
+      "loss": 0.6649,
+      "step": 1402
+    },
+    {
+      "epoch": 4.330246913580247,
+      "grad_norm": 7.25,
+      "learning_rate": 6.28421052631579e-06,
+      "loss": 0.6611,
+      "step": 1403
+    },
+    {
+      "epoch": 4.333333333333333,
+      "grad_norm": 12.0625,
+      "learning_rate": 6.2736842105263165e-06,
+      "loss": 0.6464,
+      "step": 1404
+    },
+    {
+      "epoch": 4.33641975308642,
+      "grad_norm": 8.75,
+      "learning_rate": 6.263157894736842e-06,
+      "loss": 0.6726,
+      "step": 1405
+    },
+    {
+      "epoch": 4.339506172839506,
+      "grad_norm": 11.25,
+      "learning_rate": 6.252631578947369e-06,
+      "loss": 0.7384,
+      "step": 1406
+    },
+    {
+      "epoch": 4.342592592592593,
+      "grad_norm": 7.4375,
+      "learning_rate": 6.242105263157895e-06,
+      "loss": 0.6795,
+      "step": 1407
+    },
+    {
+      "epoch": 4.345679012345679,
+      "grad_norm": 8.0625,
+      "learning_rate": 6.231578947368422e-06,
+      "loss": 0.7161,
+      "step": 1408
+    },
+    {
+      "epoch": 4.348765432098766,
+      "grad_norm": 7.40625,
+      "learning_rate": 6.221052631578947e-06,
+      "loss": 0.713,
+      "step": 1409
+    },
+    {
+      "epoch": 4.351851851851852,
+      "grad_norm": 9.0625,
+      "learning_rate": 6.2105263157894745e-06,
+      "loss": 0.7286,
+      "step": 1410
+    },
+    {
+      "epoch": 4.354938271604938,
+      "grad_norm": 6.28125,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.7143,
+      "step": 1411
+    },
+    {
+      "epoch": 4.3580246913580245,
+      "grad_norm": 7.1875,
+      "learning_rate": 6.189473684210526e-06,
+      "loss": 0.666,
+      "step": 1412
+    },
+    {
+      "epoch": 4.361111111111111,
+      "grad_norm": 8.4375,
+      "learning_rate": 6.1789473684210534e-06,
+      "loss": 0.7044,
+      "step": 1413
+    },
+    {
+      "epoch": 4.364197530864198,
+      "grad_norm": 7.125,
+      "learning_rate": 6.168421052631579e-06,
+      "loss": 0.67,
+      "step": 1414
+    },
+    {
+      "epoch": 4.367283950617284,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.157894736842106e-06,
+      "loss": 0.7057,
+      "step": 1415
+    },
+    {
+      "epoch": 4.37037037037037,
+      "grad_norm": 7.03125,
+      "learning_rate": 6.1473684210526316e-06,
+      "loss": 0.7129,
+      "step": 1416
+    },
+    {
+      "epoch": 4.3734567901234565,
+      "grad_norm": 10.375,
+      "learning_rate": 6.136842105263159e-06,
+      "loss": 0.7411,
+      "step": 1417
+    },
+    {
+      "epoch": 4.3765432098765435,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.126315789473685e-06,
+      "loss": 0.711,
+      "step": 1418
+    },
+    {
+      "epoch": 4.37962962962963,
+      "grad_norm": 8.375,
+      "learning_rate": 6.1157894736842106e-06,
+      "loss": 0.6677,
+      "step": 1419
+    },
+    {
+      "epoch": 4.382716049382716,
+      "grad_norm": 7.78125,
+      "learning_rate": 6.105263157894738e-06,
+      "loss": 0.6937,
+      "step": 1420
+    },
+    {
+      "epoch": 4.385802469135802,
+      "grad_norm": 8.4375,
+      "learning_rate": 6.094736842105263e-06,
+      "loss": 0.67,
+      "step": 1421
+    },
+    {
+      "epoch": 4.388888888888889,
+      "grad_norm": 6.5625,
+      "learning_rate": 6.08421052631579e-06,
+      "loss": 0.6913,
+      "step": 1422
+    },
+    {
+      "epoch": 4.3919753086419755,
+      "grad_norm": 7.40625,
+      "learning_rate": 6.073684210526316e-06,
+      "loss": 0.7134,
+      "step": 1423
+    },
+    {
+      "epoch": 4.395061728395062,
+      "grad_norm": 10.5,
+      "learning_rate": 6.063157894736843e-06,
+      "loss": 0.6456,
+      "step": 1424
+    },
+    {
+      "epoch": 4.398148148148148,
+      "grad_norm": 6.21875,
+      "learning_rate": 6.0526315789473685e-06,
+      "loss": 0.6833,
+      "step": 1425
+    },
+    {
+      "epoch": 4.401234567901234,
+      "grad_norm": 8.875,
+      "learning_rate": 6.042105263157895e-06,
+      "loss": 0.7267,
+      "step": 1426
+    },
+    {
+      "epoch": 4.404320987654321,
+      "grad_norm": 7.71875,
+      "learning_rate": 6.031578947368422e-06,
+      "loss": 0.7243,
+      "step": 1427
+    },
+    {
+      "epoch": 4.407407407407407,
+      "grad_norm": 8.0,
+      "learning_rate": 6.0210526315789475e-06,
+      "loss": 0.7118,
+      "step": 1428
+    },
+    {
+      "epoch": 4.410493827160494,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.010526315789475e-06,
+      "loss": 0.7129,
+      "step": 1429
+    },
+    {
+      "epoch": 4.41358024691358,
+      "grad_norm": 6.875,
+      "learning_rate": 6e-06,
+      "loss": 0.6988,
+      "step": 1430
+    },
+    {
+      "epoch": 4.416666666666667,
+      "grad_norm": 13.75,
+      "learning_rate": 5.989473684210527e-06,
+      "loss": 0.6613,
+      "step": 1431
+    },
+    {
+      "epoch": 4.419753086419753,
+      "grad_norm": 6.65625,
+      "learning_rate": 5.978947368421053e-06,
+      "loss": 0.6978,
+      "step": 1432
+    },
+    {
+      "epoch": 4.422839506172839,
+      "grad_norm": 7.8125,
+      "learning_rate": 5.968421052631579e-06,
+      "loss": 0.705,
+      "step": 1433
+    },
+    {
+      "epoch": 4.425925925925926,
+      "grad_norm": 10.6875,
+      "learning_rate": 5.9578947368421055e-06,
+      "loss": 0.7149,
+      "step": 1434
+    },
+    {
+      "epoch": 4.429012345679013,
+      "grad_norm": 7.1875,
+      "learning_rate": 5.947368421052632e-06,
+      "loss": 0.7286,
+      "step": 1435
+    },
+    {
+      "epoch": 4.432098765432099,
+      "grad_norm": 7.5625,
+      "learning_rate": 5.936842105263159e-06,
+      "loss": 0.7386,
+      "step": 1436
+    },
+    {
+      "epoch": 4.435185185185185,
+      "grad_norm": 8.0,
+      "learning_rate": 5.9263157894736844e-06,
+      "loss": 0.6403,
+      "step": 1437
+    },
+    {
+      "epoch": 4.438271604938271,
+      "grad_norm": 9.6875,
+      "learning_rate": 5.915789473684212e-06,
+      "loss": 0.6998,
+      "step": 1438
+    },
+    {
+      "epoch": 4.441358024691358,
+      "grad_norm": 6.9375,
+      "learning_rate": 5.905263157894737e-06,
+      "loss": 0.6584,
+      "step": 1439
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 7.0,
+      "learning_rate": 5.8947368421052634e-06,
+      "loss": 0.6882,
+      "step": 1440
+    },
+    {
+      "epoch": 4.447530864197531,
+      "grad_norm": 6.875,
+      "learning_rate": 5.88421052631579e-06,
+      "loss": 0.601,
+      "step": 1441
+    },
+    {
+      "epoch": 4.450617283950617,
+      "grad_norm": 8.9375,
+      "learning_rate": 5.873684210526316e-06,
+      "loss": 0.7079,
+      "step": 1442
+    },
+    {
+      "epoch": 4.453703703703704,
+      "grad_norm": 6.9375,
+      "learning_rate": 5.863157894736842e-06,
+      "loss": 0.7057,
+      "step": 1443
+    },
+    {
+      "epoch": 4.45679012345679,
+      "grad_norm": 8.375,
+      "learning_rate": 5.852631578947369e-06,
+      "loss": 0.6498,
+      "step": 1444
+    },
+    {
+      "epoch": 4.459876543209877,
+      "grad_norm": 6.625,
+      "learning_rate": 5.842105263157896e-06,
+      "loss": 0.6802,
+      "step": 1445
+    },
+    {
+      "epoch": 4.462962962962963,
+      "grad_norm": 7.71875,
+      "learning_rate": 5.831578947368421e-06,
+      "loss": 0.7028,
+      "step": 1446
+    },
+    {
+      "epoch": 4.466049382716049,
+      "grad_norm": 7.3125,
+      "learning_rate": 5.8210526315789486e-06,
+      "loss": 0.6424,
+      "step": 1447
+    },
+    {
+      "epoch": 4.469135802469136,
+      "grad_norm": 7.40625,
+      "learning_rate": 5.810526315789474e-06,
+      "loss": 0.6848,
+      "step": 1448
+    },
+    {
+      "epoch": 4.472222222222222,
+      "grad_norm": 8.0,
+      "learning_rate": 5.8e-06,
+      "loss": 0.7289,
+      "step": 1449
+    },
+    {
+      "epoch": 4.4753086419753085,
+      "grad_norm": 8.4375,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.7631,
+      "step": 1450
+    },
+    {
+      "epoch": 4.478395061728395,
+      "grad_norm": 8.5625,
+      "learning_rate": 5.778947368421053e-06,
+      "loss": 0.7297,
+      "step": 1451
+    },
+    {
+      "epoch": 4.481481481481482,
+      "grad_norm": 7.4375,
+      "learning_rate": 5.76842105263158e-06,
+      "loss": 0.6557,
+      "step": 1452
+    },
+    {
+      "epoch": 4.484567901234568,
+      "grad_norm": 9.375,
+      "learning_rate": 5.757894736842106e-06,
+      "loss": 0.6948,
+      "step": 1453
+    },
+    {
+      "epoch": 4.487654320987654,
+      "grad_norm": 7.03125,
+      "learning_rate": 5.747368421052633e-06,
+      "loss": 0.6958,
+      "step": 1454
+    },
+    {
+      "epoch": 4.4907407407407405,
+      "grad_norm": 6.21875,
+      "learning_rate": 5.736842105263158e-06,
+      "loss": 0.693,
+      "step": 1455
+    },
+    {
+      "epoch": 4.493827160493828,
+      "grad_norm": 7.1875,
+      "learning_rate": 5.726315789473685e-06,
+      "loss": 0.7026,
+      "step": 1456
+    },
+    {
+      "epoch": 4.496913580246914,
+      "grad_norm": 10.5625,
+      "learning_rate": 5.715789473684211e-06,
+      "loss": 0.6767,
+      "step": 1457
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 7.8125,
+      "learning_rate": 5.705263157894737e-06,
+      "loss": 0.5933,
+      "step": 1458
+    },
+    {
+      "epoch": 4.503086419753086,
+      "grad_norm": 7.59375,
+      "learning_rate": 5.694736842105264e-06,
+      "loss": 0.7468,
+      "step": 1459
+    },
+    {
+      "epoch": 4.506172839506172,
+      "grad_norm": 8.125,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 0.6667,
+      "step": 1460
+    },
+    {
+      "epoch": 4.5092592592592595,
+      "grad_norm": 7.65625,
+      "learning_rate": 5.673684210526317e-06,
+      "loss": 0.7096,
+      "step": 1461
+    },
+    {
+      "epoch": 4.512345679012346,
+      "grad_norm": 9.75,
+      "learning_rate": 5.663157894736843e-06,
+      "loss": 0.681,
+      "step": 1462
+    },
+    {
+      "epoch": 4.515432098765432,
+      "grad_norm": 8.6875,
+      "learning_rate": 5.652631578947368e-06,
+      "loss": 0.6509,
+      "step": 1463
+    },
+    {
+      "epoch": 4.518518518518518,
+      "grad_norm": 8.625,
+      "learning_rate": 5.642105263157895e-06,
+      "loss": 0.6885,
+      "step": 1464
+    },
+    {
+      "epoch": 4.521604938271605,
+      "grad_norm": 10.625,
+      "learning_rate": 5.631578947368422e-06,
+      "loss": 0.6945,
+      "step": 1465
+    },
+    {
+      "epoch": 4.5246913580246915,
+      "grad_norm": 7.5,
+      "learning_rate": 5.621052631578948e-06,
+      "loss": 0.6749,
+      "step": 1466
+    },
+    {
+      "epoch": 4.527777777777778,
+      "grad_norm": 7.03125,
+      "learning_rate": 5.610526315789474e-06,
+      "loss": 0.7388,
+      "step": 1467
+    },
+    {
+      "epoch": 4.530864197530864,
+      "grad_norm": 7.15625,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.688,
+      "step": 1468
+    },
+    {
+      "epoch": 4.533950617283951,
+      "grad_norm": 6.96875,
+      "learning_rate": 5.589473684210527e-06,
+      "loss": 0.7091,
+      "step": 1469
+    },
+    {
+      "epoch": 4.537037037037037,
+      "grad_norm": 7.25,
+      "learning_rate": 5.578947368421052e-06,
+      "loss": 0.6312,
+      "step": 1470
+    },
+    {
+      "epoch": 4.540123456790123,
+      "grad_norm": 11.875,
+      "learning_rate": 5.5684210526315796e-06,
+      "loss": 0.6836,
+      "step": 1471
+    },
+    {
+      "epoch": 4.54320987654321,
+      "grad_norm": 6.78125,
+      "learning_rate": 5.557894736842105e-06,
+      "loss": 0.6743,
+      "step": 1472
+    },
+    {
+      "epoch": 4.546296296296296,
+      "grad_norm": 11.6875,
+      "learning_rate": 5.547368421052632e-06,
+      "loss": 0.6953,
+      "step": 1473
+    },
+    {
+      "epoch": 4.549382716049383,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.5368421052631586e-06,
+      "loss": 0.7267,
+      "step": 1474
+    },
+    {
+      "epoch": 4.552469135802469,
+      "grad_norm": 9.0,
+      "learning_rate": 5.526315789473685e-06,
+      "loss": 0.7173,
+      "step": 1475
+    },
+    {
+      "epoch": 4.555555555555555,
+      "grad_norm": 8.0625,
+      "learning_rate": 5.515789473684211e-06,
+      "loss": 0.71,
+      "step": 1476
+    },
+    {
+      "epoch": 4.5586419753086425,
+      "grad_norm": 7.0,
+      "learning_rate": 5.505263157894737e-06,
+      "loss": 0.6864,
+      "step": 1477
+    },
+    {
+      "epoch": 4.561728395061729,
+      "grad_norm": 7.28125,
+      "learning_rate": 5.494736842105264e-06,
+      "loss": 0.6507,
+      "step": 1478
+    },
+    {
+      "epoch": 4.564814814814815,
+      "grad_norm": 13.9375,
+      "learning_rate": 5.484210526315789e-06,
+      "loss": 0.6851,
+      "step": 1479
+    },
+    {
+      "epoch": 4.567901234567901,
+      "grad_norm": 8.5,
+      "learning_rate": 5.4736842105263165e-06,
+      "loss": 0.706,
+      "step": 1480
+    },
+    {
+      "epoch": 4.570987654320987,
+      "grad_norm": 9.1875,
+      "learning_rate": 5.463157894736843e-06,
+      "loss": 0.7272,
+      "step": 1481
+    },
+    {
+      "epoch": 4.574074074074074,
+      "grad_norm": 9.6875,
+      "learning_rate": 5.452631578947369e-06,
+      "loss": 0.7011,
+      "step": 1482
+    },
+    {
+      "epoch": 4.577160493827161,
+      "grad_norm": 7.8125,
+      "learning_rate": 5.4421052631578955e-06,
+      "loss": 0.6613,
+      "step": 1483
+    },
+    {
+      "epoch": 4.580246913580247,
+      "grad_norm": 7.59375,
+      "learning_rate": 5.431578947368421e-06,
+      "loss": 0.7162,
+      "step": 1484
+    },
+    {
+      "epoch": 4.583333333333333,
+      "grad_norm": 9.375,
+      "learning_rate": 5.421052631578948e-06,
+      "loss": 0.7315,
+      "step": 1485
+    },
+    {
+      "epoch": 4.58641975308642,
+      "grad_norm": 31.125,
+      "learning_rate": 5.410526315789474e-06,
+      "loss": 0.687,
+      "step": 1486
+    },
+    {
+      "epoch": 4.589506172839506,
+      "grad_norm": 6.75,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.673,
+      "step": 1487
+    },
+    {
+      "epoch": 4.592592592592593,
+      "grad_norm": 16.0,
+      "learning_rate": 5.389473684210526e-06,
+      "loss": 0.6864,
+      "step": 1488
+    },
+    {
+      "epoch": 4.595679012345679,
+      "grad_norm": 7.71875,
+      "learning_rate": 5.3789473684210535e-06,
+      "loss": 0.6976,
+      "step": 1489
+    },
+    {
+      "epoch": 4.598765432098766,
+      "grad_norm": 7.59375,
+      "learning_rate": 5.36842105263158e-06,
+      "loss": 0.6753,
+      "step": 1490
+    },
+    {
+      "epoch": 4.601851851851852,
+      "grad_norm": 8.375,
+      "learning_rate": 5.357894736842105e-06,
+      "loss": 0.6792,
+      "step": 1491
+    },
+    {
+      "epoch": 4.604938271604938,
+      "grad_norm": 18.0,
+      "learning_rate": 5.3473684210526325e-06,
+      "loss": 0.6431,
+      "step": 1492
+    },
+    {
+      "epoch": 4.6080246913580245,
+      "grad_norm": 10.1875,
+      "learning_rate": 5.336842105263158e-06,
+      "loss": 0.6742,
+      "step": 1493
+    },
+    {
+      "epoch": 4.611111111111111,
+      "grad_norm": 11.0625,
+      "learning_rate": 5.326315789473685e-06,
+      "loss": 0.6886,
+      "step": 1494
+    },
+    {
+      "epoch": 4.614197530864198,
+      "grad_norm": 10.8125,
+      "learning_rate": 5.315789473684211e-06,
+      "loss": 0.6721,
+      "step": 1495
+    },
+    {
+      "epoch": 4.617283950617284,
+      "grad_norm": 6.625,
+      "learning_rate": 5.305263157894738e-06,
+      "loss": 0.6691,
+      "step": 1496
+    },
+    {
+      "epoch": 4.62037037037037,
+      "grad_norm": 7.625,
+      "learning_rate": 5.294736842105263e-06,
+      "loss": 0.7218,
+      "step": 1497
+    },
+    {
+      "epoch": 4.6234567901234565,
+      "grad_norm": 8.3125,
+      "learning_rate": 5.2842105263157896e-06,
+      "loss": 0.7456,
+      "step": 1498
+    },
+    {
+      "epoch": 4.6265432098765435,
+      "grad_norm": 6.0,
+      "learning_rate": 5.273684210526317e-06,
+      "loss": 0.6962,
+      "step": 1499
+    },
+    {
+      "epoch": 4.62962962962963,
+      "grad_norm": 9.25,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.697,
+      "step": 1500
+    },
+    {
+      "epoch": 4.632716049382716,
+      "grad_norm": 6.96875,
+      "learning_rate": 5.252631578947369e-06,
+      "loss": 0.6672,
+      "step": 1501
+    },
+    {
+      "epoch": 4.635802469135802,
+      "grad_norm": 14.875,
+      "learning_rate": 5.242105263157895e-06,
+      "loss": 0.6954,
+      "step": 1502
+    },
+    {
+      "epoch": 4.638888888888889,
+      "grad_norm": 6.78125,
+      "learning_rate": 5.231578947368422e-06,
+      "loss": 0.6534,
+      "step": 1503
+    },
+    {
+      "epoch": 4.6419753086419755,
+      "grad_norm": 6.84375,
+      "learning_rate": 5.2210526315789475e-06,
+      "loss": 0.6764,
+      "step": 1504
+    },
+    {
+      "epoch": 4.645061728395062,
+      "grad_norm": 7.6875,
+      "learning_rate": 5.210526315789474e-06,
+      "loss": 0.6745,
+      "step": 1505
+    },
+    {
+      "epoch": 4.648148148148148,
+      "grad_norm": 14.9375,
+      "learning_rate": 5.2e-06,
+      "loss": 0.7009,
+      "step": 1506
+    },
+    {
+      "epoch": 4.651234567901234,
+      "grad_norm": 6.1875,
+      "learning_rate": 5.1894736842105265e-06,
+      "loss": 0.6733,
+      "step": 1507
+    },
+    {
+      "epoch": 4.654320987654321,
+      "grad_norm": 6.90625,
+      "learning_rate": 5.178947368421054e-06,
+      "loss": 0.6763,
+      "step": 1508
+    },
+    {
+      "epoch": 4.657407407407407,
+      "grad_norm": 6.71875,
+      "learning_rate": 5.168421052631579e-06,
+      "loss": 0.6697,
+      "step": 1509
+    },
+    {
+      "epoch": 4.660493827160494,
+      "grad_norm": 7.40625,
+      "learning_rate": 5.157894736842106e-06,
+      "loss": 0.7158,
+      "step": 1510
+    },
+    {
+      "epoch": 4.66358024691358,
+      "grad_norm": 9.25,
+      "learning_rate": 5.147368421052632e-06,
+      "loss": 0.7207,
+      "step": 1511
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 8.5,
+      "learning_rate": 5.136842105263158e-06,
+      "loss": 0.6952,
+      "step": 1512
+    },
+    {
+      "epoch": 4.669753086419753,
+      "grad_norm": 6.96875,
+      "learning_rate": 5.1263157894736845e-06,
+      "loss": 0.6919,
+      "step": 1513
+    },
+    {
+      "epoch": 4.672839506172839,
+      "grad_norm": 7.375,
+      "learning_rate": 5.115789473684211e-06,
+      "loss": 0.722,
+      "step": 1514
+    },
+    {
+      "epoch": 4.675925925925926,
+      "grad_norm": 6.78125,
+      "learning_rate": 5.105263157894738e-06,
+      "loss": 0.6695,
+      "step": 1515
+    },
+    {
+      "epoch": 4.679012345679013,
+      "grad_norm": 9.5625,
+      "learning_rate": 5.0947368421052635e-06,
+      "loss": 0.7365,
+      "step": 1516
+    },
+    {
+      "epoch": 4.682098765432099,
+      "grad_norm": 8.4375,
+      "learning_rate": 5.084210526315791e-06,
+      "loss": 0.66,
+      "step": 1517
+    },
+    {
+      "epoch": 4.685185185185185,
+      "grad_norm": 7.15625,
+      "learning_rate": 5.073684210526316e-06,
+      "loss": 0.6688,
+      "step": 1518
+    },
+    {
+      "epoch": 4.688271604938271,
+      "grad_norm": 8.5625,
+      "learning_rate": 5.0631578947368424e-06,
+      "loss": 0.7037,
+      "step": 1519
+    },
+    {
+      "epoch": 4.6913580246913575,
+      "grad_norm": 6.75,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 0.7061,
+      "step": 1520
+    },
+    {
+      "epoch": 4.694444444444445,
+      "grad_norm": 8.25,
+      "learning_rate": 5.042105263157895e-06,
+      "loss": 0.6332,
+      "step": 1521
+    },
+    {
+      "epoch": 4.697530864197531,
+      "grad_norm": 7.28125,
+      "learning_rate": 5.0315789473684214e-06,
+      "loss": 0.7027,
+      "step": 1522
+    },
+    {
+      "epoch": 4.700617283950617,
+      "grad_norm": 9.1875,
+      "learning_rate": 5.021052631578948e-06,
+      "loss": 0.6843,
+      "step": 1523
+    },
+    {
+      "epoch": 4.703703703703704,
+      "grad_norm": 7.03125,
+      "learning_rate": 5.010526315789475e-06,
+      "loss": 0.6633,
+      "step": 1524
+    },
+    {
+      "epoch": 4.70679012345679,
+      "grad_norm": 10.4375,
+      "learning_rate": 5e-06,
+      "loss": 0.7542,
+      "step": 1525
+    },
+    {
+      "epoch": 4.709876543209877,
+      "grad_norm": 6.53125,
+      "learning_rate": 4.989473684210527e-06,
+      "loss": 0.6611,
+      "step": 1526
+    },
+    {
+      "epoch": 4.712962962962963,
+      "grad_norm": 7.40625,
+      "learning_rate": 4.978947368421053e-06,
+      "loss": 0.7399,
+      "step": 1527
+    },
+    {
+      "epoch": 4.716049382716049,
+      "grad_norm": 8.25,
+      "learning_rate": 4.968421052631579e-06,
+      "loss": 0.6832,
+      "step": 1528
+    },
+    {
+      "epoch": 4.719135802469136,
+      "grad_norm": 6.8125,
+      "learning_rate": 4.957894736842106e-06,
+      "loss": 0.6301,
+      "step": 1529
+    },
+    {
+      "epoch": 4.722222222222222,
+      "grad_norm": 7.28125,
+      "learning_rate": 4.947368421052632e-06,
+      "loss": 0.6852,
+      "step": 1530
+    },
+    {
+      "epoch": 4.7253086419753085,
+      "grad_norm": 9.25,
+      "learning_rate": 4.936842105263158e-06,
+      "loss": 0.6776,
+      "step": 1531
+    },
+    {
+      "epoch": 4.728395061728395,
+      "grad_norm": 9.125,
+      "learning_rate": 4.926315789473685e-06,
+      "loss": 0.686,
+      "step": 1532
+    },
+    {
+      "epoch": 4.731481481481482,
+      "grad_norm": 8.125,
+      "learning_rate": 4.915789473684211e-06,
+      "loss": 0.7088,
+      "step": 1533
+    },
+    {
+      "epoch": 4.734567901234568,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.905263157894737e-06,
+      "loss": 0.7097,
+      "step": 1534
+    },
+    {
+      "epoch": 4.737654320987654,
+      "grad_norm": 6.96875,
+      "learning_rate": 4.894736842105264e-06,
+      "loss": 0.7426,
+      "step": 1535
+    },
+    {
+      "epoch": 4.7407407407407405,
+      "grad_norm": 7.15625,
+      "learning_rate": 4.88421052631579e-06,
+      "loss": 0.7459,
+      "step": 1536
+    },
+    {
+      "epoch": 4.743827160493828,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.873684210526316e-06,
+      "loss": 0.687,
+      "step": 1537
+    },
+    {
+      "epoch": 4.746913580246914,
+      "grad_norm": 7.59375,
+      "learning_rate": 4.863157894736843e-06,
+      "loss": 0.6749,
+      "step": 1538
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 6.875,
+      "learning_rate": 4.852631578947369e-06,
+      "loss": 0.7104,
+      "step": 1539
+    },
+    {
+      "epoch": 4.753086419753086,
+      "grad_norm": 6.6875,
+      "learning_rate": 4.842105263157895e-06,
+      "loss": 0.6514,
+      "step": 1540
+    },
+    {
+      "epoch": 4.756172839506172,
+      "grad_norm": 7.71875,
+      "learning_rate": 4.831578947368422e-06,
+      "loss": 0.6991,
+      "step": 1541
+    },
+    {
+      "epoch": 4.7592592592592595,
+      "grad_norm": 6.75,
+      "learning_rate": 4.821052631578948e-06,
+      "loss": 0.6793,
+      "step": 1542
+    },
+    {
+      "epoch": 4.762345679012346,
+      "grad_norm": 6.28125,
+      "learning_rate": 4.8105263157894735e-06,
+      "loss": 0.6516,
+      "step": 1543
+    },
+    {
+      "epoch": 4.765432098765432,
+      "grad_norm": 7.40625,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.7063,
+      "step": 1544
+    },
+    {
+      "epoch": 4.768518518518518,
+      "grad_norm": 7.28125,
+      "learning_rate": 4.789473684210527e-06,
+      "loss": 0.6539,
+      "step": 1545
+    },
+    {
+      "epoch": 4.771604938271605,
+      "grad_norm": 7.5,
+      "learning_rate": 4.778947368421053e-06,
+      "loss": 0.7186,
+      "step": 1546
+    },
+    {
+      "epoch": 4.7746913580246915,
+      "grad_norm": 8.0625,
+      "learning_rate": 4.76842105263158e-06,
+      "loss": 0.6543,
+      "step": 1547
+    },
+    {
+      "epoch": 4.777777777777778,
+      "grad_norm": 7.90625,
+      "learning_rate": 4.757894736842106e-06,
+      "loss": 0.6825,
+      "step": 1548
+    },
+    {
+      "epoch": 4.780864197530864,
+      "grad_norm": 10.4375,
+      "learning_rate": 4.747368421052632e-06,
+      "loss": 0.6895,
+      "step": 1549
+    },
+    {
+      "epoch": 4.783950617283951,
+      "grad_norm": 7.40625,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 0.6591,
+      "step": 1550
+    },
+    {
+      "epoch": 4.787037037037037,
+      "grad_norm": 7.875,
+      "learning_rate": 4.726315789473684e-06,
+      "loss": 0.6959,
+      "step": 1551
+    },
+    {
+      "epoch": 4.790123456790123,
+      "grad_norm": 8.3125,
+      "learning_rate": 4.71578947368421e-06,
+      "loss": 0.7074,
+      "step": 1552
+    },
+    {
+      "epoch": 4.79320987654321,
+      "grad_norm": 9.4375,
+      "learning_rate": 4.705263157894738e-06,
+      "loss": 0.7371,
+      "step": 1553
+    },
+    {
+      "epoch": 4.796296296296296,
+      "grad_norm": 7.28125,
+      "learning_rate": 4.694736842105264e-06,
+      "loss": 0.6664,
+      "step": 1554
+    },
+    {
+      "epoch": 4.799382716049383,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.68421052631579e-06,
+      "loss": 0.6723,
+      "step": 1555
+    },
+    {
+      "epoch": 4.802469135802469,
+      "grad_norm": 11.4375,
+      "learning_rate": 4.6736842105263166e-06,
+      "loss": 0.7024,
+      "step": 1556
+    },
+    {
+      "epoch": 4.805555555555555,
+      "grad_norm": 8.875,
+      "learning_rate": 4.663157894736842e-06,
+      "loss": 0.7014,
+      "step": 1557
+    },
+    {
+      "epoch": 4.8086419753086425,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.652631578947368e-06,
+      "loss": 0.6916,
+      "step": 1558
+    },
+    {
+      "epoch": 4.811728395061729,
+      "grad_norm": 8.5,
+      "learning_rate": 4.642105263157895e-06,
+      "loss": 0.6858,
+      "step": 1559
+    },
+    {
+      "epoch": 4.814814814814815,
+      "grad_norm": 8.0,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 0.6984,
+      "step": 1560
+    },
+    {
+      "epoch": 4.817901234567901,
+      "grad_norm": 7.3125,
+      "learning_rate": 4.621052631578948e-06,
+      "loss": 0.6056,
+      "step": 1561
+    },
+    {
+      "epoch": 4.820987654320987,
+      "grad_norm": 7.09375,
+      "learning_rate": 4.6105263157894745e-06,
+      "loss": 0.7138,
+      "step": 1562
+    },
+    {
+      "epoch": 4.824074074074074,
+      "grad_norm": 7.28125,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.6328,
+      "step": 1563
+    },
+    {
+      "epoch": 4.827160493827161,
+      "grad_norm": 9.4375,
+      "learning_rate": 4.589473684210526e-06,
+      "loss": 0.7647,
+      "step": 1564
+    },
+    {
+      "epoch": 4.830246913580247,
+      "grad_norm": 6.40625,
+      "learning_rate": 4.578947368421053e-06,
+      "loss": 0.7049,
+      "step": 1565
+    },
+    {
+      "epoch": 4.833333333333333,
+      "grad_norm": 7.46875,
+      "learning_rate": 4.568421052631579e-06,
+      "loss": 0.6969,
+      "step": 1566
+    },
+    {
+      "epoch": 4.83641975308642,
+      "grad_norm": 8.1875,
+      "learning_rate": 4.557894736842105e-06,
+      "loss": 0.7254,
+      "step": 1567
+    },
+    {
+      "epoch": 4.839506172839506,
+      "grad_norm": 7.375,
+      "learning_rate": 4.547368421052632e-06,
+      "loss": 0.6472,
+      "step": 1568
+    },
+    {
+      "epoch": 4.842592592592593,
+      "grad_norm": 7.0625,
+      "learning_rate": 4.536842105263158e-06,
+      "loss": 0.7117,
+      "step": 1569
+    },
+    {
+      "epoch": 4.845679012345679,
+      "grad_norm": 7.6875,
+      "learning_rate": 4.526315789473685e-06,
+      "loss": 0.6194,
+      "step": 1570
+    },
+    {
+      "epoch": 4.848765432098766,
+      "grad_norm": 7.53125,
+      "learning_rate": 4.5157894736842115e-06,
+      "loss": 0.69,
+      "step": 1571
+    },
+    {
+      "epoch": 4.851851851851852,
+      "grad_norm": 7.15625,
+      "learning_rate": 4.505263157894737e-06,
+      "loss": 0.663,
+      "step": 1572
+    },
+    {
+      "epoch": 4.854938271604938,
+      "grad_norm": 22.5,
+      "learning_rate": 4.494736842105263e-06,
+      "loss": 0.6863,
+      "step": 1573
+    },
+    {
+      "epoch": 4.8580246913580245,
+      "grad_norm": 8.4375,
+      "learning_rate": 4.48421052631579e-06,
+      "loss": 0.7032,
+      "step": 1574
+    },
+    {
+      "epoch": 4.861111111111111,
+      "grad_norm": 10.4375,
+      "learning_rate": 4.473684210526316e-06,
+      "loss": 0.6357,
+      "step": 1575
+    },
+    {
+      "epoch": 4.864197530864198,
+      "grad_norm": 9.3125,
+      "learning_rate": 4.463157894736842e-06,
+      "loss": 0.6767,
+      "step": 1576
+    },
+    {
+      "epoch": 4.867283950617284,
+      "grad_norm": 7.28125,
+      "learning_rate": 4.452631578947369e-06,
+      "loss": 0.7243,
+      "step": 1577
+    },
+    {
+      "epoch": 4.87037037037037,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.442105263157896e-06,
+      "loss": 0.7096,
+      "step": 1578
+    },
+    {
+      "epoch": 4.8734567901234565,
+      "grad_norm": 8.0625,
+      "learning_rate": 4.431578947368421e-06,
+      "loss": 0.6269,
+      "step": 1579
+    },
+    {
+      "epoch": 4.8765432098765435,
+      "grad_norm": 8.8125,
+      "learning_rate": 4.4210526315789476e-06,
+      "loss": 0.6455,
+      "step": 1580
+    },
+    {
+      "epoch": 4.87962962962963,
+      "grad_norm": 7.5,
+      "learning_rate": 4.410526315789474e-06,
+      "loss": 0.6863,
+      "step": 1581
+    },
+    {
+      "epoch": 4.882716049382716,
+      "grad_norm": 6.40625,
+      "learning_rate": 4.4e-06,
+      "loss": 0.6828,
+      "step": 1582
+    },
+    {
+      "epoch": 4.885802469135802,
+      "grad_norm": 8.5,
+      "learning_rate": 4.3894736842105266e-06,
+      "loss": 0.6905,
+      "step": 1583
+    },
+    {
+      "epoch": 4.888888888888889,
+      "grad_norm": 9.4375,
+      "learning_rate": 4.378947368421053e-06,
+      "loss": 0.7097,
+      "step": 1584
+    },
+    {
+      "epoch": 4.8919753086419755,
+      "grad_norm": 7.75,
+      "learning_rate": 4.368421052631579e-06,
+      "loss": 0.6713,
+      "step": 1585
+    },
+    {
+      "epoch": 4.895061728395062,
+      "grad_norm": 10.1875,
+      "learning_rate": 4.3578947368421055e-06,
+      "loss": 0.6666,
+      "step": 1586
+    },
+    {
+      "epoch": 4.898148148148148,
+      "grad_norm": 6.0625,
+      "learning_rate": 4.347368421052632e-06,
+      "loss": 0.5941,
+      "step": 1587
+    },
+    {
+      "epoch": 4.901234567901234,
+      "grad_norm": 11.75,
+      "learning_rate": 4.336842105263158e-06,
+      "loss": 0.6762,
+      "step": 1588
+    },
+    {
+      "epoch": 4.904320987654321,
+      "grad_norm": 7.3125,
+      "learning_rate": 4.3263157894736845e-06,
+      "loss": 0.7021,
+      "step": 1589
+    },
+    {
+      "epoch": 4.907407407407407,
+      "grad_norm": 7.71875,
+      "learning_rate": 4.315789473684211e-06,
+      "loss": 0.6733,
+      "step": 1590
+    },
+    {
+      "epoch": 4.910493827160494,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.305263157894737e-06,
+      "loss": 0.6557,
+      "step": 1591
+    },
+    {
+      "epoch": 4.91358024691358,
+      "grad_norm": 9.0625,
+      "learning_rate": 4.2947368421052635e-06,
+      "loss": 0.7364,
+      "step": 1592
+    },
+    {
+      "epoch": 4.916666666666667,
+      "grad_norm": 7.1875,
+      "learning_rate": 4.28421052631579e-06,
+      "loss": 0.7169,
+      "step": 1593
+    },
+    {
+      "epoch": 4.919753086419753,
+      "grad_norm": 7.21875,
+      "learning_rate": 4.273684210526316e-06,
+      "loss": 0.6915,
+      "step": 1594
+    },
+    {
+      "epoch": 4.922839506172839,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.2631578947368425e-06,
+      "loss": 0.6865,
+      "step": 1595
+    },
+    {
+      "epoch": 4.925925925925926,
+      "grad_norm": 7.96875,
+      "learning_rate": 4.252631578947369e-06,
+      "loss": 0.683,
+      "step": 1596
+    },
+    {
+      "epoch": 4.929012345679013,
+      "grad_norm": 9.0625,
+      "learning_rate": 4.242105263157895e-06,
+      "loss": 0.6625,
+      "step": 1597
+    },
+    {
+      "epoch": 4.932098765432099,
+      "grad_norm": 8.125,
+      "learning_rate": 4.2315789473684215e-06,
+      "loss": 0.6426,
+      "step": 1598
+    },
+    {
+      "epoch": 4.935185185185185,
+      "grad_norm": 6.78125,
+      "learning_rate": 4.221052631578948e-06,
+      "loss": 0.6709,
+      "step": 1599
+    },
+    {
+      "epoch": 4.938271604938271,
+      "grad_norm": 6.78125,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.6569,
+      "step": 1600
+    },
+    {
+      "epoch": 4.9413580246913575,
+      "grad_norm": 7.125,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.672,
+      "step": 1601
+    },
+    {
+      "epoch": 4.944444444444445,
+      "grad_norm": 7.25,
+      "learning_rate": 4.189473684210527e-06,
+      "loss": 0.7019,
+      "step": 1602
+    },
+    {
+      "epoch": 4.947530864197531,
+      "grad_norm": 7.625,
+      "learning_rate": 4.178947368421053e-06,
+      "loss": 0.6805,
+      "step": 1603
+    },
+    {
+      "epoch": 4.950617283950617,
+      "grad_norm": 6.5,
+      "learning_rate": 4.1684210526315794e-06,
+      "loss": 0.6915,
+      "step": 1604
+    },
+    {
+      "epoch": 4.953703703703704,
+      "grad_norm": 6.875,
+      "learning_rate": 4.157894736842106e-06,
+      "loss": 0.7088,
+      "step": 1605
+    },
+    {
+      "epoch": 4.95679012345679,
+      "grad_norm": 7.40625,
+      "learning_rate": 4.147368421052632e-06,
+      "loss": 0.657,
+      "step": 1606
+    },
+    {
+      "epoch": 4.959876543209877,
+      "grad_norm": 8.625,
+      "learning_rate": 4.136842105263158e-06,
+      "loss": 0.6841,
+      "step": 1607
+    },
+    {
+      "epoch": 4.962962962962963,
+      "grad_norm": 8.0625,
+      "learning_rate": 4.126315789473685e-06,
+      "loss": 0.7172,
+      "step": 1608
+    },
+    {
+      "epoch": 4.966049382716049,
+      "grad_norm": 9.0625,
+      "learning_rate": 4.115789473684211e-06,
+      "loss": 0.7073,
+      "step": 1609
+    },
+    {
+      "epoch": 4.969135802469136,
+      "grad_norm": 7.5625,
+      "learning_rate": 4.105263157894737e-06,
+      "loss": 0.6593,
+      "step": 1610
+    },
+    {
+      "epoch": 4.972222222222222,
+      "grad_norm": 29.875,
+      "learning_rate": 4.094736842105264e-06,
+      "loss": 0.7014,
+      "step": 1611
+    },
+    {
+      "epoch": 4.9753086419753085,
+      "grad_norm": 8.25,
+      "learning_rate": 4.08421052631579e-06,
+      "loss": 0.7064,
+      "step": 1612
+    },
+    {
+      "epoch": 4.978395061728395,
+      "grad_norm": 6.71875,
+      "learning_rate": 4.073684210526316e-06,
+      "loss": 0.6808,
+      "step": 1613
+    },
+    {
+      "epoch": 4.981481481481482,
+      "grad_norm": 7.09375,
+      "learning_rate": 4.063157894736842e-06,
+      "loss": 0.6727,
+      "step": 1614
+    },
+    {
+      "epoch": 4.984567901234568,
+      "grad_norm": 7.8125,
+      "learning_rate": 4.052631578947368e-06,
+      "loss": 0.6807,
+      "step": 1615
+    },
+    {
+      "epoch": 4.987654320987654,
+      "grad_norm": 8.75,
+      "learning_rate": 4.042105263157895e-06,
+      "loss": 0.7397,
+      "step": 1616
+    },
+    {
+      "epoch": 4.9907407407407405,
+      "grad_norm": 6.84375,
+      "learning_rate": 4.031578947368422e-06,
+      "loss": 0.7259,
+      "step": 1617
+    },
+    {
+      "epoch": 4.993827160493828,
+      "grad_norm": 6.59375,
+      "learning_rate": 4.021052631578948e-06,
+      "loss": 0.6554,
+      "step": 1618
+    },
+    {
+      "epoch": 4.996913580246914,
+      "grad_norm": 7.625,
+      "learning_rate": 4.010526315789474e-06,
+      "loss": 0.7078,
+      "step": 1619
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 8.25,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6784,
+      "step": 1620
+    },
+    {
+      "epoch": 5.003086419753086,
+      "grad_norm": 6.8125,
+      "learning_rate": 3.989473684210526e-06,
+      "loss": 0.6965,
+      "step": 1621
+    },
+    {
+      "epoch": 5.006172839506172,
+      "grad_norm": 6.90625,
+      "learning_rate": 3.9789473684210525e-06,
+      "loss": 0.7049,
+      "step": 1622
+    },
+    {
+      "epoch": 5.0092592592592595,
+      "grad_norm": 9.5,
+      "learning_rate": 3.968421052631579e-06,
+      "loss": 0.693,
+      "step": 1623
+    },
+    {
+      "epoch": 5.012345679012346,
+      "grad_norm": 9.5625,
+      "learning_rate": 3.957894736842106e-06,
+      "loss": 0.7368,
+      "step": 1624
+    },
+    {
+      "epoch": 5.015432098765432,
+      "grad_norm": 7.96875,
+      "learning_rate": 3.947368421052632e-06,
+      "loss": 0.6922,
+      "step": 1625
+    },
+    {
+      "epoch": 5.018518518518518,
+      "grad_norm": 7.4375,
+      "learning_rate": 3.936842105263159e-06,
+      "loss": 0.7196,
+      "step": 1626
+    },
+    {
+      "epoch": 5.021604938271605,
+      "grad_norm": 6.46875,
+      "learning_rate": 3.926315789473685e-06,
+      "loss": 0.6761,
+      "step": 1627
+    },
+    {
+      "epoch": 5.0246913580246915,
+      "grad_norm": 7.5625,
+      "learning_rate": 3.9157894736842104e-06,
+      "loss": 0.6837,
+      "step": 1628
+    },
+    {
+      "epoch": 5.027777777777778,
+      "grad_norm": 10.3125,
+      "learning_rate": 3.905263157894737e-06,
+      "loss": 0.6465,
+      "step": 1629
+    },
+    {
+      "epoch": 5.030864197530864,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.894736842105263e-06,
+      "loss": 0.7195,
+      "step": 1630
+    },
+    {
+      "epoch": 5.033950617283951,
+      "grad_norm": 6.125,
+      "learning_rate": 3.884210526315789e-06,
+      "loss": 0.639,
+      "step": 1631
+    },
+    {
+      "epoch": 5.037037037037037,
+      "grad_norm": 8.5625,
+      "learning_rate": 3.873684210526316e-06,
+      "loss": 0.6287,
+      "step": 1632
+    },
+    {
+      "epoch": 5.040123456790123,
+      "grad_norm": 6.375,
+      "learning_rate": 3.863157894736843e-06,
+      "loss": 0.6769,
+      "step": 1633
+    },
+    {
+      "epoch": 5.04320987654321,
+      "grad_norm": 8.0625,
+      "learning_rate": 3.852631578947369e-06,
+      "loss": 0.7142,
+      "step": 1634
+    },
+    {
+      "epoch": 5.046296296296297,
+      "grad_norm": 7.3125,
+      "learning_rate": 3.842105263157895e-06,
+      "loss": 0.6555,
+      "step": 1635
+    },
+    {
+      "epoch": 5.049382716049383,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.831578947368421e-06,
+      "loss": 0.692,
+      "step": 1636
+    },
+    {
+      "epoch": 5.052469135802469,
+      "grad_norm": 10.0,
+      "learning_rate": 3.821052631578947e-06,
+      "loss": 0.6954,
+      "step": 1637
+    },
+    {
+      "epoch": 5.055555555555555,
+      "grad_norm": 7.28125,
+      "learning_rate": 3.810526315789474e-06,
+      "loss": 0.7027,
+      "step": 1638
+    },
+    {
+      "epoch": 5.058641975308642,
+      "grad_norm": 8.0,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.6966,
+      "step": 1639
+    },
+    {
+      "epoch": 5.061728395061729,
+      "grad_norm": 9.8125,
+      "learning_rate": 3.789473684210527e-06,
+      "loss": 0.7469,
+      "step": 1640
+    },
+    {
+      "epoch": 5.064814814814815,
+      "grad_norm": 7.59375,
+      "learning_rate": 3.778947368421053e-06,
+      "loss": 0.6812,
+      "step": 1641
+    },
+    {
+      "epoch": 5.067901234567901,
+      "grad_norm": 7.0,
+      "learning_rate": 3.768421052631579e-06,
+      "loss": 0.6271,
+      "step": 1642
+    },
+    {
+      "epoch": 5.070987654320987,
+      "grad_norm": 8.1875,
+      "learning_rate": 3.7578947368421053e-06,
+      "loss": 0.7453,
+      "step": 1643
+    },
+    {
+      "epoch": 5.074074074074074,
+      "grad_norm": 7.65625,
+      "learning_rate": 3.7473684210526317e-06,
+      "loss": 0.6406,
+      "step": 1644
+    },
+    {
+      "epoch": 5.077160493827161,
+      "grad_norm": 7.0625,
+      "learning_rate": 3.736842105263158e-06,
+      "loss": 0.6995,
+      "step": 1645
+    },
+    {
+      "epoch": 5.080246913580247,
+      "grad_norm": 6.65625,
+      "learning_rate": 3.7263157894736848e-06,
+      "loss": 0.6943,
+      "step": 1646
+    },
+    {
+      "epoch": 5.083333333333333,
+      "grad_norm": 7.4375,
+      "learning_rate": 3.715789473684211e-06,
+      "loss": 0.6229,
+      "step": 1647
+    },
+    {
+      "epoch": 5.08641975308642,
+      "grad_norm": 6.4375,
+      "learning_rate": 3.7052631578947374e-06,
+      "loss": 0.6796,
+      "step": 1648
+    },
+    {
+      "epoch": 5.089506172839506,
+      "grad_norm": 6.34375,
+      "learning_rate": 3.6947368421052637e-06,
+      "loss": 0.6694,
+      "step": 1649
+    },
+    {
+      "epoch": 5.092592592592593,
+      "grad_norm": 7.5625,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 0.7165,
+      "step": 1650
+    },
+    {
+      "epoch": 5.095679012345679,
+      "grad_norm": 6.875,
+      "learning_rate": 3.673684210526316e-06,
+      "loss": 0.7128,
+      "step": 1651
+    },
+    {
+      "epoch": 5.098765432098766,
+      "grad_norm": 7.9375,
+      "learning_rate": 3.6631578947368423e-06,
+      "loss": 0.6683,
+      "step": 1652
+    },
+    {
+      "epoch": 5.101851851851852,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.6526315789473686e-06,
+      "loss": 0.6708,
+      "step": 1653
+    },
+    {
+      "epoch": 5.104938271604938,
+      "grad_norm": 11.75,
+      "learning_rate": 3.642105263157895e-06,
+      "loss": 0.7227,
+      "step": 1654
+    },
+    {
+      "epoch": 5.1080246913580245,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.6315789473684217e-06,
+      "loss": 0.7371,
+      "step": 1655
+    },
+    {
+      "epoch": 5.111111111111111,
+      "grad_norm": 9.375,
+      "learning_rate": 3.621052631578948e-06,
+      "loss": 0.6771,
+      "step": 1656
+    },
+    {
+      "epoch": 5.114197530864198,
+      "grad_norm": 7.65625,
+      "learning_rate": 3.610526315789474e-06,
+      "loss": 0.6729,
+      "step": 1657
+    },
+    {
+      "epoch": 5.117283950617284,
+      "grad_norm": 7.25,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.6892,
+      "step": 1658
+    },
+    {
+      "epoch": 5.12037037037037,
+      "grad_norm": 10.4375,
+      "learning_rate": 3.5894736842105266e-06,
+      "loss": 0.7127,
+      "step": 1659
+    },
+    {
+      "epoch": 5.1234567901234565,
+      "grad_norm": 7.5,
+      "learning_rate": 3.578947368421053e-06,
+      "loss": 0.7055,
+      "step": 1660
+    },
+    {
+      "epoch": 5.1265432098765435,
+      "grad_norm": 8.625,
+      "learning_rate": 3.5684210526315792e-06,
+      "loss": 0.6597,
+      "step": 1661
+    },
+    {
+      "epoch": 5.12962962962963,
+      "grad_norm": 7.96875,
+      "learning_rate": 3.5578947368421056e-06,
+      "loss": 0.6591,
+      "step": 1662
+    },
+    {
+      "epoch": 5.132716049382716,
+      "grad_norm": 7.0625,
+      "learning_rate": 3.5473684210526323e-06,
+      "loss": 0.7143,
+      "step": 1663
+    },
+    {
+      "epoch": 5.135802469135802,
+      "grad_norm": 6.59375,
+      "learning_rate": 3.536842105263158e-06,
+      "loss": 0.6118,
+      "step": 1664
+    },
+    {
+      "epoch": 5.138888888888889,
+      "grad_norm": 7.8125,
+      "learning_rate": 3.5263157894736846e-06,
+      "loss": 0.7126,
+      "step": 1665
+    },
+    {
+      "epoch": 5.1419753086419755,
+      "grad_norm": 15.25,
+      "learning_rate": 3.515789473684211e-06,
+      "loss": 0.7289,
+      "step": 1666
+    },
+    {
+      "epoch": 5.145061728395062,
+      "grad_norm": 8.8125,
+      "learning_rate": 3.505263157894737e-06,
+      "loss": 0.7259,
+      "step": 1667
+    },
+    {
+      "epoch": 5.148148148148148,
+      "grad_norm": 7.75,
+      "learning_rate": 3.4947368421052635e-06,
+      "loss": 0.7336,
+      "step": 1668
+    },
+    {
+      "epoch": 5.151234567901234,
+      "grad_norm": 9.6875,
+      "learning_rate": 3.48421052631579e-06,
+      "loss": 0.6703,
+      "step": 1669
+    },
+    {
+      "epoch": 5.154320987654321,
+      "grad_norm": 8.5,
+      "learning_rate": 3.473684210526316e-06,
+      "loss": 0.6417,
+      "step": 1670
+    },
+    {
+      "epoch": 5.157407407407407,
+      "grad_norm": 8.3125,
+      "learning_rate": 3.463157894736842e-06,
+      "loss": 0.6719,
+      "step": 1671
+    },
+    {
+      "epoch": 5.160493827160494,
+      "grad_norm": 11.625,
+      "learning_rate": 3.4526315789473684e-06,
+      "loss": 0.666,
+      "step": 1672
+    },
+    {
+      "epoch": 5.16358024691358,
+      "grad_norm": 6.40625,
+      "learning_rate": 3.4421052631578947e-06,
+      "loss": 0.6731,
+      "step": 1673
+    },
+    {
+      "epoch": 5.166666666666667,
+      "grad_norm": 7.15625,
+      "learning_rate": 3.4315789473684215e-06,
+      "loss": 0.6803,
+      "step": 1674
+    },
+    {
+      "epoch": 5.169753086419753,
+      "grad_norm": 7.5625,
+      "learning_rate": 3.421052631578948e-06,
+      "loss": 0.7046,
+      "step": 1675
+    },
+    {
+      "epoch": 5.172839506172839,
+      "grad_norm": 7.6875,
+      "learning_rate": 3.410526315789474e-06,
+      "loss": 0.6384,
+      "step": 1676
+    },
+    {
+      "epoch": 5.175925925925926,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.7465,
+      "step": 1677
+    },
+    {
+      "epoch": 5.179012345679013,
+      "grad_norm": 7.15625,
+      "learning_rate": 3.3894736842105264e-06,
+      "loss": 0.6857,
+      "step": 1678
+    },
+    {
+      "epoch": 5.182098765432099,
+      "grad_norm": 8.375,
+      "learning_rate": 3.3789473684210527e-06,
+      "loss": 0.6933,
+      "step": 1679
+    },
+    {
+      "epoch": 5.185185185185185,
+      "grad_norm": 6.71875,
+      "learning_rate": 3.368421052631579e-06,
+      "loss": 0.6705,
+      "step": 1680
+    },
+    {
+      "epoch": 5.188271604938271,
+      "grad_norm": 8.125,
+      "learning_rate": 3.3578947368421054e-06,
+      "loss": 0.6727,
+      "step": 1681
+    },
+    {
+      "epoch": 5.191358024691358,
+      "grad_norm": 8.0,
+      "learning_rate": 3.347368421052632e-06,
+      "loss": 0.6332,
+      "step": 1682
+    },
+    {
+      "epoch": 5.194444444444445,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.3368421052631584e-06,
+      "loss": 0.7038,
+      "step": 1683
+    },
+    {
+      "epoch": 5.197530864197531,
+      "grad_norm": 10.4375,
+      "learning_rate": 3.3263157894736848e-06,
+      "loss": 0.7139,
+      "step": 1684
+    },
+    {
+      "epoch": 5.200617283950617,
+      "grad_norm": 9.625,
+      "learning_rate": 3.3157894736842107e-06,
+      "loss": 0.6781,
+      "step": 1685
+    },
+    {
+      "epoch": 5.203703703703703,
+      "grad_norm": 6.4375,
+      "learning_rate": 3.305263157894737e-06,
+      "loss": 0.6898,
+      "step": 1686
+    },
+    {
+      "epoch": 5.20679012345679,
+      "grad_norm": 7.125,
+      "learning_rate": 3.2947368421052633e-06,
+      "loss": 0.6904,
+      "step": 1687
+    },
+    {
+      "epoch": 5.209876543209877,
+      "grad_norm": 7.65625,
+      "learning_rate": 3.2842105263157897e-06,
+      "loss": 0.6556,
+      "step": 1688
+    },
+    {
+      "epoch": 5.212962962962963,
+      "grad_norm": 8.0,
+      "learning_rate": 3.273684210526316e-06,
+      "loss": 0.6938,
+      "step": 1689
+    },
+    {
+      "epoch": 5.216049382716049,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.2631578947368423e-06,
+      "loss": 0.6434,
+      "step": 1690
+    },
+    {
+      "epoch": 5.219135802469136,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.252631578947369e-06,
+      "loss": 0.7199,
+      "step": 1691
+    },
+    {
+      "epoch": 5.222222222222222,
+      "grad_norm": 8.25,
+      "learning_rate": 3.2421052631578945e-06,
+      "loss": 0.6955,
+      "step": 1692
+    },
+    {
+      "epoch": 5.2253086419753085,
+      "grad_norm": 9.875,
+      "learning_rate": 3.2315789473684213e-06,
+      "loss": 0.6928,
+      "step": 1693
+    },
+    {
+      "epoch": 5.228395061728395,
+      "grad_norm": 6.65625,
+      "learning_rate": 3.2210526315789476e-06,
+      "loss": 0.6842,
+      "step": 1694
+    },
+    {
+      "epoch": 5.231481481481482,
+      "grad_norm": 7.15625,
+      "learning_rate": 3.210526315789474e-06,
+      "loss": 0.6378,
+      "step": 1695
+    },
+    {
+      "epoch": 5.234567901234568,
+      "grad_norm": 7.28125,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.6613,
+      "step": 1696
+    },
+    {
+      "epoch": 5.237654320987654,
+      "grad_norm": 8.0,
+      "learning_rate": 3.1894736842105266e-06,
+      "loss": 0.6001,
+      "step": 1697
+    },
+    {
+      "epoch": 5.2407407407407405,
+      "grad_norm": 7.875,
+      "learning_rate": 3.178947368421053e-06,
+      "loss": 0.6725,
+      "step": 1698
+    },
+    {
+      "epoch": 5.243827160493828,
+      "grad_norm": 19.875,
+      "learning_rate": 3.168421052631579e-06,
+      "loss": 0.6571,
+      "step": 1699
+    },
+    {
+      "epoch": 5.246913580246914,
+      "grad_norm": 6.65625,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.655,
+      "step": 1700
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 6.1875,
+      "learning_rate": 3.147368421052632e-06,
+      "loss": 0.7022,
+      "step": 1701
+    },
+    {
+      "epoch": 5.253086419753086,
+      "grad_norm": 8.4375,
+      "learning_rate": 3.1368421052631582e-06,
+      "loss": 0.6045,
+      "step": 1702
+    },
+    {
+      "epoch": 5.256172839506172,
+      "grad_norm": 9.1875,
+      "learning_rate": 3.1263157894736846e-06,
+      "loss": 0.6881,
+      "step": 1703
+    },
+    {
+      "epoch": 5.2592592592592595,
+      "grad_norm": 7.125,
+      "learning_rate": 3.115789473684211e-06,
+      "loss": 0.6861,
+      "step": 1704
+    },
+    {
+      "epoch": 5.262345679012346,
+      "grad_norm": 7.875,
+      "learning_rate": 3.1052631578947372e-06,
+      "loss": 0.7,
+      "step": 1705
+    },
+    {
+      "epoch": 5.265432098765432,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.094736842105263e-06,
+      "loss": 0.6729,
+      "step": 1706
+    },
+    {
+      "epoch": 5.268518518518518,
+      "grad_norm": 7.6875,
+      "learning_rate": 3.0842105263157895e-06,
+      "loss": 0.6688,
+      "step": 1707
+    },
+    {
+      "epoch": 5.271604938271605,
+      "grad_norm": 7.09375,
+      "learning_rate": 3.0736842105263158e-06,
+      "loss": 0.6266,
+      "step": 1708
+    },
+    {
+      "epoch": 5.2746913580246915,
+      "grad_norm": 6.75,
+      "learning_rate": 3.0631578947368425e-06,
+      "loss": 0.6289,
+      "step": 1709
+    },
+    {
+      "epoch": 5.277777777777778,
+      "grad_norm": 7.34375,
+      "learning_rate": 3.052631578947369e-06,
+      "loss": 0.6546,
+      "step": 1710
+    },
+    {
+      "epoch": 5.280864197530864,
+      "grad_norm": 6.65625,
+      "learning_rate": 3.042105263157895e-06,
+      "loss": 0.6475,
+      "step": 1711
+    },
+    {
+      "epoch": 5.283950617283951,
+      "grad_norm": 7.34375,
+      "learning_rate": 3.0315789473684215e-06,
+      "loss": 0.6469,
+      "step": 1712
+    },
+    {
+      "epoch": 5.287037037037037,
+      "grad_norm": 7.625,
+      "learning_rate": 3.0210526315789474e-06,
+      "loss": 0.6587,
+      "step": 1713
+    },
+    {
+      "epoch": 5.290123456790123,
+      "grad_norm": 7.09375,
+      "learning_rate": 3.0105263157894737e-06,
+      "loss": 0.7064,
+      "step": 1714
+    },
+    {
+      "epoch": 5.29320987654321,
+      "grad_norm": 9.0625,
+      "learning_rate": 3e-06,
+      "loss": 0.6896,
+      "step": 1715
+    },
+    {
+      "epoch": 5.296296296296296,
+      "grad_norm": 6.53125,
+      "learning_rate": 2.9894736842105264e-06,
+      "loss": 0.6458,
+      "step": 1716
+    },
+    {
+      "epoch": 5.299382716049383,
+      "grad_norm": 8.4375,
+      "learning_rate": 2.9789473684210527e-06,
+      "loss": 0.6643,
+      "step": 1717
+    },
+    {
+      "epoch": 5.302469135802469,
+      "grad_norm": 6.625,
+      "learning_rate": 2.9684210526315795e-06,
+      "loss": 0.6974,
+      "step": 1718
+    },
+    {
+      "epoch": 5.305555555555555,
+      "grad_norm": 10.125,
+      "learning_rate": 2.957894736842106e-06,
+      "loss": 0.6851,
+      "step": 1719
+    },
+    {
+      "epoch": 5.308641975308642,
+      "grad_norm": 7.875,
+      "learning_rate": 2.9473684210526317e-06,
+      "loss": 0.6604,
+      "step": 1720
+    },
+    {
+      "epoch": 5.311728395061729,
+      "grad_norm": 7.28125,
+      "learning_rate": 2.936842105263158e-06,
+      "loss": 0.6786,
+      "step": 1721
+    },
+    {
+      "epoch": 5.314814814814815,
+      "grad_norm": 7.90625,
+      "learning_rate": 2.9263157894736844e-06,
+      "loss": 0.7541,
+      "step": 1722
+    },
+    {
+      "epoch": 5.317901234567901,
+      "grad_norm": 7.125,
+      "learning_rate": 2.9157894736842107e-06,
+      "loss": 0.6844,
+      "step": 1723
+    },
+    {
+      "epoch": 5.320987654320987,
+      "grad_norm": 6.8125,
+      "learning_rate": 2.905263157894737e-06,
+      "loss": 0.6444,
+      "step": 1724
+    },
+    {
+      "epoch": 5.324074074074074,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.8947368421052634e-06,
+      "loss": 0.7107,
+      "step": 1725
+    },
+    {
+      "epoch": 5.327160493827161,
+      "grad_norm": 6.65625,
+      "learning_rate": 2.88421052631579e-06,
+      "loss": 0.6566,
+      "step": 1726
+    },
+    {
+      "epoch": 5.330246913580247,
+      "grad_norm": 7.0,
+      "learning_rate": 2.8736842105263164e-06,
+      "loss": 0.6824,
+      "step": 1727
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.8631578947368423e-06,
+      "loss": 0.6953,
+      "step": 1728
+    },
+    {
+      "epoch": 5.33641975308642,
+      "grad_norm": 6.75,
+      "learning_rate": 2.8526315789473687e-06,
+      "loss": 0.645,
+      "step": 1729
+    },
+    {
+      "epoch": 5.339506172839506,
+      "grad_norm": 6.84375,
+      "learning_rate": 2.842105263157895e-06,
+      "loss": 0.6893,
+      "step": 1730
+    },
+    {
+      "epoch": 5.342592592592593,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.8315789473684213e-06,
+      "loss": 0.7241,
+      "step": 1731
+    },
+    {
+      "epoch": 5.345679012345679,
+      "grad_norm": 7.1875,
+      "learning_rate": 2.8210526315789476e-06,
+      "loss": 0.6449,
+      "step": 1732
+    },
+    {
+      "epoch": 5.348765432098766,
+      "grad_norm": 6.3125,
+      "learning_rate": 2.810526315789474e-06,
+      "loss": 0.7601,
+      "step": 1733
+    },
+    {
+      "epoch": 5.351851851851852,
+      "grad_norm": 8.875,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.7075,
+      "step": 1734
+    },
+    {
+      "epoch": 5.354938271604938,
+      "grad_norm": 6.59375,
+      "learning_rate": 2.789473684210526e-06,
+      "loss": 0.6246,
+      "step": 1735
+    },
+    {
+      "epoch": 5.3580246913580245,
+      "grad_norm": 8.0625,
+      "learning_rate": 2.7789473684210525e-06,
+      "loss": 0.7088,
+      "step": 1736
+    },
+    {
+      "epoch": 5.361111111111111,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.7684210526315793e-06,
+      "loss": 0.671,
+      "step": 1737
+    },
+    {
+      "epoch": 5.364197530864198,
+      "grad_norm": 6.875,
+      "learning_rate": 2.7578947368421056e-06,
+      "loss": 0.6831,
+      "step": 1738
+    },
+    {
+      "epoch": 5.367283950617284,
+      "grad_norm": 7.90625,
+      "learning_rate": 2.747368421052632e-06,
+      "loss": 0.6523,
+      "step": 1739
+    },
+    {
+      "epoch": 5.37037037037037,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.7368421052631583e-06,
+      "loss": 0.7088,
+      "step": 1740
+    },
+    {
+      "epoch": 5.3734567901234565,
+      "grad_norm": 6.84375,
+      "learning_rate": 2.7263157894736846e-06,
+      "loss": 0.6296,
+      "step": 1741
+    },
+    {
+      "epoch": 5.3765432098765435,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.7157894736842105e-06,
+      "loss": 0.6585,
+      "step": 1742
+    },
+    {
+      "epoch": 5.37962962962963,
+      "grad_norm": 8.625,
+      "learning_rate": 2.705263157894737e-06,
+      "loss": 0.6473,
+      "step": 1743
+    },
+    {
+      "epoch": 5.382716049382716,
+      "grad_norm": 7.1875,
+      "learning_rate": 2.694736842105263e-06,
+      "loss": 0.6367,
+      "step": 1744
+    },
+    {
+      "epoch": 5.385802469135802,
+      "grad_norm": 6.6875,
+      "learning_rate": 2.68421052631579e-06,
+      "loss": 0.6815,
+      "step": 1745
+    },
+    {
+      "epoch": 5.388888888888889,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.6736842105263162e-06,
+      "loss": 0.7038,
+      "step": 1746
+    },
+    {
+      "epoch": 5.3919753086419755,
+      "grad_norm": 6.96875,
+      "learning_rate": 2.6631578947368426e-06,
+      "loss": 0.6817,
+      "step": 1747
+    },
+    {
+      "epoch": 5.395061728395062,
+      "grad_norm": 9.625,
+      "learning_rate": 2.652631578947369e-06,
+      "loss": 0.6896,
+      "step": 1748
+    },
+    {
+      "epoch": 5.398148148148148,
+      "grad_norm": 6.59375,
+      "learning_rate": 2.6421052631578948e-06,
+      "loss": 0.7078,
+      "step": 1749
+    },
+    {
+      "epoch": 5.401234567901234,
+      "grad_norm": 7.21875,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 0.6555,
+      "step": 1750
+    },
+    {
+      "epoch": 5.404320987654321,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.6210526315789474e-06,
+      "loss": 0.6807,
+      "step": 1751
+    },
+    {
+      "epoch": 5.407407407407407,
+      "grad_norm": 7.71875,
+      "learning_rate": 2.6105263157894738e-06,
+      "loss": 0.7198,
+      "step": 1752
+    },
+    {
+      "epoch": 5.410493827160494,
+      "grad_norm": 6.90625,
+      "learning_rate": 2.6e-06,
+      "loss": 0.6837,
+      "step": 1753
+    },
+    {
+      "epoch": 5.41358024691358,
+      "grad_norm": 7.59375,
+      "learning_rate": 2.589473684210527e-06,
+      "loss": 0.6732,
+      "step": 1754
+    },
+    {
+      "epoch": 5.416666666666667,
+      "grad_norm": 8.5,
+      "learning_rate": 2.578947368421053e-06,
+      "loss": 0.6507,
+      "step": 1755
+    },
+    {
+      "epoch": 5.419753086419753,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.568421052631579e-06,
+      "loss": 0.6646,
+      "step": 1756
+    },
+    {
+      "epoch": 5.422839506172839,
+      "grad_norm": 6.625,
+      "learning_rate": 2.5578947368421054e-06,
+      "loss": 0.7129,
+      "step": 1757
+    },
+    {
+      "epoch": 5.425925925925926,
+      "grad_norm": 8.125,
+      "learning_rate": 2.5473684210526317e-06,
+      "loss": 0.6912,
+      "step": 1758
+    },
+    {
+      "epoch": 5.429012345679013,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.536842105263158e-06,
+      "loss": 0.7024,
+      "step": 1759
+    },
+    {
+      "epoch": 5.432098765432099,
+      "grad_norm": 6.59375,
+      "learning_rate": 2.5263157894736844e-06,
+      "loss": 0.6796,
+      "step": 1760
+    },
+    {
+      "epoch": 5.435185185185185,
+      "grad_norm": 7.59375,
+      "learning_rate": 2.5157894736842107e-06,
+      "loss": 0.6644,
+      "step": 1761
+    },
+    {
+      "epoch": 5.438271604938271,
+      "grad_norm": 6.3125,
+      "learning_rate": 2.5052631578947375e-06,
+      "loss": 0.6836,
+      "step": 1762
+    },
+    {
+      "epoch": 5.441358024691358,
+      "grad_norm": 6.53125,
+      "learning_rate": 2.4947368421052634e-06,
+      "loss": 0.6515,
+      "step": 1763
+    },
+    {
+      "epoch": 5.444444444444445,
+      "grad_norm": 8.5625,
+      "learning_rate": 2.4842105263157897e-06,
+      "loss": 0.6193,
+      "step": 1764
+    },
+    {
+      "epoch": 5.447530864197531,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.473684210526316e-06,
+      "loss": 0.7108,
+      "step": 1765
+    },
+    {
+      "epoch": 5.450617283950617,
+      "grad_norm": 8.0625,
+      "learning_rate": 2.4631578947368424e-06,
+      "loss": 0.6578,
+      "step": 1766
+    },
+    {
+      "epoch": 5.453703703703704,
+      "grad_norm": 8.125,
+      "learning_rate": 2.4526315789473687e-06,
+      "loss": 0.7109,
+      "step": 1767
+    },
+    {
+      "epoch": 5.45679012345679,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.442105263157895e-06,
+      "loss": 0.6846,
+      "step": 1768
+    },
+    {
+      "epoch": 5.459876543209877,
+      "grad_norm": 6.15625,
+      "learning_rate": 2.4315789473684213e-06,
+      "loss": 0.7107,
+      "step": 1769
+    },
+    {
+      "epoch": 5.462962962962963,
+      "grad_norm": 7.53125,
+      "learning_rate": 2.4210526315789477e-06,
+      "loss": 0.678,
+      "step": 1770
+    },
+    {
+      "epoch": 5.466049382716049,
+      "grad_norm": 7.28125,
+      "learning_rate": 2.410526315789474e-06,
+      "loss": 0.6737,
+      "step": 1771
+    },
+    {
+      "epoch": 5.469135802469136,
+      "grad_norm": 10.5,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.6429,
+      "step": 1772
+    },
+    {
+      "epoch": 5.472222222222222,
+      "grad_norm": 9.0625,
+      "learning_rate": 2.3894736842105266e-06,
+      "loss": 0.6648,
+      "step": 1773
+    },
+    {
+      "epoch": 5.4753086419753085,
+      "grad_norm": 6.4375,
+      "learning_rate": 2.378947368421053e-06,
+      "loss": 0.6735,
+      "step": 1774
+    },
+    {
+      "epoch": 5.478395061728395,
+      "grad_norm": 8.5,
+      "learning_rate": 2.368421052631579e-06,
+      "loss": 0.671,
+      "step": 1775
+    },
+    {
+      "epoch": 5.481481481481482,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.357894736842105e-06,
+      "loss": 0.7535,
+      "step": 1776
+    },
+    {
+      "epoch": 5.484567901234568,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.347368421052632e-06,
+      "loss": 0.7257,
+      "step": 1777
+    },
+    {
+      "epoch": 5.487654320987654,
+      "grad_norm": 9.375,
+      "learning_rate": 2.3368421052631583e-06,
+      "loss": 0.736,
+      "step": 1778
+    },
+    {
+      "epoch": 5.4907407407407405,
+      "grad_norm": 7.40625,
+      "learning_rate": 2.326315789473684e-06,
+      "loss": 0.7064,
+      "step": 1779
+    },
+    {
+      "epoch": 5.493827160493828,
+      "grad_norm": 9.625,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 0.6818,
+      "step": 1780
+    },
+    {
+      "epoch": 5.496913580246914,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.3052631578947373e-06,
+      "loss": 0.6874,
+      "step": 1781
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 6.875,
+      "learning_rate": 2.294736842105263e-06,
+      "loss": 0.6931,
+      "step": 1782
+    },
+    {
+      "epoch": 5.503086419753086,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.2842105263157895e-06,
+      "loss": 0.7058,
+      "step": 1783
+    },
+    {
+      "epoch": 5.506172839506172,
+      "grad_norm": 6.84375,
+      "learning_rate": 2.273684210526316e-06,
+      "loss": 0.6413,
+      "step": 1784
+    },
+    {
+      "epoch": 5.5092592592592595,
+      "grad_norm": 7.1875,
+      "learning_rate": 2.2631578947368426e-06,
+      "loss": 0.677,
+      "step": 1785
+    },
+    {
+      "epoch": 5.512345679012346,
+      "grad_norm": 7.9375,
+      "learning_rate": 2.2526315789473685e-06,
+      "loss": 0.6756,
+      "step": 1786
+    },
+    {
+      "epoch": 5.515432098765432,
+      "grad_norm": 6.96875,
+      "learning_rate": 2.242105263157895e-06,
+      "loss": 0.6926,
+      "step": 1787
+    },
+    {
+      "epoch": 5.518518518518518,
+      "grad_norm": 12.125,
+      "learning_rate": 2.231578947368421e-06,
+      "loss": 0.6486,
+      "step": 1788
+    },
+    {
+      "epoch": 5.521604938271605,
+      "grad_norm": 10.5,
+      "learning_rate": 2.221052631578948e-06,
+      "loss": 0.6788,
+      "step": 1789
+    },
+    {
+      "epoch": 5.5246913580246915,
+      "grad_norm": 6.65625,
+      "learning_rate": 2.2105263157894738e-06,
+      "loss": 0.7425,
+      "step": 1790
+    },
+    {
+      "epoch": 5.527777777777778,
+      "grad_norm": 6.84375,
+      "learning_rate": 2.2e-06,
+      "loss": 0.6591,
+      "step": 1791
+    },
+    {
+      "epoch": 5.530864197530864,
+      "grad_norm": 6.53125,
+      "learning_rate": 2.1894736842105264e-06,
+      "loss": 0.6963,
+      "step": 1792
+    },
+    {
+      "epoch": 5.533950617283951,
+      "grad_norm": 9.3125,
+      "learning_rate": 2.1789473684210528e-06,
+      "loss": 0.7211,
+      "step": 1793
+    },
+    {
+      "epoch": 5.537037037037037,
+      "grad_norm": 12.75,
+      "learning_rate": 2.168421052631579e-06,
+      "loss": 0.7562,
+      "step": 1794
+    },
+    {
+      "epoch": 5.540123456790123,
+      "grad_norm": 7.4375,
+      "learning_rate": 2.1578947368421054e-06,
+      "loss": 0.6408,
+      "step": 1795
+    },
+    {
+      "epoch": 5.54320987654321,
+      "grad_norm": 7.09375,
+      "learning_rate": 2.1473684210526317e-06,
+      "loss": 0.7059,
+      "step": 1796
+    },
+    {
+      "epoch": 5.546296296296296,
+      "grad_norm": 9.125,
+      "learning_rate": 2.136842105263158e-06,
+      "loss": 0.671,
+      "step": 1797
+    },
+    {
+      "epoch": 5.549382716049383,
+      "grad_norm": 6.1875,
+      "learning_rate": 2.1263157894736844e-06,
+      "loss": 0.6675,
+      "step": 1798
+    },
+    {
+      "epoch": 5.552469135802469,
+      "grad_norm": 8.25,
+      "learning_rate": 2.1157894736842107e-06,
+      "loss": 0.6654,
+      "step": 1799
+    },
+    {
+      "epoch": 5.555555555555555,
+      "grad_norm": 7.125,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.6614,
+      "step": 1800
+    },
+    {
+      "epoch": 5.5586419753086425,
+      "grad_norm": 6.53125,
+      "learning_rate": 2.0947368421052634e-06,
+      "loss": 0.7215,
+      "step": 1801
+    },
+    {
+      "epoch": 5.561728395061729,
+      "grad_norm": 7.78125,
+      "learning_rate": 2.0842105263157897e-06,
+      "loss": 0.6477,
+      "step": 1802
+    },
+    {
+      "epoch": 5.564814814814815,
+      "grad_norm": 7.03125,
+      "learning_rate": 2.073684210526316e-06,
+      "loss": 0.6691,
+      "step": 1803
+    },
+    {
+      "epoch": 5.567901234567901,
+      "grad_norm": 6.6875,
+      "learning_rate": 2.0631578947368424e-06,
+      "loss": 0.7373,
+      "step": 1804
+    },
+    {
+      "epoch": 5.570987654320987,
+      "grad_norm": 8.75,
+      "learning_rate": 2.0526315789473687e-06,
+      "loss": 0.6439,
+      "step": 1805
+    },
+    {
+      "epoch": 5.574074074074074,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.042105263157895e-06,
+      "loss": 0.614,
+      "step": 1806
+    },
+    {
+      "epoch": 5.577160493827161,
+      "grad_norm": 6.9375,
+      "learning_rate": 2.031578947368421e-06,
+      "loss": 0.669,
+      "step": 1807
+    },
+    {
+      "epoch": 5.580246913580247,
+      "grad_norm": 7.09375,
+      "learning_rate": 2.0210526315789477e-06,
+      "loss": 0.6497,
+      "step": 1808
+    },
+    {
+      "epoch": 5.583333333333333,
+      "grad_norm": 8.25,
+      "learning_rate": 2.010526315789474e-06,
+      "loss": 0.665,
+      "step": 1809
+    },
+    {
+      "epoch": 5.58641975308642,
+      "grad_norm": 7.28125,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.6645,
+      "step": 1810
+    },
+    {
+      "epoch": 5.589506172839506,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.9894736842105262e-06,
+      "loss": 0.6915,
+      "step": 1811
+    },
+    {
+      "epoch": 5.592592592592593,
+      "grad_norm": 8.5,
+      "learning_rate": 1.978947368421053e-06,
+      "loss": 0.6735,
+      "step": 1812
+    },
+    {
+      "epoch": 5.595679012345679,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.9684210526315793e-06,
+      "loss": 0.7362,
+      "step": 1813
+    },
+    {
+      "epoch": 5.598765432098766,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.9578947368421052e-06,
+      "loss": 0.6946,
+      "step": 1814
+    },
+    {
+      "epoch": 5.601851851851852,
+      "grad_norm": 18.125,
+      "learning_rate": 1.9473684210526315e-06,
+      "loss": 0.692,
+      "step": 1815
+    },
+    {
+      "epoch": 5.604938271604938,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.936842105263158e-06,
+      "loss": 0.6878,
+      "step": 1816
+    },
+    {
+      "epoch": 5.6080246913580245,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.9263157894736846e-06,
+      "loss": 0.7026,
+      "step": 1817
+    },
+    {
+      "epoch": 5.611111111111111,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.9157894736842105e-06,
+      "loss": 0.723,
+      "step": 1818
+    },
+    {
+      "epoch": 5.614197530864198,
+      "grad_norm": 8.625,
+      "learning_rate": 1.905263157894737e-06,
+      "loss": 0.6513,
+      "step": 1819
+    },
+    {
+      "epoch": 5.617283950617284,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.8947368421052634e-06,
+      "loss": 0.6548,
+      "step": 1820
+    },
+    {
+      "epoch": 5.62037037037037,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8842105263157895e-06,
+      "loss": 0.6441,
+      "step": 1821
+    },
+    {
+      "epoch": 5.6234567901234565,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.8736842105263158e-06,
+      "loss": 0.6594,
+      "step": 1822
+    },
+    {
+      "epoch": 5.6265432098765435,
+      "grad_norm": 8.25,
+      "learning_rate": 1.8631578947368424e-06,
+      "loss": 0.6597,
+      "step": 1823
+    },
+    {
+      "epoch": 5.62962962962963,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.8526315789473687e-06,
+      "loss": 0.6819,
+      "step": 1824
+    },
+    {
+      "epoch": 5.632716049382716,
+      "grad_norm": 8.625,
+      "learning_rate": 1.8421052631578948e-06,
+      "loss": 0.6935,
+      "step": 1825
+    },
+    {
+      "epoch": 5.635802469135802,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.8315789473684211e-06,
+      "loss": 0.6894,
+      "step": 1826
+    },
+    {
+      "epoch": 5.638888888888889,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8210526315789475e-06,
+      "loss": 0.6589,
+      "step": 1827
+    },
+    {
+      "epoch": 5.6419753086419755,
+      "grad_norm": 7.125,
+      "learning_rate": 1.810526315789474e-06,
+      "loss": 0.7514,
+      "step": 1828
+    },
+    {
+      "epoch": 5.645061728395062,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.6925,
+      "step": 1829
+    },
+    {
+      "epoch": 5.648148148148148,
+      "grad_norm": 7.28125,
+      "learning_rate": 1.7894736842105265e-06,
+      "loss": 0.6444,
+      "step": 1830
+    },
+    {
+      "epoch": 5.651234567901234,
+      "grad_norm": 7.25,
+      "learning_rate": 1.7789473684210528e-06,
+      "loss": 0.654,
+      "step": 1831
+    },
+    {
+      "epoch": 5.654320987654321,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.768421052631579e-06,
+      "loss": 0.6307,
+      "step": 1832
+    },
+    {
+      "epoch": 5.657407407407407,
+      "grad_norm": 8.25,
+      "learning_rate": 1.7578947368421054e-06,
+      "loss": 0.6624,
+      "step": 1833
+    },
+    {
+      "epoch": 5.660493827160494,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.7473684210526318e-06,
+      "loss": 0.6817,
+      "step": 1834
+    },
+    {
+      "epoch": 5.66358024691358,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.736842105263158e-06,
+      "loss": 0.7157,
+      "step": 1835
+    },
+    {
+      "epoch": 5.666666666666667,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.7263157894736842e-06,
+      "loss": 0.6196,
+      "step": 1836
+    },
+    {
+      "epoch": 5.669753086419753,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.7157894736842107e-06,
+      "loss": 0.6899,
+      "step": 1837
+    },
+    {
+      "epoch": 5.672839506172839,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.705263157894737e-06,
+      "loss": 0.6615,
+      "step": 1838
+    },
+    {
+      "epoch": 5.675925925925926,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.6947368421052632e-06,
+      "loss": 0.7287,
+      "step": 1839
+    },
+    {
+      "epoch": 5.679012345679013,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.6842105263157895e-06,
+      "loss": 0.6757,
+      "step": 1840
+    },
+    {
+      "epoch": 5.682098765432099,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.673684210526316e-06,
+      "loss": 0.7142,
+      "step": 1841
+    },
+    {
+      "epoch": 5.685185185185185,
+      "grad_norm": 7.625,
+      "learning_rate": 1.6631578947368424e-06,
+      "loss": 0.6658,
+      "step": 1842
+    },
+    {
+      "epoch": 5.688271604938271,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.6526315789473685e-06,
+      "loss": 0.6265,
+      "step": 1843
+    },
+    {
+      "epoch": 5.6913580246913575,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.6421052631578948e-06,
+      "loss": 0.7204,
+      "step": 1844
+    },
+    {
+      "epoch": 5.694444444444445,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.6315789473684212e-06,
+      "loss": 0.6732,
+      "step": 1845
+    },
+    {
+      "epoch": 5.697530864197531,
+      "grad_norm": 7.75,
+      "learning_rate": 1.6210526315789473e-06,
+      "loss": 0.7194,
+      "step": 1846
+    },
+    {
+      "epoch": 5.700617283950617,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.6105263157894738e-06,
+      "loss": 0.6641,
+      "step": 1847
+    },
+    {
+      "epoch": 5.703703703703704,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.6991,
+      "step": 1848
+    },
+    {
+      "epoch": 5.70679012345679,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.5894736842105265e-06,
+      "loss": 0.6792,
+      "step": 1849
+    },
+    {
+      "epoch": 5.709876543209877,
+      "grad_norm": 9.75,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 0.7226,
+      "step": 1850
+    },
+    {
+      "epoch": 5.712962962962963,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.5684210526315791e-06,
+      "loss": 0.6645,
+      "step": 1851
+    },
+    {
+      "epoch": 5.716049382716049,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.5578947368421054e-06,
+      "loss": 0.7099,
+      "step": 1852
+    },
+    {
+      "epoch": 5.719135802469136,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.5473684210526316e-06,
+      "loss": 0.6751,
+      "step": 1853
+    },
+    {
+      "epoch": 5.722222222222222,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.5368421052631579e-06,
+      "loss": 0.6844,
+      "step": 1854
+    },
+    {
+      "epoch": 5.7253086419753085,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.5263157894736844e-06,
+      "loss": 0.7015,
+      "step": 1855
+    },
+    {
+      "epoch": 5.728395061728395,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.5157894736842108e-06,
+      "loss": 0.6747,
+      "step": 1856
+    },
+    {
+      "epoch": 5.731481481481482,
+      "grad_norm": 9.875,
+      "learning_rate": 1.5052631578947369e-06,
+      "loss": 0.6968,
+      "step": 1857
+    },
+    {
+      "epoch": 5.734567901234568,
+      "grad_norm": 6.625,
+      "learning_rate": 1.4947368421052632e-06,
+      "loss": 0.7261,
+      "step": 1858
+    },
+    {
+      "epoch": 5.737654320987654,
+      "grad_norm": 7.0,
+      "learning_rate": 1.4842105263157897e-06,
+      "loss": 0.6448,
+      "step": 1859
+    },
+    {
+      "epoch": 5.7407407407407405,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.4736842105263159e-06,
+      "loss": 0.682,
+      "step": 1860
+    },
+    {
+      "epoch": 5.743827160493828,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.4631578947368422e-06,
+      "loss": 0.6488,
+      "step": 1861
+    },
+    {
+      "epoch": 5.746913580246914,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.4526315789473685e-06,
+      "loss": 0.6384,
+      "step": 1862
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.442105263157895e-06,
+      "loss": 0.6511,
+      "step": 1863
+    },
+    {
+      "epoch": 5.753086419753086,
+      "grad_norm": 6.375,
+      "learning_rate": 1.4315789473684212e-06,
+      "loss": 0.6547,
+      "step": 1864
+    },
+    {
+      "epoch": 5.756172839506172,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.4210526315789475e-06,
+      "loss": 0.6784,
+      "step": 1865
+    },
+    {
+      "epoch": 5.7592592592592595,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.4105263157894738e-06,
+      "loss": 0.7109,
+      "step": 1866
+    },
+    {
+      "epoch": 5.762345679012346,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.6963,
+      "step": 1867
+    },
+    {
+      "epoch": 5.765432098765432,
+      "grad_norm": 8.625,
+      "learning_rate": 1.3894736842105263e-06,
+      "loss": 0.6133,
+      "step": 1868
+    },
+    {
+      "epoch": 5.768518518518518,
+      "grad_norm": 8.125,
+      "learning_rate": 1.3789473684210528e-06,
+      "loss": 0.6867,
+      "step": 1869
+    },
+    {
+      "epoch": 5.771604938271605,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.3684210526315791e-06,
+      "loss": 0.6938,
+      "step": 1870
+    },
+    {
+      "epoch": 5.7746913580246915,
+      "grad_norm": 10.375,
+      "learning_rate": 1.3578947368421052e-06,
+      "loss": 0.6997,
+      "step": 1871
+    },
+    {
+      "epoch": 5.777777777777778,
+      "grad_norm": 6.75,
+      "learning_rate": 1.3473684210526316e-06,
+      "loss": 0.6154,
+      "step": 1872
+    },
+    {
+      "epoch": 5.780864197530864,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.3368421052631581e-06,
+      "loss": 0.6765,
+      "step": 1873
+    },
+    {
+      "epoch": 5.783950617283951,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.3263157894736844e-06,
+      "loss": 0.7109,
+      "step": 1874
+    },
+    {
+      "epoch": 5.787037037037037,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.3157894736842106e-06,
+      "loss": 0.6958,
+      "step": 1875
+    },
+    {
+      "epoch": 5.790123456790123,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.3052631578947369e-06,
+      "loss": 0.7064,
+      "step": 1876
+    },
+    {
+      "epoch": 5.79320987654321,
+      "grad_norm": 7.125,
+      "learning_rate": 1.2947368421052634e-06,
+      "loss": 0.6865,
+      "step": 1877
+    },
+    {
+      "epoch": 5.796296296296296,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.2842105263157895e-06,
+      "loss": 0.7251,
+      "step": 1878
+    },
+    {
+      "epoch": 5.799382716049383,
+      "grad_norm": 12.0,
+      "learning_rate": 1.2736842105263159e-06,
+      "loss": 0.6643,
+      "step": 1879
+    },
+    {
+      "epoch": 5.802469135802469,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 0.6347,
+      "step": 1880
+    },
+    {
+      "epoch": 5.805555555555555,
+      "grad_norm": 26.75,
+      "learning_rate": 1.2526315789473687e-06,
+      "loss": 0.733,
+      "step": 1881
+    },
+    {
+      "epoch": 5.8086419753086425,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.2421052631578948e-06,
+      "loss": 0.672,
+      "step": 1882
+    },
+    {
+      "epoch": 5.811728395061729,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.2315789473684212e-06,
+      "loss": 0.6521,
+      "step": 1883
+    },
+    {
+      "epoch": 5.814814814814815,
+      "grad_norm": 6.53125,
+      "learning_rate": 1.2210526315789475e-06,
+      "loss": 0.633,
+      "step": 1884
+    },
+    {
+      "epoch": 5.817901234567901,
+      "grad_norm": 20.25,
+      "learning_rate": 1.2105263157894738e-06,
+      "loss": 0.717,
+      "step": 1885
+    },
+    {
+      "epoch": 5.820987654320987,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.6878,
+      "step": 1886
+    },
+    {
+      "epoch": 5.824074074074074,
+      "grad_norm": 6.75,
+      "learning_rate": 1.1894736842105265e-06,
+      "loss": 0.7346,
+      "step": 1887
+    },
+    {
+      "epoch": 5.827160493827161,
+      "grad_norm": 6.875,
+      "learning_rate": 1.1789473684210526e-06,
+      "loss": 0.6511,
+      "step": 1888
+    },
+    {
+      "epoch": 5.830246913580247,
+      "grad_norm": 33.75,
+      "learning_rate": 1.1684210526315791e-06,
+      "loss": 0.6967,
+      "step": 1889
+    },
+    {
+      "epoch": 5.833333333333333,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.1578947368421053e-06,
+      "loss": 0.6317,
+      "step": 1890
+    },
+    {
+      "epoch": 5.83641975308642,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.1473684210526316e-06,
+      "loss": 0.6564,
+      "step": 1891
+    },
+    {
+      "epoch": 5.839506172839506,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.136842105263158e-06,
+      "loss": 0.709,
+      "step": 1892
+    },
+    {
+      "epoch": 5.842592592592593,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.1263157894736842e-06,
+      "loss": 0.6662,
+      "step": 1893
+    },
+    {
+      "epoch": 5.845679012345679,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.1157894736842106e-06,
+      "loss": 0.709,
+      "step": 1894
+    },
+    {
+      "epoch": 5.848765432098766,
+      "grad_norm": 7.0,
+      "learning_rate": 1.1052631578947369e-06,
+      "loss": 0.6444,
+      "step": 1895
+    },
+    {
+      "epoch": 5.851851851851852,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.0947368421052632e-06,
+      "loss": 0.672,
+      "step": 1896
+    },
+    {
+      "epoch": 5.854938271604938,
+      "grad_norm": 6.5,
+      "learning_rate": 1.0842105263157895e-06,
+      "loss": 0.639,
+      "step": 1897
+    },
+    {
+      "epoch": 5.8580246913580245,
+      "grad_norm": 6.65625,
+      "learning_rate": 1.0736842105263159e-06,
+      "loss": 0.6602,
+      "step": 1898
+    },
+    {
+      "epoch": 5.861111111111111,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.0631578947368422e-06,
+      "loss": 0.6488,
+      "step": 1899
+    },
+    {
+      "epoch": 5.864197530864198,
+      "grad_norm": 6.375,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.7208,
+      "step": 1900
+    },
+    {
+      "epoch": 5.867283950617284,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.0421052631578949e-06,
+      "loss": 0.6596,
+      "step": 1901
+    },
+    {
+      "epoch": 5.87037037037037,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.0315789473684212e-06,
+      "loss": 0.6958,
+      "step": 1902
+    },
+    {
+      "epoch": 5.8734567901234565,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.0210526315789475e-06,
+      "loss": 0.6502,
+      "step": 1903
+    },
+    {
+      "epoch": 5.8765432098765435,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.0105263157894738e-06,
+      "loss": 0.6697,
+      "step": 1904
+    },
+    {
+      "epoch": 5.87962962962963,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7013,
+      "step": 1905
+    },
+    {
+      "epoch": 5.882716049382716,
+      "grad_norm": 7.625,
+      "learning_rate": 9.894736842105265e-07,
+      "loss": 0.7201,
+      "step": 1906
+    },
+    {
+      "epoch": 5.885802469135802,
+      "grad_norm": 8.5,
+      "learning_rate": 9.789473684210526e-07,
+      "loss": 0.6887,
+      "step": 1907
+    },
+    {
+      "epoch": 5.888888888888889,
+      "grad_norm": 7.8125,
+      "learning_rate": 9.68421052631579e-07,
+      "loss": 0.6712,
+      "step": 1908
+    },
+    {
+      "epoch": 5.8919753086419755,
+      "grad_norm": 7.65625,
+      "learning_rate": 9.578947368421053e-07,
+      "loss": 0.6914,
+      "step": 1909
+    },
+    {
+      "epoch": 5.895061728395062,
+      "grad_norm": 6.90625,
+      "learning_rate": 9.473684210526317e-07,
+      "loss": 0.6855,
+      "step": 1910
+    },
+    {
+      "epoch": 5.898148148148148,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.368421052631579e-07,
+      "loss": 0.682,
+      "step": 1911
+    },
+    {
+      "epoch": 5.901234567901234,
+      "grad_norm": 7.46875,
+      "learning_rate": 9.263157894736844e-07,
+      "loss": 0.6891,
+      "step": 1912
+    },
+    {
+      "epoch": 5.904320987654321,
+      "grad_norm": 7.0625,
+      "learning_rate": 9.157894736842106e-07,
+      "loss": 0.6756,
+      "step": 1913
+    },
+    {
+      "epoch": 5.907407407407407,
+      "grad_norm": 7.78125,
+      "learning_rate": 9.05263157894737e-07,
+      "loss": 0.6431,
+      "step": 1914
+    },
+    {
+      "epoch": 5.910493827160494,
+      "grad_norm": 7.59375,
+      "learning_rate": 8.947368421052632e-07,
+      "loss": 0.7237,
+      "step": 1915
+    },
+    {
+      "epoch": 5.91358024691358,
+      "grad_norm": 7.4375,
+      "learning_rate": 8.842105263157895e-07,
+      "loss": 0.6984,
+      "step": 1916
+    },
+    {
+      "epoch": 5.916666666666667,
+      "grad_norm": 8.8125,
+      "learning_rate": 8.736842105263159e-07,
+      "loss": 0.7097,
+      "step": 1917
+    },
+    {
+      "epoch": 5.919753086419753,
+      "grad_norm": 7.8125,
+      "learning_rate": 8.631578947368421e-07,
+      "loss": 0.7156,
+      "step": 1918
+    },
+    {
+      "epoch": 5.922839506172839,
+      "grad_norm": 8.6875,
+      "learning_rate": 8.526315789473685e-07,
+      "loss": 0.6476,
+      "step": 1919
+    },
+    {
+      "epoch": 5.925925925925926,
+      "grad_norm": 7.6875,
+      "learning_rate": 8.421052631578948e-07,
+      "loss": 0.6789,
+      "step": 1920
+    },
+    {
+      "epoch": 5.929012345679013,
+      "grad_norm": 6.625,
+      "learning_rate": 8.315789473684212e-07,
+      "loss": 0.6942,
+      "step": 1921
+    },
+    {
+      "epoch": 5.932098765432099,
+      "grad_norm": 6.25,
+      "learning_rate": 8.210526315789474e-07,
+      "loss": 0.6114,
+      "step": 1922
+    },
+    {
+      "epoch": 5.935185185185185,
+      "grad_norm": 7.3125,
+      "learning_rate": 8.105263157894736e-07,
+      "loss": 0.6643,
+      "step": 1923
+    },
+    {
+      "epoch": 5.938271604938271,
+      "grad_norm": 7.34375,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.7205,
+      "step": 1924
+    },
+    {
+      "epoch": 5.9413580246913575,
+      "grad_norm": 6.1875,
+      "learning_rate": 7.894736842105263e-07,
+      "loss": 0.6792,
+      "step": 1925
+    },
+    {
+      "epoch": 5.944444444444445,
+      "grad_norm": 9.375,
+      "learning_rate": 7.789473684210527e-07,
+      "loss": 0.6937,
+      "step": 1926
+    },
+    {
+      "epoch": 5.947530864197531,
+      "grad_norm": 7.5625,
+      "learning_rate": 7.684210526315789e-07,
+      "loss": 0.6875,
+      "step": 1927
+    },
+    {
+      "epoch": 5.950617283950617,
+      "grad_norm": 7.21875,
+      "learning_rate": 7.578947368421054e-07,
+      "loss": 0.6943,
+      "step": 1928
+    },
+    {
+      "epoch": 5.953703703703704,
+      "grad_norm": 8.5,
+      "learning_rate": 7.473684210526316e-07,
+      "loss": 0.6622,
+      "step": 1929
+    },
+    {
+      "epoch": 5.95679012345679,
+      "grad_norm": 9.875,
+      "learning_rate": 7.368421052631579e-07,
+      "loss": 0.6518,
+      "step": 1930
+    },
+    {
+      "epoch": 5.959876543209877,
+      "grad_norm": 7.40625,
+      "learning_rate": 7.263157894736843e-07,
+      "loss": 0.719,
+      "step": 1931
+    },
+    {
+      "epoch": 5.962962962962963,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.157894736842106e-07,
+      "loss": 0.6681,
+      "step": 1932
+    },
+    {
+      "epoch": 5.966049382716049,
+      "grad_norm": 7.21875,
+      "learning_rate": 7.052631578947369e-07,
+      "loss": 0.6528,
+      "step": 1933
+    },
+    {
+      "epoch": 5.969135802469136,
+      "grad_norm": 7.53125,
+      "learning_rate": 6.947368421052631e-07,
+      "loss": 0.6624,
+      "step": 1934
+    },
+    {
+      "epoch": 5.972222222222222,
+      "grad_norm": 11.1875,
+      "learning_rate": 6.842105263157896e-07,
+      "loss": 0.7066,
+      "step": 1935
+    },
+    {
+      "epoch": 5.9753086419753085,
+      "grad_norm": 7.3125,
+      "learning_rate": 6.736842105263158e-07,
+      "loss": 0.7439,
+      "step": 1936
+    },
+    {
+      "epoch": 5.978395061728395,
+      "grad_norm": 6.8125,
+      "learning_rate": 6.631578947368422e-07,
+      "loss": 0.7277,
+      "step": 1937
+    },
+    {
+      "epoch": 5.981481481481482,
+      "grad_norm": 7.15625,
+      "learning_rate": 6.526315789473684e-07,
+      "loss": 0.7332,
+      "step": 1938
+    },
+    {
+      "epoch": 5.984567901234568,
+      "grad_norm": 6.8125,
+      "learning_rate": 6.421052631578948e-07,
+      "loss": 0.6615,
+      "step": 1939
+    },
+    {
+      "epoch": 5.987654320987654,
+      "grad_norm": 7.4375,
+      "learning_rate": 6.315789473684211e-07,
+      "loss": 0.7158,
+      "step": 1940
+    },
+    {
+      "epoch": 5.9907407407407405,
+      "grad_norm": 12.0,
+      "learning_rate": 6.210526315789474e-07,
+      "loss": 0.6889,
+      "step": 1941
+    },
+    {
+      "epoch": 5.993827160493828,
+      "grad_norm": 6.96875,
+      "learning_rate": 6.105263157894738e-07,
+      "loss": 0.6887,
+      "step": 1942
+    },
+    {
+      "epoch": 5.996913580246914,
+      "grad_norm": 8.75,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.6594,
+      "step": 1943
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 8.4375,
+      "learning_rate": 5.894736842105263e-07,
+      "loss": 0.6945,
+      "step": 1944
+    },
+    {
+      "epoch": 6.003086419753086,
+      "grad_norm": 8.875,
+      "learning_rate": 5.789473684210526e-07,
+      "loss": 0.6025,
+      "step": 1945
+    },
+    {
+      "epoch": 6.006172839506172,
+      "grad_norm": 7.59375,
+      "learning_rate": 5.68421052631579e-07,
+      "loss": 0.6842,
+      "step": 1946
+    },
+    {
+      "epoch": 6.0092592592592595,
+      "grad_norm": 7.34375,
+      "learning_rate": 5.578947368421053e-07,
+      "loss": 0.654,
+      "step": 1947
+    },
+    {
+      "epoch": 6.012345679012346,
+      "grad_norm": 8.8125,
+      "learning_rate": 5.473684210526316e-07,
+      "loss": 0.6581,
+      "step": 1948
+    },
+    {
+      "epoch": 6.015432098765432,
+      "grad_norm": 9.0,
+      "learning_rate": 5.368421052631579e-07,
+      "loss": 0.6616,
+      "step": 1949
+    },
+    {
+      "epoch": 6.018518518518518,
+      "grad_norm": 7.875,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 0.6723,
+      "step": 1950
+    },
+    {
+      "epoch": 6.021604938271605,
+      "grad_norm": 6.8125,
+      "learning_rate": 5.157894736842106e-07,
+      "loss": 0.6886,
+      "step": 1951
+    },
+    {
+      "epoch": 6.0246913580246915,
+      "grad_norm": 7.875,
+      "learning_rate": 5.052631578947369e-07,
+      "loss": 0.6914,
+      "step": 1952
+    },
+    {
+      "epoch": 6.027777777777778,
+      "grad_norm": 7.125,
+      "learning_rate": 4.947368421052632e-07,
+      "loss": 0.7147,
+      "step": 1953
+    },
+    {
+      "epoch": 6.030864197530864,
+      "grad_norm": 7.84375,
+      "learning_rate": 4.842105263157895e-07,
+      "loss": 0.6747,
+      "step": 1954
+    },
+    {
+      "epoch": 6.033950617283951,
+      "grad_norm": 7.1875,
+      "learning_rate": 4.7368421052631585e-07,
+      "loss": 0.6463,
+      "step": 1955
+    },
+    {
+      "epoch": 6.037037037037037,
+      "grad_norm": 7.53125,
+      "learning_rate": 4.631578947368422e-07,
+      "loss": 0.7016,
+      "step": 1956
+    },
+    {
+      "epoch": 6.040123456790123,
+      "grad_norm": 6.96875,
+      "learning_rate": 4.526315789473685e-07,
+      "loss": 0.733,
+      "step": 1957
+    },
+    {
+      "epoch": 6.04320987654321,
+      "grad_norm": 7.0,
+      "learning_rate": 4.421052631578947e-07,
+      "loss": 0.6932,
+      "step": 1958
+    },
+    {
+      "epoch": 6.046296296296297,
+      "grad_norm": 6.5,
+      "learning_rate": 4.3157894736842105e-07,
+      "loss": 0.709,
+      "step": 1959
+    },
+    {
+      "epoch": 6.049382716049383,
+      "grad_norm": 7.34375,
+      "learning_rate": 4.210526315789474e-07,
+      "loss": 0.6775,
+      "step": 1960
+    },
+    {
+      "epoch": 6.052469135802469,
+      "grad_norm": 8.8125,
+      "learning_rate": 4.105263157894737e-07,
+      "loss": 0.6507,
+      "step": 1961
+    },
+    {
+      "epoch": 6.055555555555555,
+      "grad_norm": 6.375,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.6522,
+      "step": 1962
+    },
+    {
+      "epoch": 6.058641975308642,
+      "grad_norm": 6.90625,
+      "learning_rate": 3.8947368421052636e-07,
+      "loss": 0.6465,
+      "step": 1963
+    },
+    {
+      "epoch": 6.061728395061729,
+      "grad_norm": 7.625,
+      "learning_rate": 3.789473684210527e-07,
+      "loss": 0.6649,
+      "step": 1964
+    },
+    {
+      "epoch": 6.064814814814815,
+      "grad_norm": 6.75,
+      "learning_rate": 3.6842105263157896e-07,
+      "loss": 0.66,
+      "step": 1965
+    },
+    {
+      "epoch": 6.067901234567901,
+      "grad_norm": 8.5,
+      "learning_rate": 3.578947368421053e-07,
+      "loss": 0.6994,
+      "step": 1966
+    },
+    {
+      "epoch": 6.070987654320987,
+      "grad_norm": 7.5,
+      "learning_rate": 3.4736842105263157e-07,
+      "loss": 0.6642,
+      "step": 1967
+    },
+    {
+      "epoch": 6.074074074074074,
+      "grad_norm": 7.25,
+      "learning_rate": 3.368421052631579e-07,
+      "loss": 0.6858,
+      "step": 1968
+    },
+    {
+      "epoch": 6.077160493827161,
+      "grad_norm": 6.375,
+      "learning_rate": 3.263157894736842e-07,
+      "loss": 0.6914,
+      "step": 1969
+    },
+    {
+      "epoch": 6.080246913580247,
+      "grad_norm": 7.4375,
+      "learning_rate": 3.1578947368421055e-07,
+      "loss": 0.6949,
+      "step": 1970
+    },
+    {
+      "epoch": 6.083333333333333,
+      "grad_norm": 6.65625,
+      "learning_rate": 3.052631578947369e-07,
+      "loss": 0.7099,
+      "step": 1971
+    },
+    {
+      "epoch": 6.08641975308642,
+      "grad_norm": 6.875,
+      "learning_rate": 2.9473684210526315e-07,
+      "loss": 0.6627,
+      "step": 1972
+    },
+    {
+      "epoch": 6.089506172839506,
+      "grad_norm": 25.625,
+      "learning_rate": 2.842105263157895e-07,
+      "loss": 0.6713,
+      "step": 1973
+    },
+    {
+      "epoch": 6.092592592592593,
+      "grad_norm": 6.90625,
+      "learning_rate": 2.736842105263158e-07,
+      "loss": 0.6928,
+      "step": 1974
+    },
+    {
+      "epoch": 6.095679012345679,
+      "grad_norm": 9.625,
+      "learning_rate": 2.6315789473684213e-07,
+      "loss": 0.7051,
+      "step": 1975
+    },
+    {
+      "epoch": 6.098765432098766,
+      "grad_norm": 8.375,
+      "learning_rate": 2.5263157894736846e-07,
+      "loss": 0.6697,
+      "step": 1976
+    },
+    {
+      "epoch": 6.101851851851852,
+      "grad_norm": 6.71875,
+      "learning_rate": 2.4210526315789473e-07,
+      "loss": 0.6925,
+      "step": 1977
+    },
+    {
+      "epoch": 6.104938271604938,
+      "grad_norm": 6.875,
+      "learning_rate": 2.315789473684211e-07,
+      "loss": 0.6245,
+      "step": 1978
+    },
+    {
+      "epoch": 6.1080246913580245,
+      "grad_norm": 7.65625,
+      "learning_rate": 2.2105263157894736e-07,
+      "loss": 0.6174,
+      "step": 1979
+    },
+    {
+      "epoch": 6.111111111111111,
+      "grad_norm": 8.25,
+      "learning_rate": 2.105263157894737e-07,
+      "loss": 0.6994,
+      "step": 1980
+    },
+    {
+      "epoch": 6.114197530864198,
+      "grad_norm": 7.0625,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.6644,
+      "step": 1981
+    },
+    {
+      "epoch": 6.117283950617284,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.8947368421052634e-07,
+      "loss": 0.7637,
+      "step": 1982
+    },
+    {
+      "epoch": 6.12037037037037,
+      "grad_norm": 13.375,
+      "learning_rate": 1.7894736842105265e-07,
+      "loss": 0.7004,
+      "step": 1983
+    },
+    {
+      "epoch": 6.1234567901234565,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.6842105263157895e-07,
+      "loss": 0.6722,
+      "step": 1984
+    },
+    {
+      "epoch": 6.1265432098765435,
+      "grad_norm": 8.875,
+      "learning_rate": 1.5789473684210527e-07,
+      "loss": 0.6542,
+      "step": 1985
+    },
+    {
+      "epoch": 6.12962962962963,
+      "grad_norm": 7.0,
+      "learning_rate": 1.4736842105263158e-07,
+      "loss": 0.6913,
+      "step": 1986
+    },
+    {
+      "epoch": 6.132716049382716,
+      "grad_norm": 8.375,
+      "learning_rate": 1.368421052631579e-07,
+      "loss": 0.6444,
+      "step": 1987
+    },
+    {
+      "epoch": 6.135802469135802,
+      "grad_norm": 9.5,
+      "learning_rate": 1.2631578947368423e-07,
+      "loss": 0.6559,
+      "step": 1988
+    },
+    {
+      "epoch": 6.138888888888889,
+      "grad_norm": 6.5,
+      "learning_rate": 1.1578947368421054e-07,
+      "loss": 0.6911,
+      "step": 1989
+    },
+    {
+      "epoch": 6.1419753086419755,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.0526315789473685e-07,
+      "loss": 0.7319,
+      "step": 1990
+    },
+    {
+      "epoch": 6.145061728395062,
+      "grad_norm": 7.75,
+      "learning_rate": 9.473684210526317e-08,
+      "loss": 0.6858,
+      "step": 1991
+    },
+    {
+      "epoch": 6.148148148148148,
+      "grad_norm": 6.6875,
+      "learning_rate": 8.421052631578947e-08,
+      "loss": 0.6637,
+      "step": 1992
+    },
+    {
+      "epoch": 6.151234567901234,
+      "grad_norm": 7.5,
+      "learning_rate": 7.368421052631579e-08,
+      "loss": 0.6992,
+      "step": 1993
+    },
+    {
+      "epoch": 6.154320987654321,
+      "grad_norm": 6.59375,
+      "learning_rate": 6.315789473684211e-08,
+      "loss": 0.6645,
+      "step": 1994
+    },
+    {
+      "epoch": 6.157407407407407,
+      "grad_norm": 8.125,
+      "learning_rate": 5.263157894736842e-08,
+      "loss": 0.6455,
+      "step": 1995
+    },
+    {
+      "epoch": 6.160493827160494,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.2105263157894737e-08,
+      "loss": 0.6354,
+      "step": 1996
+    },
+    {
+      "epoch": 6.16358024691358,
+      "grad_norm": 6.46875,
+      "learning_rate": 3.157894736842106e-08,
+      "loss": 0.723,
+      "step": 1997
+    },
+    {
+      "epoch": 6.166666666666667,
+      "grad_norm": 7.78125,
+      "learning_rate": 2.1052631578947368e-08,
+      "loss": 0.7276,
+      "step": 1998
+    },
+    {
+      "epoch": 6.169753086419753,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.0526315789473684e-08,
+      "loss": 0.6575,
+      "step": 1999
+    },
+    {
+      "epoch": 6.172839506172839,
+      "grad_norm": 6.65625,
+      "learning_rate": 0.0,
+      "loss": 0.6649,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 7,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 256,
+  "trial_name": null,
+  "trial_params": null
+}