diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -3,7 +3,7 @@
   "best_model_checkpoint": null,
   "epoch": 20.0,
   "eval_steps": 500,
-  "global_step": 1100,
+  "global_step": 1080,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -11,6779 +11,6659 @@
     {
       "epoch": 0.02,
       "learning_rate": 0.0,
-      "loss": 1.4737,
+      "loss": 1.4621,
       "step": 1
     },
     {
       "epoch": 0.04,
       "learning_rate": 3.9647972634112114e-05,
-      "loss": 1.2108,
+      "loss": 1.251,
       "step": 2
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.06,
       "learning_rate": 6.28405498546863e-05,
-      "loss": 1.2581,
+      "loss": 1.2407,
       "step": 3
     },
     {
       "epoch": 0.07,
       "learning_rate": 7.929594526822423e-05,
-      "loss": 1.139,
+      "loss": 1.168,
       "step": 4
     },
     {
       "epoch": 0.09,
       "learning_rate": 9.20597415644702e-05,
-      "loss": 1.1407,
+      "loss": 1.1146,
       "step": 5
     },
     {
       "epoch": 0.11,
       "learning_rate": 0.00010248852248879841,
-      "loss": 1.0303,
+      "loss": 1.1669,
       "step": 6
     },
     {
       "epoch": 0.13,
       "learning_rate": 0.00011130593112397982,
-      "loss": 1.1241,
+      "loss": 1.0961,
       "step": 7
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.00011894391790233632,
-      "loss": 0.9747,
+      "loss": 1.064,
       "step": 8
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.17,
       "learning_rate": 0.0001256810997093726,
-      "loss": 1.0138,
+      "loss": 1.0397,
       "step": 9
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.19,
       "learning_rate": 0.00013170771419858233,
-      "loss": 0.9869,
+      "loss": 1.0446,
       "step": 10
     },
     {
       "epoch": 0.2,
       "learning_rate": 0.00013715945014531372,
-      "loss": 0.983,
+      "loss": 1.0325,
       "step": 11
     },
     {
       "epoch": 0.22,
       "learning_rate": 0.0001421364951229105,
-      "loss": 0.8718,
+      "loss": 0.9755,
       "step": 12
     },
     {
       "epoch": 0.24,
       "learning_rate": 0.00014671493267903954,
-      "loss": 0.8891,
+      "loss": 0.9744,
       "step": 13
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.26,
       "learning_rate": 0.00015095390375809194,
-      "loss": 0.8584,
+      "loss": 0.9678,
       "step": 14
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.28,
       "learning_rate": 0.00015490029141915652,
-      "loss": 0.908,
+      "loss": 0.9336,
       "step": 15
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.3,
       "learning_rate": 0.00015859189053644846,
-      "loss": 0.8369,
+      "loss": 0.9485,
       "step": 16
     },
     {
       "epoch": 0.31,
       "learning_rate": 0.0001620596148728436,
-      "loss": 0.8441,
+      "loss": 0.9815,
       "step": 17
     },
     {
       "epoch": 0.33,
       "learning_rate": 0.0001653290723434847,
-      "loss": 0.8768,
+      "loss": 0.9364,
       "step": 18
     },
     {
       "epoch": 0.35,
       "learning_rate": 0.00016842171380470318,
-      "loss": 0.8319,
+      "loss": 0.9414,
       "step": 19
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.37,
       "learning_rate": 0.00017135568683269444,
-      "loss": 0.8437,
+      "loss": 0.8375,
       "step": 20
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.39,
       "learning_rate": 0.00017414648097866613,
-      "loss": 0.7532,
+      "loss": 0.8996,
       "step": 21
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.41,
       "learning_rate": 0.00017680742277942585,
-      "loss": 0.7573,
+      "loss": 0.8257,
       "step": 22
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.43,
       "learning_rate": 0.0001793500606424591,
-      "loss": 0.7576,
+      "loss": 0.8045,
       "step": 23
     },
     {
       "epoch": 0.44,
       "learning_rate": 0.00018178446775702264,
-      "loss": 0.5811,
+      "loss": 0.8317,
       "step": 24
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.46,
       "learning_rate": 0.0001841194831289404,
-      "loss": 0.707,
+      "loss": 0.8321,
       "step": 25
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.48,
       "learning_rate": 0.00018636290531315168,
-      "loss": 0.6758,
+      "loss": 0.798,
       "step": 26
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.5,
       "learning_rate": 0.00018852164956405888,
-      "loss": 0.5306,
+      "loss": 0.7013,
       "step": 27
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.52,
       "learning_rate": 0.00019060187639220405,
-      "loss": 0.6441,
+      "loss": 0.7181,
       "step": 28
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.54,
       "learning_rate": 0.0001926090975518547,
-      "loss": 0.5951,
+      "loss": 0.7503,
       "step": 29
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.56,
       "learning_rate": 0.00019454826405326863,
-      "loss": 0.6004,
+      "loss": 0.6465,
       "step": 30
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.57,
       "learning_rate": 0.00019642383973823802,
-      "loss": 0.561,
+      "loss": 0.6445,
       "step": 31
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.59,
       "learning_rate": 0.00019823986317056056,
-      "loss": 0.539,
+      "loss": 0.6802,
       "step": 32
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.61,
       "learning_rate": 0.0002,
-      "loss": 0.4335,
+      "loss": 0.6426,
       "step": 33
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.63,
       "learning_rate": 0.0002,
-      "loss": 0.5391,
+      "loss": 0.6134,
       "step": 34
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.65,
       "learning_rate": 0.0002,
-      "loss": 0.5088,
+      "loss": 0.5961,
       "step": 35
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.67,
       "learning_rate": 0.0002,
-      "loss": 0.4994,
+      "loss": 0.7114,
       "step": 36
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.69,
       "learning_rate": 0.0002,
-      "loss": 0.4375,
+      "loss": 0.5714,
       "step": 37
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.7,
       "learning_rate": 0.0002,
-      "loss": 0.4979,
+      "loss": 0.5757,
       "step": 38
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.72,
       "learning_rate": 0.0002,
-      "loss": 0.5967,
+      "loss": 0.4163,
       "step": 39
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.74,
       "learning_rate": 0.0002,
-      "loss": 0.4788,
+      "loss": 0.576,
       "step": 40
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.76,
       "learning_rate": 0.0002,
-      "loss": 0.322,
+      "loss": 0.4897,
       "step": 41
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.78,
       "learning_rate": 0.0002,
-      "loss": 0.3617,
+      "loss": 0.522,
       "step": 42
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.8,
       "learning_rate": 0.0002,
-      "loss": 0.337,
+      "loss": 0.3833,
       "step": 43
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.81,
       "learning_rate": 0.0002,
-      "loss": 0.3428,
+      "loss": 0.4566,
       "step": 44
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.83,
       "learning_rate": 0.0002,
-      "loss": 0.3332,
+      "loss": 0.4844,
       "step": 45
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.85,
       "learning_rate": 0.0002,
-      "loss": 0.3049,
+      "loss": 0.5017,
       "step": 46
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.87,
       "learning_rate": 0.0002,
-      "loss": 0.1916,
+      "loss": 0.3339,
       "step": 47
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.89,
       "learning_rate": 0.0002,
-      "loss": 0.2789,
+      "loss": 0.4174,
       "step": 48
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.91,
       "learning_rate": 0.0002,
-      "loss": 0.3114,
+      "loss": 0.4256,
       "step": 49
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.93,
       "learning_rate": 0.0002,
-      "loss": 0.3547,
+      "loss": 0.3593,
       "step": 50
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.94,
       "learning_rate": 0.0002,
-      "loss": 0.2256,
+      "loss": 0.3484,
       "step": 51
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.96,
       "learning_rate": 0.0002,
-      "loss": 0.241,
+      "loss": 0.4084,
       "step": 52
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.98,
       "learning_rate": 0.0002,
-      "loss": 0.2266,
+      "loss": 0.3884,
       "step": 53
     },
     {
-      "epoch": 0.98,
+      "epoch": 1.0,
       "learning_rate": 0.0002,
-      "loss": 0.3486,
+      "loss": 0.3284,
       "step": 54
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002,
-      "loss": 0.265,
-      "step": 55
+      "eval_loss": 0.23234400153160095,
+      "eval_runtime": 137.6148,
+      "eval_samples_per_second": 12.433,
+      "eval_steps_per_second": 0.392,
+      "step": 54
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 0.16694895923137665,
-      "eval_runtime": 96.4796,
-      "eval_samples_per_second": 8.986,
-      "eval_steps_per_second": 0.57,
+      "epoch": 1.02,
+      "learning_rate": 0.0002,
+      "loss": 0.2345,
       "step": 55
     },
     {
-      "epoch": 1.02,
+      "epoch": 1.04,
       "learning_rate": 0.0002,
-      "loss": 0.1636,
+      "loss": 0.1947,
       "step": 56
     },
     {
-      "epoch": 1.04,
+      "epoch": 1.06,
       "learning_rate": 0.0002,
-      "loss": 0.1318,
+      "loss": 0.1926,
       "step": 57
     },
     {
-      "epoch": 1.05,
+      "epoch": 1.07,
       "learning_rate": 0.0002,
-      "loss": 0.1962,
+      "loss": 0.1856,
       "step": 58
     },
     {
-      "epoch": 1.07,
+      "epoch": 1.09,
       "learning_rate": 0.0002,
-      "loss": 0.1107,
+      "loss": 0.1538,
       "step": 59
     },
     {
-      "epoch": 1.09,
+      "epoch": 1.11,
       "learning_rate": 0.0002,
-      "loss": 0.1811,
+      "loss": 0.1599,
       "step": 60
     },
     {
-      "epoch": 1.11,
+      "epoch": 1.13,
       "learning_rate": 0.0002,
-      "loss": 0.104,
+      "loss": 0.1503,
       "step": 61
     },
     {
-      "epoch": 1.13,
+      "epoch": 1.15,
       "learning_rate": 0.0002,
-      "loss": 0.1203,
+      "loss": 0.1949,
       "step": 62
     },
     {
-      "epoch": 1.15,
+      "epoch": 1.17,
       "learning_rate": 0.0002,
-      "loss": 0.1434,
+      "loss": 0.1367,
       "step": 63
     },
     {
-      "epoch": 1.16,
+      "epoch": 1.19,
       "learning_rate": 0.0002,
-      "loss": 0.124,
+      "loss": 0.125,
       "step": 64
     },
     {
-      "epoch": 1.18,
+      "epoch": 1.2,
       "learning_rate": 0.0002,
-      "loss": 0.1906,
+      "loss": 0.2634,
       "step": 65
     },
     {
-      "epoch": 1.2,
+      "epoch": 1.22,
       "learning_rate": 0.0002,
-      "loss": 0.1756,
+      "loss": 0.1706,
       "step": 66
     },
     {
-      "epoch": 1.22,
+      "epoch": 1.24,
       "learning_rate": 0.0002,
-      "loss": 0.1128,
+      "loss": 0.1482,
       "step": 67
     },
     {
-      "epoch": 1.24,
+      "epoch": 1.26,
       "learning_rate": 0.0002,
-      "loss": 0.14,
+      "loss": 0.1726,
       "step": 68
     },
     {
-      "epoch": 1.25,
+      "epoch": 1.28,
       "learning_rate": 0.0002,
-      "loss": 0.1985,
+      "loss": 0.2005,
       "step": 69
     },
     {
-      "epoch": 1.27,
+      "epoch": 1.3,
       "learning_rate": 0.0002,
-      "loss": 0.088,
+      "loss": 0.1389,
       "step": 70
     },
     {
-      "epoch": 1.29,
+      "epoch": 1.31,
       "learning_rate": 0.0002,
-      "loss": 0.1121,
+      "loss": 0.1511,
       "step": 71
     },
     {
-      "epoch": 1.31,
+      "epoch": 1.33,
       "learning_rate": 0.0002,
-      "loss": 0.1268,
+      "loss": 0.137,
       "step": 72
     },
     {
-      "epoch": 1.33,
+      "epoch": 1.35,
       "learning_rate": 0.0002,
-      "loss": 0.1002,
+      "loss": 0.1589,
       "step": 73
     },
     {
-      "epoch": 1.35,
+      "epoch": 1.37,
       "learning_rate": 0.0002,
-      "loss": 0.0681,
+      "loss": 0.1268,
       "step": 74
     },
     {
-      "epoch": 1.36,
+      "epoch": 1.39,
       "learning_rate": 0.0002,
-      "loss": 0.074,
+      "loss": 0.1187,
       "step": 75
     },
     {
-      "epoch": 1.38,
+      "epoch": 1.41,
       "learning_rate": 0.0002,
-      "loss": 0.0744,
+      "loss": 0.1458,
       "step": 76
     },
     {
-      "epoch": 1.4,
+      "epoch": 1.43,
       "learning_rate": 0.0002,
-      "loss": 0.0657,
+      "loss": 0.1791,
       "step": 77
     },
     {
-      "epoch": 1.42,
+      "epoch": 1.44,
       "learning_rate": 0.0002,
-      "loss": 0.1125,
+      "loss": 0.1491,
       "step": 78
     },
     {
-      "epoch": 1.44,
+      "epoch": 1.46,
       "learning_rate": 0.0002,
-      "loss": 0.1592,
+      "loss": 0.1334,
       "step": 79
     },
     {
-      "epoch": 1.45,
+      "epoch": 1.48,
       "learning_rate": 0.0002,
-      "loss": 0.1034,
+      "loss": 0.1302,
       "step": 80
     },
     {
-      "epoch": 1.47,
+      "epoch": 1.5,
       "learning_rate": 0.0002,
-      "loss": 0.0739,
+      "loss": 0.0984,
       "step": 81
     },
     {
-      "epoch": 1.49,
+      "epoch": 1.52,
       "learning_rate": 0.0002,
-      "loss": 0.0639,
+      "loss": 0.1044,
       "step": 82
     },
     {
-      "epoch": 1.51,
+      "epoch": 1.54,
       "learning_rate": 0.0002,
-      "loss": 0.0834,
+      "loss": 0.1213,
       "step": 83
     },
     {
-      "epoch": 1.53,
+      "epoch": 1.56,
       "learning_rate": 0.0002,
-      "loss": 0.063,
+      "loss": 0.0859,
       "step": 84
     },
     {
-      "epoch": 1.55,
+      "epoch": 1.57,
       "learning_rate": 0.0002,
-      "loss": 0.114,
+      "loss": 0.1816,
       "step": 85
     },
     {
-      "epoch": 1.56,
+      "epoch": 1.59,
       "learning_rate": 0.0002,
-      "loss": 0.0989,
+      "loss": 0.1104,
       "step": 86
     },
     {
-      "epoch": 1.58,
+      "epoch": 1.61,
       "learning_rate": 0.0002,
-      "loss": 0.112,
+      "loss": 0.1067,
       "step": 87
     },
     {
-      "epoch": 1.6,
+      "epoch": 1.63,
       "learning_rate": 0.0002,
-      "loss": 0.0425,
+      "loss": 0.1222,
       "step": 88
     },
     {
-      "epoch": 1.62,
+      "epoch": 1.65,
       "learning_rate": 0.0002,
-      "loss": 0.0764,
+      "loss": 0.0972,
       "step": 89
     },
     {
-      "epoch": 1.64,
+      "epoch": 1.67,
       "learning_rate": 0.0002,
-      "loss": 0.0672,
+      "loss": 0.0844,
       "step": 90
     },
     {
-      "epoch": 1.65,
+      "epoch": 1.69,
       "learning_rate": 0.0002,
-      "loss": 0.0477,
+      "loss": 0.1204,
       "step": 91
     },
     {
-      "epoch": 1.67,
+      "epoch": 1.7,
       "learning_rate": 0.0002,
-      "loss": 0.0808,
+      "loss": 0.1156,
       "step": 92
     },
     {
-      "epoch": 1.69,
+      "epoch": 1.72,
       "learning_rate": 0.0002,
-      "loss": 0.0695,
+      "loss": 0.0846,
       "step": 93
     },
     {
-      "epoch": 1.71,
+      "epoch": 1.74,
       "learning_rate": 0.0002,
-      "loss": 0.1155,
+      "loss": 0.1278,
       "step": 94
     },
     {
-      "epoch": 1.73,
+      "epoch": 1.76,
       "learning_rate": 0.0002,
-      "loss": 0.0265,
+      "loss": 0.0594,
       "step": 95
     },
     {
-      "epoch": 1.75,
+      "epoch": 1.78,
       "learning_rate": 0.0002,
-      "loss": 0.1261,
+      "loss": 0.0551,
       "step": 96
     },
     {
-      "epoch": 1.76,
+      "epoch": 1.8,
       "learning_rate": 0.0002,
-      "loss": 0.0479,
+      "loss": 0.103,
       "step": 97
     },
     {
-      "epoch": 1.78,
+      "epoch": 1.81,
       "learning_rate": 0.0002,
-      "loss": 0.0346,
+      "loss": 0.1273,
       "step": 98
     },
     {
-      "epoch": 1.8,
+      "epoch": 1.83,
       "learning_rate": 0.0002,
-      "loss": 0.0484,
+      "loss": 0.1304,
       "step": 99
     },
     {
-      "epoch": 1.82,
+      "epoch": 1.85,
       "learning_rate": 0.0002,
-      "loss": 0.0606,
+      "loss": 0.1319,
       "step": 100
     },
     {
-      "epoch": 1.84,
+      "epoch": 1.87,
       "learning_rate": 0.0002,
-      "loss": 0.1539,
+      "loss": 0.1322,
       "step": 101
     },
     {
-      "epoch": 1.85,
+      "epoch": 1.89,
       "learning_rate": 0.0002,
-      "loss": 0.0377,
+      "loss": 0.0869,
       "step": 102
     },
     {
-      "epoch": 1.87,
+      "epoch": 1.91,
       "learning_rate": 0.0002,
-      "loss": 0.0496,
+      "loss": 0.1111,
       "step": 103
     },
     {
-      "epoch": 1.89,
+      "epoch": 1.93,
       "learning_rate": 0.0002,
-      "loss": 0.068,
+      "loss": 0.0734,
       "step": 104
     },
     {
-      "epoch": 1.91,
+      "epoch": 1.94,
       "learning_rate": 0.0002,
-      "loss": 0.0968,
+      "loss": 0.1519,
       "step": 105
     },
     {
-      "epoch": 1.93,
+      "epoch": 1.96,
       "learning_rate": 0.0002,
-      "loss": 0.1314,
+      "loss": 0.0852,
       "step": 106
     },
     {
-      "epoch": 1.95,
+      "epoch": 1.98,
       "learning_rate": 0.0002,
-      "loss": 0.1178,
+      "loss": 0.062,
       "step": 107
     },
     {
-      "epoch": 1.96,
+      "epoch": 2.0,
       "learning_rate": 0.0002,
-      "loss": 0.1022,
+      "loss": 0.0315,
       "step": 108
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 0.0002,
-      "loss": 0.0802,
-      "step": 109
+      "epoch": 2.0,
+      "eval_loss": 0.04475187137722969,
+      "eval_runtime": 137.7441,
+      "eval_samples_per_second": 12.422,
+      "eval_steps_per_second": 0.392,
+      "step": 108
     },
     {
-      "epoch": 2.0,
+      "epoch": 2.02,
       "learning_rate": 0.0002,
-      "loss": 0.0327,
-      "step": 110
+      "loss": 0.0425,
+      "step": 109
     },
     {
-      "epoch": 2.0,
-      "eval_loss": 0.03490348532795906,
-      "eval_runtime": 97.2717,
-      "eval_samples_per_second": 8.913,
-      "eval_steps_per_second": 0.565,
+      "epoch": 2.04,
+      "learning_rate": 0.0002,
+      "loss": 0.0314,
       "step": 110
     },
     {
-      "epoch": 2.02,
+      "epoch": 2.06,
       "learning_rate": 0.0002,
-      "loss": 0.019,
+      "loss": 0.0326,
       "step": 111
     },
     {
-      "epoch": 2.04,
+      "epoch": 2.07,
       "learning_rate": 0.0002,
-      "loss": 0.0342,
+      "loss": 0.0515,
       "step": 112
     },
     {
-      "epoch": 2.05,
+      "epoch": 2.09,
       "learning_rate": 0.0002,
-      "loss": 0.0269,
+      "loss": 0.0331,
       "step": 113
     },
     {
-      "epoch": 2.07,
+      "epoch": 2.11,
       "learning_rate": 0.0002,
-      "loss": 0.019,
+      "loss": 0.0504,
       "step": 114
     },
     {
-      "epoch": 2.09,
+      "epoch": 2.13,
       "learning_rate": 0.0002,
-      "loss": 0.0341,
+      "loss": 0.0393,
       "step": 115
     },
     {
-      "epoch": 2.11,
+      "epoch": 2.15,
       "learning_rate": 0.0002,
-      "loss": 0.019,
+      "loss": 0.0284,
       "step": 116
     },
     {
-      "epoch": 2.13,
+      "epoch": 2.17,
       "learning_rate": 0.0002,
-      "loss": 0.0379,
+      "loss": 0.0227,
       "step": 117
     },
     {
-      "epoch": 2.15,
+      "epoch": 2.19,
       "learning_rate": 0.0002,
-      "loss": 0.0329,
+      "loss": 0.0301,
       "step": 118
     },
     {
-      "epoch": 2.16,
+      "epoch": 2.2,
       "learning_rate": 0.0002,
-      "loss": 0.0175,
+      "loss": 0.0229,
       "step": 119
     },
     {
-      "epoch": 2.18,
+      "epoch": 2.22,
       "learning_rate": 0.0002,
-      "loss": 0.0213,
+      "loss": 0.0307,
       "step": 120
     },
     {
-      "epoch": 2.2,
+      "epoch": 2.24,
       "learning_rate": 0.0002,
-      "loss": 0.0328,
+      "loss": 0.0386,
       "step": 121
     },
     {
-      "epoch": 2.22,
+      "epoch": 2.26,
       "learning_rate": 0.0002,
-      "loss": 0.0405,
+      "loss": 0.0336,
       "step": 122
     },
     {
-      "epoch": 2.24,
+      "epoch": 2.28,
       "learning_rate": 0.0002,
-      "loss": 0.018,
+      "loss": 0.0287,
       "step": 123
     },
     {
-      "epoch": 2.25,
+      "epoch": 2.3,
       "learning_rate": 0.0002,
-      "loss": 0.0306,
+      "loss": 0.0262,
       "step": 124
     },
     {
-      "epoch": 2.27,
+      "epoch": 2.31,
       "learning_rate": 0.0002,
-      "loss": 0.0093,
+      "loss": 0.017,
       "step": 125
     },
     {
-      "epoch": 2.29,
+      "epoch": 2.33,
       "learning_rate": 0.0002,
-      "loss": 0.028,
+      "loss": 0.0401,
       "step": 126
     },
     {
-      "epoch": 2.31,
+      "epoch": 2.35,
       "learning_rate": 0.0002,
-      "loss": 0.0238,
+      "loss": 0.0285,
       "step": 127
     },
     {
-      "epoch": 2.33,
+      "epoch": 2.37,
       "learning_rate": 0.0002,
-      "loss": 0.0143,
+      "loss": 0.0207,
       "step": 128
     },
     {
-      "epoch": 2.35,
+      "epoch": 2.39,
       "learning_rate": 0.0002,
-      "loss": 0.0303,
+      "loss": 0.0218,
       "step": 129
     },
     {
-      "epoch": 2.36,
+      "epoch": 2.41,
       "learning_rate": 0.0002,
-      "loss": 0.0228,
+      "loss": 0.0332,
       "step": 130
     },
     {
-      "epoch": 2.38,
+      "epoch": 2.43,
       "learning_rate": 0.0002,
-      "loss": 0.0146,
+      "loss": 0.0281,
       "step": 131
     },
     {
-      "epoch": 2.4,
+      "epoch": 2.44,
       "learning_rate": 0.0002,
-      "loss": 0.0163,
+      "loss": 0.0307,
       "step": 132
     },
     {
-      "epoch": 2.42,
+      "epoch": 2.46,
       "learning_rate": 0.0002,
-      "loss": 0.0374,
+      "loss": 0.0392,
       "step": 133
     },
     {
-      "epoch": 2.44,
+      "epoch": 2.48,
       "learning_rate": 0.0002,
-      "loss": 0.0352,
+      "loss": 0.0279,
       "step": 134
     },
     {
-      "epoch": 2.45,
+      "epoch": 2.5,
       "learning_rate": 0.0002,
-      "loss": 0.0296,
+      "loss": 0.0246,
       "step": 135
     },
     {
-      "epoch": 2.47,
+      "epoch": 2.52,
       "learning_rate": 0.0002,
-      "loss": 0.0326,
+      "loss": 0.0239,
       "step": 136
     },
     {
-      "epoch": 2.49,
+      "epoch": 2.54,
       "learning_rate": 0.0002,
-      "loss": 0.0174,
+      "loss": 0.0328,
       "step": 137
     },
     {
-      "epoch": 2.51,
+      "epoch": 2.56,
       "learning_rate": 0.0002,
-      "loss": 0.0273,
+      "loss": 0.0239,
       "step": 138
     },
     {
-      "epoch": 2.53,
+      "epoch": 2.57,
       "learning_rate": 0.0002,
-      "loss": 0.0158,
+      "loss": 0.0199,
       "step": 139
     },
     {
-      "epoch": 2.55,
+      "epoch": 2.59,
       "learning_rate": 0.0002,
-      "loss": 0.0249,
+      "loss": 0.0236,
       "step": 140
     },
     {
-      "epoch": 2.56,
+      "epoch": 2.61,
       "learning_rate": 0.0002,
-      "loss": 0.0245,
+      "loss": 0.0144,
       "step": 141
     },
     {
-      "epoch": 2.58,
+      "epoch": 2.63,
       "learning_rate": 0.0002,
-      "loss": 0.0147,
+      "loss": 0.0328,
       "step": 142
     },
     {
-      "epoch": 2.6,
+      "epoch": 2.65,
       "learning_rate": 0.0002,
-      "loss": 0.0108,
+      "loss": 0.0393,
       "step": 143
     },
     {
-      "epoch": 2.62,
+      "epoch": 2.67,
       "learning_rate": 0.0002,
-      "loss": 0.0184,
+      "loss": 0.0213,
       "step": 144
     },
     {
-      "epoch": 2.64,
+      "epoch": 2.69,
       "learning_rate": 0.0002,
-      "loss": 0.0249,
+      "loss": 0.0168,
       "step": 145
     },
     {
-      "epoch": 2.65,
+      "epoch": 2.7,
       "learning_rate": 0.0002,
-      "loss": 0.0366,
+      "loss": 0.0191,
       "step": 146
     },
     {
-      "epoch": 2.67,
+      "epoch": 2.72,
       "learning_rate": 0.0002,
-      "loss": 0.0168,
+      "loss": 0.0302,
       "step": 147
     },
     {
-      "epoch": 2.69,
+      "epoch": 2.74,
       "learning_rate": 0.0002,
-      "loss": 0.0309,
+      "loss": 0.0247,
       "step": 148
     },
     {
-      "epoch": 2.71,
+      "epoch": 2.76,
       "learning_rate": 0.0002,
-      "loss": 0.0207,
+      "loss": 0.041,
       "step": 149
     },
     {
-      "epoch": 2.73,
+      "epoch": 2.78,
       "learning_rate": 0.0002,
-      "loss": 0.0173,
+      "loss": 0.0287,
       "step": 150
     },
     {
-      "epoch": 2.75,
+      "epoch": 2.8,
       "learning_rate": 0.0002,
-      "loss": 0.017,
+      "loss": 0.0334,
       "step": 151
     },
     {
-      "epoch": 2.76,
+      "epoch": 2.81,
       "learning_rate": 0.0002,
-      "loss": 0.0063,
+      "loss": 0.0207,
       "step": 152
     },
     {
-      "epoch": 2.78,
+      "epoch": 2.83,
       "learning_rate": 0.0002,
-      "loss": 0.0239,
+      "loss": 0.0265,
       "step": 153
     },
     {
-      "epoch": 2.8,
+      "epoch": 2.85,
       "learning_rate": 0.0002,
-      "loss": 0.0264,
+      "loss": 0.037,
       "step": 154
     },
     {
-      "epoch": 2.82,
+      "epoch": 2.87,
       "learning_rate": 0.0002,
-      "loss": 0.0336,
+      "loss": 0.0276,
       "step": 155
     },
     {
-      "epoch": 2.84,
+      "epoch": 2.89,
       "learning_rate": 0.0002,
-      "loss": 0.0272,
+      "loss": 0.0229,
       "step": 156
     },
     {
-      "epoch": 2.85,
+      "epoch": 2.91,
       "learning_rate": 0.0002,
-      "loss": 0.0271,
+      "loss": 0.0458,
       "step": 157
     },
     {
-      "epoch": 2.87,
+      "epoch": 2.93,
       "learning_rate": 0.0002,
-      "loss": 0.0093,
+      "loss": 0.0428,
       "step": 158
     },
     {
-      "epoch": 2.89,
+      "epoch": 2.94,
       "learning_rate": 0.0002,
-      "loss": 0.0162,
+      "loss": 0.024,
       "step": 159
     },
     {
-      "epoch": 2.91,
+      "epoch": 2.96,
       "learning_rate": 0.0002,
-      "loss": 0.0152,
+      "loss": 0.0194,
       "step": 160
     },
     {
-      "epoch": 2.93,
+      "epoch": 2.98,
       "learning_rate": 0.0002,
-      "loss": 0.0128,
+      "loss": 0.0272,
       "step": 161
     },
     {
-      "epoch": 2.95,
+      "epoch": 3.0,
       "learning_rate": 0.0002,
-      "loss": 0.0234,
+      "loss": 0.0267,
       "step": 162
     },
     {
-      "epoch": 2.96,
+      "epoch": 3.0,
+      "eval_loss": 0.014697842299938202,
+      "eval_runtime": 137.708,
+      "eval_samples_per_second": 12.425,
+      "eval_steps_per_second": 0.392,
+      "step": 162
+    },
+    {
+      "epoch": 3.02,
       "learning_rate": 0.0002,
-      "loss": 0.0306,
+      "loss": 0.0172,
       "step": 163
     },
     {
-      "epoch": 2.98,
+      "epoch": 3.04,
       "learning_rate": 0.0002,
-      "loss": 0.043,
+      "loss": 0.0134,
       "step": 164
     },
     {
-      "epoch": 3.0,
+      "epoch": 3.06,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
-      "step": 165
-    },
-    {
-      "epoch": 3.0,
-      "eval_loss": 0.01406150870025158,
-      "eval_runtime": 97.3873,
-      "eval_samples_per_second": 8.903,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0147,
       "step": 165
     },
     {
-      "epoch": 3.02,
+      "epoch": 3.07,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
+      "loss": 0.0089,
       "step": 166
     },
     {
-      "epoch": 3.04,
+      "epoch": 3.09,
       "learning_rate": 0.0002,
-      "loss": 0.0075,
+      "loss": 0.0171,
       "step": 167
     },
     {
-      "epoch": 3.05,
+      "epoch": 3.11,
       "learning_rate": 0.0002,
-      "loss": 0.0114,
+      "loss": 0.0162,
       "step": 168
     },
     {
-      "epoch": 3.07,
+      "epoch": 3.13,
       "learning_rate": 0.0002,
-      "loss": 0.0154,
+      "loss": 0.0114,
       "step": 169
     },
     {
-      "epoch": 3.09,
+      "epoch": 3.15,
       "learning_rate": 0.0002,
-      "loss": 0.0188,
+      "loss": 0.006,
       "step": 170
     },
     {
-      "epoch": 3.11,
+      "epoch": 3.17,
       "learning_rate": 0.0002,
-      "loss": 0.0163,
+      "loss": 0.017,
       "step": 171
     },
     {
-      "epoch": 3.13,
+      "epoch": 3.19,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0096,
       "step": 172
     },
     {
-      "epoch": 3.15,
+      "epoch": 3.2,
       "learning_rate": 0.0002,
-      "loss": 0.024,
+      "loss": 0.0145,
       "step": 173
     },
     {
-      "epoch": 3.16,
+      "epoch": 3.22,
       "learning_rate": 0.0002,
-      "loss": 0.0098,
+      "loss": 0.0118,
       "step": 174
     },
     {
-      "epoch": 3.18,
+      "epoch": 3.24,
       "learning_rate": 0.0002,
-      "loss": 0.0238,
+      "loss": 0.0139,
       "step": 175
     },
     {
-      "epoch": 3.2,
+      "epoch": 3.26,
       "learning_rate": 0.0002,
-      "loss": 0.0138,
+      "loss": 0.0136,
       "step": 176
     },
     {
-      "epoch": 3.22,
+      "epoch": 3.28,
       "learning_rate": 0.0002,
-      "loss": 0.0219,
+      "loss": 0.0181,
       "step": 177
     },
     {
-      "epoch": 3.24,
+      "epoch": 3.3,
       "learning_rate": 0.0002,
-      "loss": 0.0064,
+      "loss": 0.0104,
       "step": 178
     },
     {
-      "epoch": 3.25,
+      "epoch": 3.31,
       "learning_rate": 0.0002,
-      "loss": 0.0158,
+      "loss": 0.0117,
       "step": 179
     },
     {
-      "epoch": 3.27,
+      "epoch": 3.33,
       "learning_rate": 0.0002,
-      "loss": 0.0104,
+      "loss": 0.0146,
       "step": 180
     },
     {
-      "epoch": 3.29,
+      "epoch": 3.35,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.0107,
       "step": 181
     },
     {
-      "epoch": 3.31,
+      "epoch": 3.37,
       "learning_rate": 0.0002,
-      "loss": 0.0136,
+      "loss": 0.0139,
       "step": 182
     },
     {
-      "epoch": 3.33,
+      "epoch": 3.39,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0083,
       "step": 183
     },
     {
-      "epoch": 3.35,
+      "epoch": 3.41,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
+      "loss": 0.0093,
       "step": 184
     },
     {
-      "epoch": 3.36,
+      "epoch": 3.43,
       "learning_rate": 0.0002,
-      "loss": 0.0126,
+      "loss": 0.01,
       "step": 185
     },
     {
-      "epoch": 3.38,
+      "epoch": 3.44,
       "learning_rate": 0.0002,
-      "loss": 0.0209,
+      "loss": 0.0163,
       "step": 186
     },
     {
-      "epoch": 3.4,
+      "epoch": 3.46,
       "learning_rate": 0.0002,
-      "loss": 0.0269,
+      "loss": 0.0142,
       "step": 187
     },
     {
-      "epoch": 3.42,
+      "epoch": 3.48,
       "learning_rate": 0.0002,
-      "loss": 0.0239,
+      "loss": 0.0073,
       "step": 188
     },
     {
-      "epoch": 3.44,
+      "epoch": 3.5,
       "learning_rate": 0.0002,
-      "loss": 0.0104,
+      "loss": 0.0106,
       "step": 189
     },
     {
-      "epoch": 3.45,
+      "epoch": 3.52,
       "learning_rate": 0.0002,
-      "loss": 0.0148,
+      "loss": 0.0133,
       "step": 190
     },
     {
-      "epoch": 3.47,
+      "epoch": 3.54,
       "learning_rate": 0.0002,
-      "loss": 0.0219,
+      "loss": 0.012,
       "step": 191
     },
     {
-      "epoch": 3.49,
+      "epoch": 3.56,
       "learning_rate": 0.0002,
-      "loss": 0.024,
+      "loss": 0.0131,
       "step": 192
     },
     {
-      "epoch": 3.51,
+      "epoch": 3.57,
       "learning_rate": 0.0002,
-      "loss": 0.0191,
+      "loss": 0.0184,
       "step": 193
     },
     {
-      "epoch": 3.53,
+      "epoch": 3.59,
       "learning_rate": 0.0002,
-      "loss": 0.0143,
+      "loss": 0.0095,
       "step": 194
     },
     {
-      "epoch": 3.55,
+      "epoch": 3.61,
       "learning_rate": 0.0002,
-      "loss": 0.0209,
+      "loss": 0.0108,
       "step": 195
     },
     {
-      "epoch": 3.56,
+      "epoch": 3.63,
       "learning_rate": 0.0002,
-      "loss": 0.0085,
+      "loss": 0.0109,
       "step": 196
     },
     {
-      "epoch": 3.58,
+      "epoch": 3.65,
       "learning_rate": 0.0002,
-      "loss": 0.0133,
+      "loss": 0.0098,
       "step": 197
     },
     {
-      "epoch": 3.6,
+      "epoch": 3.67,
       "learning_rate": 0.0002,
-      "loss": 0.0135,
+      "loss": 0.0116,
       "step": 198
     },
     {
-      "epoch": 3.62,
+      "epoch": 3.69,
       "learning_rate": 0.0002,
-      "loss": 0.0212,
+      "loss": 0.0132,
       "step": 199
     },
     {
-      "epoch": 3.64,
+      "epoch": 3.7,
       "learning_rate": 0.0002,
-      "loss": 0.0262,
+      "loss": 0.0137,
       "step": 200
     },
     {
-      "epoch": 3.65,
+      "epoch": 3.72,
       "learning_rate": 0.0002,
-      "loss": 0.0083,
+      "loss": 0.0138,
       "step": 201
     },
     {
-      "epoch": 3.67,
+      "epoch": 3.74,
       "learning_rate": 0.0002,
-      "loss": 0.0162,
+      "loss": 0.0101,
       "step": 202
     },
     {
-      "epoch": 3.69,
+      "epoch": 3.76,
       "learning_rate": 0.0002,
-      "loss": 0.0246,
+      "loss": 0.0072,
       "step": 203
     },
     {
-      "epoch": 3.71,
+      "epoch": 3.78,
       "learning_rate": 0.0002,
-      "loss": 0.0159,
+      "loss": 0.0096,
       "step": 204
     },
     {
-      "epoch": 3.73,
+      "epoch": 3.8,
       "learning_rate": 0.0002,
-      "loss": 0.0228,
+      "loss": 0.0102,
       "step": 205
     },
     {
-      "epoch": 3.75,
+      "epoch": 3.81,
       "learning_rate": 0.0002,
-      "loss": 0.0185,
+      "loss": 0.0131,
       "step": 206
     },
     {
-      "epoch": 3.76,
+      "epoch": 3.83,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.0101,
       "step": 207
     },
     {
-      "epoch": 3.78,
+      "epoch": 3.85,
       "learning_rate": 0.0002,
-      "loss": 0.0143,
+      "loss": 0.0099,
       "step": 208
     },
     {
-      "epoch": 3.8,
+      "epoch": 3.87,
       "learning_rate": 0.0002,
-      "loss": 0.0135,
+      "loss": 0.0104,
       "step": 209
     },
     {
-      "epoch": 3.82,
+      "epoch": 3.89,
       "learning_rate": 0.0002,
-      "loss": 0.0144,
+      "loss": 0.0126,
       "step": 210
     },
     {
-      "epoch": 3.84,
+      "epoch": 3.91,
       "learning_rate": 0.0002,
-      "loss": 0.0105,
+      "loss": 0.0138,
       "step": 211
     },
     {
-      "epoch": 3.85,
+      "epoch": 3.93,
       "learning_rate": 0.0002,
-      "loss": 0.01,
+      "loss": 0.0132,
       "step": 212
     },
     {
-      "epoch": 3.87,
+      "epoch": 3.94,
       "learning_rate": 0.0002,
-      "loss": 0.0079,
+      "loss": 0.0122,
       "step": 213
     },
     {
-      "epoch": 3.89,
+      "epoch": 3.96,
       "learning_rate": 0.0002,
-      "loss": 0.0124,
+      "loss": 0.0091,
       "step": 214
     },
     {
-      "epoch": 3.91,
+      "epoch": 3.98,
       "learning_rate": 0.0002,
-      "loss": 0.0083,
+      "loss": 0.0102,
       "step": 215
     },
     {
-      "epoch": 3.93,
+      "epoch": 4.0,
       "learning_rate": 0.0002,
-      "loss": 0.016,
+      "loss": 0.0096,
+      "step": 216
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.00889603141695261,
+      "eval_runtime": 138.0674,
+      "eval_samples_per_second": 12.392,
+      "eval_steps_per_second": 0.391,
       "step": 216
     },
     {
-      "epoch": 3.95,
+      "epoch": 4.02,
       "learning_rate": 0.0002,
-      "loss": 0.018,
+      "loss": 0.0087,
       "step": 217
     },
     {
-      "epoch": 3.96,
+      "epoch": 4.04,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.0124,
       "step": 218
     },
     {
-      "epoch": 3.98,
+      "epoch": 4.06,
       "learning_rate": 0.0002,
-      "loss": 0.0042,
+      "loss": 0.013,
       "step": 219
     },
     {
-      "epoch": 4.0,
+      "epoch": 4.07,
       "learning_rate": 0.0002,
-      "loss": 0.0204,
-      "step": 220
-    },
-    {
-      "epoch": 4.0,
-      "eval_loss": 0.009295277297496796,
-      "eval_runtime": 97.3995,
-      "eval_samples_per_second": 8.901,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0047,
       "step": 220
     },
     {
-      "epoch": 4.02,
+      "epoch": 4.09,
       "learning_rate": 0.0002,
-      "loss": 0.0034,
+      "loss": 0.0079,
       "step": 221
     },
     {
-      "epoch": 4.04,
+      "epoch": 4.11,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0088,
       "step": 222
     },
     {
-      "epoch": 4.05,
+      "epoch": 4.13,
       "learning_rate": 0.0002,
-      "loss": 0.0075,
+      "loss": 0.0064,
       "step": 223
     },
     {
-      "epoch": 4.07,
+      "epoch": 4.15,
       "learning_rate": 0.0002,
-      "loss": 0.0032,
+      "loss": 0.0132,
       "step": 224
     },
     {
-      "epoch": 4.09,
+      "epoch": 4.17,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0067,
       "step": 225
     },
     {
-      "epoch": 4.11,
+      "epoch": 4.19,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.0045,
       "step": 226
     },
     {
-      "epoch": 4.13,
+      "epoch": 4.2,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0086,
       "step": 227
     },
     {
-      "epoch": 4.15,
+      "epoch": 4.22,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0096,
       "step": 228
     },
     {
-      "epoch": 4.16,
+      "epoch": 4.24,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.0067,
       "step": 229
     },
     {
-      "epoch": 4.18,
+      "epoch": 4.26,
       "learning_rate": 0.0002,
-      "loss": 0.0089,
+      "loss": 0.0081,
       "step": 230
     },
     {
-      "epoch": 4.2,
+      "epoch": 4.28,
       "learning_rate": 0.0002,
-      "loss": 0.0119,
+      "loss": 0.0131,
       "step": 231
     },
     {
-      "epoch": 4.22,
+      "epoch": 4.3,
       "learning_rate": 0.0002,
-      "loss": 0.0221,
+      "loss": 0.0086,
       "step": 232
     },
     {
-      "epoch": 4.24,
+      "epoch": 4.31,
       "learning_rate": 0.0002,
-      "loss": 0.0039,
+      "loss": 0.0044,
       "step": 233
     },
     {
-      "epoch": 4.25,
+      "epoch": 4.33,
       "learning_rate": 0.0002,
-      "loss": 0.0122,
+      "loss": 0.0136,
       "step": 234
     },
     {
-      "epoch": 4.27,
+      "epoch": 4.35,
       "learning_rate": 0.0002,
-      "loss": 0.0126,
+      "loss": 0.0056,
       "step": 235
     },
     {
-      "epoch": 4.29,
+      "epoch": 4.37,
       "learning_rate": 0.0002,
-      "loss": 0.01,
+      "loss": 0.0058,
       "step": 236
     },
     {
-      "epoch": 4.31,
+      "epoch": 4.39,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.012,
       "step": 237
     },
     {
-      "epoch": 4.33,
+      "epoch": 4.41,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0093,
       "step": 238
     },
     {
-      "epoch": 4.35,
+      "epoch": 4.43,
       "learning_rate": 0.0002,
-      "loss": 0.012,
+      "loss": 0.0075,
       "step": 239
     },
     {
-      "epoch": 4.36,
+      "epoch": 4.44,
       "learning_rate": 0.0002,
-      "loss": 0.0108,
+      "loss": 0.0071,
       "step": 240
     },
     {
-      "epoch": 4.38,
+      "epoch": 4.46,
       "learning_rate": 0.0002,
-      "loss": 0.013,
+      "loss": 0.0053,
       "step": 241
     },
     {
-      "epoch": 4.4,
+      "epoch": 4.48,
       "learning_rate": 0.0002,
-      "loss": 0.0097,
+      "loss": 0.0126,
       "step": 242
     },
     {
-      "epoch": 4.42,
+      "epoch": 4.5,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0108,
       "step": 243
     },
     {
-      "epoch": 4.44,
+      "epoch": 4.52,
       "learning_rate": 0.0002,
-      "loss": 0.0043,
+      "loss": 0.0104,
       "step": 244
     },
     {
-      "epoch": 4.45,
+      "epoch": 4.54,
       "learning_rate": 0.0002,
-      "loss": 0.0084,
+      "loss": 0.0065,
       "step": 245
     },
     {
-      "epoch": 4.47,
+      "epoch": 4.56,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
+      "loss": 0.0115,
       "step": 246
     },
     {
-      "epoch": 4.49,
+      "epoch": 4.57,
       "learning_rate": 0.0002,
-      "loss": 0.0113,
+      "loss": 0.0066,
       "step": 247
     },
     {
-      "epoch": 4.51,
+      "epoch": 4.59,
       "learning_rate": 0.0002,
-      "loss": 0.0095,
+      "loss": 0.0067,
       "step": 248
     },
     {
-      "epoch": 4.53,
+      "epoch": 4.61,
       "learning_rate": 0.0002,
-      "loss": 0.0047,
+      "loss": 0.0075,
       "step": 249
     },
     {
-      "epoch": 4.55,
+      "epoch": 4.63,
       "learning_rate": 0.0002,
-      "loss": 0.0124,
+      "loss": 0.0046,
       "step": 250
     },
     {
-      "epoch": 4.56,
+      "epoch": 4.65,
       "learning_rate": 0.0002,
-      "loss": 0.0095,
+      "loss": 0.012,
       "step": 251
     },
     {
-      "epoch": 4.58,
+      "epoch": 4.67,
       "learning_rate": 0.0002,
-      "loss": 0.0108,
+      "loss": 0.0107,
       "step": 252
     },
     {
-      "epoch": 4.6,
+      "epoch": 4.69,
       "learning_rate": 0.0002,
-      "loss": 0.0115,
+      "loss": 0.0114,
       "step": 253
     },
     {
-      "epoch": 4.62,
+      "epoch": 4.7,
       "learning_rate": 0.0002,
-      "loss": 0.0063,
+      "loss": 0.0073,
       "step": 254
     },
     {
-      "epoch": 4.64,
+      "epoch": 4.72,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.007,
       "step": 255
     },
     {
-      "epoch": 4.65,
+      "epoch": 4.74,
       "learning_rate": 0.0002,
-      "loss": 0.0037,
+      "loss": 0.0085,
       "step": 256
     },
     {
-      "epoch": 4.67,
+      "epoch": 4.76,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0096,
       "step": 257
     },
     {
-      "epoch": 4.69,
+      "epoch": 4.78,
       "learning_rate": 0.0002,
-      "loss": 0.0057,
+      "loss": 0.0109,
       "step": 258
     },
     {
-      "epoch": 4.71,
+      "epoch": 4.8,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0103,
       "step": 259
     },
     {
-      "epoch": 4.73,
+      "epoch": 4.81,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0067,
       "step": 260
     },
     {
-      "epoch": 4.75,
+      "epoch": 4.83,
       "learning_rate": 0.0002,
-      "loss": 0.0156,
+      "loss": 0.0051,
       "step": 261
     },
     {
-      "epoch": 4.76,
+      "epoch": 4.85,
       "learning_rate": 0.0002,
-      "loss": 0.0069,
+      "loss": 0.0058,
       "step": 262
     },
     {
-      "epoch": 4.78,
+      "epoch": 4.87,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.0075,
       "step": 263
     },
     {
-      "epoch": 4.8,
+      "epoch": 4.89,
       "learning_rate": 0.0002,
-      "loss": 0.0073,
+      "loss": 0.0075,
       "step": 264
     },
     {
-      "epoch": 4.82,
+      "epoch": 4.91,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0141,
       "step": 265
     },
     {
-      "epoch": 4.84,
+      "epoch": 4.93,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0085,
       "step": 266
     },
     {
-      "epoch": 4.85,
+      "epoch": 4.94,
       "learning_rate": 0.0002,
-      "loss": 0.0113,
+      "loss": 0.0095,
       "step": 267
     },
     {
-      "epoch": 4.87,
+      "epoch": 4.96,
       "learning_rate": 0.0002,
-      "loss": 0.0186,
+      "loss": 0.0092,
       "step": 268
     },
     {
-      "epoch": 4.89,
+      "epoch": 4.98,
       "learning_rate": 0.0002,
-      "loss": 0.0101,
+      "loss": 0.0093,
       "step": 269
     },
     {
-      "epoch": 4.91,
+      "epoch": 5.0,
       "learning_rate": 0.0002,
-      "loss": 0.0142,
+      "loss": 0.007,
       "step": 270
     },
     {
-      "epoch": 4.93,
+      "epoch": 5.0,
+      "eval_loss": 0.006592159159481525,
+      "eval_runtime": 138.0662,
+      "eval_samples_per_second": 12.393,
+      "eval_steps_per_second": 0.391,
+      "step": 270
+    },
+    {
+      "epoch": 5.02,
       "learning_rate": 0.0002,
-      "loss": 0.0108,
+      "loss": 0.0068,
       "step": 271
     },
     {
-      "epoch": 4.95,
+      "epoch": 5.04,
       "learning_rate": 0.0002,
-      "loss": 0.0106,
+      "loss": 0.0067,
       "step": 272
     },
     {
-      "epoch": 4.96,
+      "epoch": 5.06,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0062,
       "step": 273
     },
     {
-      "epoch": 4.98,
+      "epoch": 5.07,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
+      "loss": 0.004,
       "step": 274
     },
     {
-      "epoch": 5.0,
+      "epoch": 5.09,
       "learning_rate": 0.0002,
-      "loss": 0.0063,
-      "step": 275
-    },
-    {
-      "epoch": 5.0,
-      "eval_loss": 0.008474300615489483,
-      "eval_runtime": 97.3016,
-      "eval_samples_per_second": 8.91,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0092,
       "step": 275
     },
     {
-      "epoch": 5.02,
+      "epoch": 5.11,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0062,
       "step": 276
     },
     {
-      "epoch": 5.04,
+      "epoch": 5.13,
       "learning_rate": 0.0002,
-      "loss": 0.0055,
+      "loss": 0.0078,
       "step": 277
     },
     {
-      "epoch": 5.05,
+      "epoch": 5.15,
       "learning_rate": 0.0002,
-      "loss": 0.0048,
+      "loss": 0.0066,
       "step": 278
     },
     {
-      "epoch": 5.07,
+      "epoch": 5.17,
       "learning_rate": 0.0002,
-      "loss": 0.0056,
+      "loss": 0.0096,
       "step": 279
     },
     {
-      "epoch": 5.09,
+      "epoch": 5.19,
       "learning_rate": 0.0002,
-      "loss": 0.008,
+      "loss": 0.0047,
       "step": 280
     },
     {
-      "epoch": 5.11,
+      "epoch": 5.2,
       "learning_rate": 0.0002,
-      "loss": 0.0022,
+      "loss": 0.007,
       "step": 281
     },
     {
-      "epoch": 5.13,
+      "epoch": 5.22,
       "learning_rate": 0.0002,
-      "loss": 0.0105,
+      "loss": 0.0034,
       "step": 282
     },
     {
-      "epoch": 5.15,
+      "epoch": 5.24,
       "learning_rate": 0.0002,
-      "loss": 0.021,
+      "loss": 0.0036,
       "step": 283
     },
     {
-      "epoch": 5.16,
+      "epoch": 5.26,
       "learning_rate": 0.0002,
-      "loss": 0.009,
+      "loss": 0.005,
       "step": 284
     },
     {
-      "epoch": 5.18,
+      "epoch": 5.28,
       "learning_rate": 0.0002,
-      "loss": 0.0085,
+      "loss": 0.0089,
       "step": 285
     },
     {
-      "epoch": 5.2,
+      "epoch": 5.3,
       "learning_rate": 0.0002,
-      "loss": 0.0112,
+      "loss": 0.0058,
       "step": 286
     },
     {
-      "epoch": 5.22,
+      "epoch": 5.31,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.009,
       "step": 287
     },
     {
-      "epoch": 5.24,
+      "epoch": 5.33,
       "learning_rate": 0.0002,
-      "loss": 0.0108,
+      "loss": 0.0066,
       "step": 288
     },
     {
-      "epoch": 5.25,
+      "epoch": 5.35,
       "learning_rate": 0.0002,
-      "loss": 0.0101,
+      "loss": 0.0067,
       "step": 289
     },
     {
-      "epoch": 5.27,
+      "epoch": 5.37,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.005,
       "step": 290
     },
     {
-      "epoch": 5.29,
+      "epoch": 5.39,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0101,
       "step": 291
     },
     {
-      "epoch": 5.31,
+      "epoch": 5.41,
       "learning_rate": 0.0002,
-      "loss": 0.0079,
+      "loss": 0.0073,
       "step": 292
     },
     {
-      "epoch": 5.33,
+      "epoch": 5.43,
       "learning_rate": 0.0002,
-      "loss": 0.0127,
+      "loss": 0.0114,
       "step": 293
     },
     {
-      "epoch": 5.35,
+      "epoch": 5.44,
       "learning_rate": 0.0002,
-      "loss": 0.0088,
+      "loss": 0.005,
       "step": 294
     },
     {
-      "epoch": 5.36,
+      "epoch": 5.46,
       "learning_rate": 0.0002,
-      "loss": 0.0156,
+      "loss": 0.0095,
       "step": 295
     },
     {
-      "epoch": 5.38,
+      "epoch": 5.48,
       "learning_rate": 0.0002,
-      "loss": 0.0066,
+      "loss": 0.0101,
       "step": 296
     },
     {
-      "epoch": 5.4,
+      "epoch": 5.5,
       "learning_rate": 0.0002,
-      "loss": 0.0013,
+      "loss": 0.0047,
       "step": 297
     },
     {
-      "epoch": 5.42,
+      "epoch": 5.52,
       "learning_rate": 0.0002,
-      "loss": 0.0078,
+      "loss": 0.0056,
       "step": 298
     },
     {
-      "epoch": 5.44,
+      "epoch": 5.54,
       "learning_rate": 0.0002,
-      "loss": 0.0079,
+      "loss": 0.0091,
       "step": 299
     },
     {
-      "epoch": 5.45,
+      "epoch": 5.56,
       "learning_rate": 0.0002,
-      "loss": 0.0037,
+      "loss": 0.0071,
       "step": 300
     },
     {
-      "epoch": 5.47,
+      "epoch": 5.57,
       "learning_rate": 0.0002,
-      "loss": 0.0118,
+      "loss": 0.0072,
       "step": 301
     },
     {
-      "epoch": 5.49,
+      "epoch": 5.59,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
+      "loss": 0.008,
       "step": 302
     },
     {
-      "epoch": 5.51,
+      "epoch": 5.61,
       "learning_rate": 0.0002,
-      "loss": 0.0033,
+      "loss": 0.0051,
       "step": 303
     },
     {
-      "epoch": 5.53,
+      "epoch": 5.63,
       "learning_rate": 0.0002,
-      "loss": 0.0086,
+      "loss": 0.0073,
       "step": 304
     },
     {
-      "epoch": 5.55,
+      "epoch": 5.65,
       "learning_rate": 0.0002,
-      "loss": 0.0031,
+      "loss": 0.0067,
       "step": 305
     },
     {
-      "epoch": 5.56,
+      "epoch": 5.67,
       "learning_rate": 0.0002,
-      "loss": 0.0022,
+      "loss": 0.0054,
       "step": 306
     },
     {
-      "epoch": 5.58,
+      "epoch": 5.69,
       "learning_rate": 0.0002,
-      "loss": 0.0144,
+      "loss": 0.008,
       "step": 307
     },
     {
-      "epoch": 5.6,
+      "epoch": 5.7,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.007,
       "step": 308
     },
     {
-      "epoch": 5.62,
+      "epoch": 5.72,
       "learning_rate": 0.0002,
-      "loss": 0.0078,
+      "loss": 0.006,
       "step": 309
     },
     {
-      "epoch": 5.64,
+      "epoch": 5.74,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.0024,
       "step": 310
     },
     {
-      "epoch": 5.65,
+      "epoch": 5.76,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.007,
       "step": 311
     },
     {
-      "epoch": 5.67,
+      "epoch": 5.78,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0047,
       "step": 312
     },
     {
-      "epoch": 5.69,
+      "epoch": 5.8,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0081,
       "step": 313
     },
     {
-      "epoch": 5.71,
+      "epoch": 5.81,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0081,
       "step": 314
     },
     {
-      "epoch": 5.73,
+      "epoch": 5.83,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.0059,
       "step": 315
     },
     {
-      "epoch": 5.75,
+      "epoch": 5.85,
       "learning_rate": 0.0002,
-      "loss": 0.0034,
+      "loss": 0.0045,
       "step": 316
     },
     {
-      "epoch": 5.76,
+      "epoch": 5.87,
       "learning_rate": 0.0002,
-      "loss": 0.0025,
+      "loss": 0.0088,
       "step": 317
     },
     {
-      "epoch": 5.78,
+      "epoch": 5.89,
       "learning_rate": 0.0002,
-      "loss": 0.0099,
+      "loss": 0.0082,
       "step": 318
     },
     {
-      "epoch": 5.8,
+      "epoch": 5.91,
       "learning_rate": 0.0002,
-      "loss": 0.0022,
+      "loss": 0.0159,
       "step": 319
     },
     {
-      "epoch": 5.82,
+      "epoch": 5.93,
       "learning_rate": 0.0002,
-      "loss": 0.0025,
+      "loss": 0.0066,
       "step": 320
     },
     {
-      "epoch": 5.84,
+      "epoch": 5.94,
       "learning_rate": 0.0002,
-      "loss": 0.0074,
+      "loss": 0.0068,
       "step": 321
     },
     {
-      "epoch": 5.85,
+      "epoch": 5.96,
       "learning_rate": 0.0002,
-      "loss": 0.0064,
+      "loss": 0.0082,
       "step": 322
     },
     {
-      "epoch": 5.87,
+      "epoch": 5.98,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.0057,
       "step": 323
     },
     {
-      "epoch": 5.89,
+      "epoch": 6.0,
       "learning_rate": 0.0002,
-      "loss": 0.003,
+      "loss": 0.005,
       "step": 324
     },
     {
-      "epoch": 5.91,
+      "epoch": 6.0,
+      "eval_loss": 0.006087442860007286,
+      "eval_runtime": 138.075,
+      "eval_samples_per_second": 12.392,
+      "eval_steps_per_second": 0.391,
+      "step": 324
+    },
+    {
+      "epoch": 6.02,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0054,
       "step": 325
     },
     {
-      "epoch": 5.93,
+      "epoch": 6.04,
       "learning_rate": 0.0002,
-      "loss": 0.0058,
+      "loss": 0.006,
       "step": 326
     },
     {
-      "epoch": 5.95,
+      "epoch": 6.06,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
+      "loss": 0.0059,
       "step": 327
     },
     {
-      "epoch": 5.96,
+      "epoch": 6.07,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.0045,
       "step": 328
     },
     {
-      "epoch": 5.98,
+      "epoch": 6.09,
       "learning_rate": 0.0002,
-      "loss": 0.0204,
+      "loss": 0.0038,
       "step": 329
     },
     {
-      "epoch": 6.0,
+      "epoch": 6.11,
       "learning_rate": 0.0002,
-      "loss": 0.0142,
-      "step": 330
-    },
-    {
-      "epoch": 6.0,
-      "eval_loss": 0.004596048966050148,
-      "eval_runtime": 97.3891,
-      "eval_samples_per_second": 8.902,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0041,
       "step": 330
     },
     {
-      "epoch": 6.02,
+      "epoch": 6.13,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0059,
       "step": 331
     },
     {
-      "epoch": 6.04,
+      "epoch": 6.15,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0094,
       "step": 332
     },
     {
-      "epoch": 6.05,
+      "epoch": 6.17,
       "learning_rate": 0.0002,
-      "loss": 0.0056,
+      "loss": 0.0078,
       "step": 333
     },
     {
-      "epoch": 6.07,
+      "epoch": 6.19,
       "learning_rate": 0.0002,
-      "loss": 0.0052,
+      "loss": 0.0049,
       "step": 334
     },
     {
-      "epoch": 6.09,
+      "epoch": 6.2,
       "learning_rate": 0.0002,
-      "loss": 0.0188,
+      "loss": 0.0159,
       "step": 335
     },
     {
-      "epoch": 6.11,
+      "epoch": 6.22,
       "learning_rate": 0.0002,
-      "loss": 0.0106,
+      "loss": 0.0054,
       "step": 336
     },
     {
-      "epoch": 6.13,
+      "epoch": 6.24,
       "learning_rate": 0.0002,
-      "loss": 0.0198,
+      "loss": 0.0096,
       "step": 337
     },
     {
-      "epoch": 6.15,
+      "epoch": 6.26,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0114,
       "step": 338
     },
     {
-      "epoch": 6.16,
+      "epoch": 6.28,
       "learning_rate": 0.0002,
-      "loss": 0.0133,
+      "loss": 0.007,
       "step": 339
     },
     {
-      "epoch": 6.18,
+      "epoch": 6.3,
       "learning_rate": 0.0002,
-      "loss": 0.0107,
+      "loss": 0.0098,
       "step": 340
     },
     {
-      "epoch": 6.2,
+      "epoch": 6.31,
       "learning_rate": 0.0002,
-      "loss": 0.0075,
+      "loss": 0.0041,
       "step": 341
     },
     {
-      "epoch": 6.22,
+      "epoch": 6.33,
       "learning_rate": 0.0002,
-      "loss": 0.0126,
+      "loss": 0.004,
       "step": 342
     },
     {
-      "epoch": 6.24,
+      "epoch": 6.35,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.005,
       "step": 343
     },
     {
-      "epoch": 6.25,
+      "epoch": 6.37,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0168,
       "step": 344
     },
     {
-      "epoch": 6.27,
+      "epoch": 6.39,
       "learning_rate": 0.0002,
-      "loss": 0.027,
+      "loss": 0.0091,
       "step": 345
     },
     {
-      "epoch": 6.29,
+      "epoch": 6.41,
       "learning_rate": 0.0002,
-      "loss": 0.008,
+      "loss": 0.0069,
       "step": 346
     },
     {
-      "epoch": 6.31,
+      "epoch": 6.43,
       "learning_rate": 0.0002,
-      "loss": 0.0031,
+      "loss": 0.0053,
       "step": 347
     },
     {
-      "epoch": 6.33,
+      "epoch": 6.44,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
+      "loss": 0.0103,
       "step": 348
     },
     {
-      "epoch": 6.35,
+      "epoch": 6.46,
       "learning_rate": 0.0002,
-      "loss": 0.0104,
+      "loss": 0.0107,
       "step": 349
     },
     {
-      "epoch": 6.36,
+      "epoch": 6.48,
       "learning_rate": 0.0002,
-      "loss": 0.0036,
+      "loss": 0.0098,
       "step": 350
     },
     {
-      "epoch": 6.38,
+      "epoch": 6.5,
       "learning_rate": 0.0002,
-      "loss": 0.0064,
+      "loss": 0.006,
       "step": 351
     },
     {
-      "epoch": 6.4,
+      "epoch": 6.52,
       "learning_rate": 0.0002,
-      "loss": 0.0167,
+      "loss": 0.0078,
       "step": 352
     },
     {
-      "epoch": 6.42,
+      "epoch": 6.54,
       "learning_rate": 0.0002,
-      "loss": 0.0037,
+      "loss": 0.0112,
       "step": 353
     },
     {
-      "epoch": 6.44,
+      "epoch": 6.56,
       "learning_rate": 0.0002,
-      "loss": 0.0212,
+      "loss": 0.011,
       "step": 354
     },
     {
-      "epoch": 6.45,
+      "epoch": 6.57,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0049,
       "step": 355
     },
     {
-      "epoch": 6.47,
+      "epoch": 6.59,
       "learning_rate": 0.0002,
-      "loss": 0.0063,
+      "loss": 0.0072,
       "step": 356
     },
     {
-      "epoch": 6.49,
+      "epoch": 6.61,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.005,
       "step": 357
     },
     {
-      "epoch": 6.51,
+      "epoch": 6.63,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.0075,
       "step": 358
     },
     {
-      "epoch": 6.53,
+      "epoch": 6.65,
       "learning_rate": 0.0002,
-      "loss": 0.0132,
+      "loss": 0.0064,
       "step": 359
     },
     {
-      "epoch": 6.55,
+      "epoch": 6.67,
       "learning_rate": 0.0002,
-      "loss": 0.0025,
+      "loss": 0.0046,
       "step": 360
     },
     {
-      "epoch": 6.56,
+      "epoch": 6.69,
       "learning_rate": 0.0002,
       "loss": 0.0064,
       "step": 361
     },
     {
-      "epoch": 6.58,
+      "epoch": 6.7,
       "learning_rate": 0.0002,
-      "loss": 0.0093,
+      "loss": 0.0066,
       "step": 362
     },
     {
-      "epoch": 6.6,
+      "epoch": 6.72,
       "learning_rate": 0.0002,
-      "loss": 0.0032,
+      "loss": 0.0077,
       "step": 363
     },
     {
-      "epoch": 6.62,
+      "epoch": 6.74,
       "learning_rate": 0.0002,
-      "loss": 0.007,
+      "loss": 0.009,
       "step": 364
     },
     {
-      "epoch": 6.64,
+      "epoch": 6.76,
       "learning_rate": 0.0002,
-      "loss": 0.0083,
+      "loss": 0.009,
       "step": 365
     },
     {
-      "epoch": 6.65,
+      "epoch": 6.78,
       "learning_rate": 0.0002,
-      "loss": 0.0045,
+      "loss": 0.0076,
       "step": 366
     },
     {
-      "epoch": 6.67,
+      "epoch": 6.8,
       "learning_rate": 0.0002,
-      "loss": 0.0084,
+      "loss": 0.0053,
       "step": 367
     },
     {
-      "epoch": 6.69,
+      "epoch": 6.81,
       "learning_rate": 0.0002,
-      "loss": 0.0043,
+      "loss": 0.0077,
       "step": 368
     },
     {
-      "epoch": 6.71,
+      "epoch": 6.83,
       "learning_rate": 0.0002,
-      "loss": 0.0024,
+      "loss": 0.0072,
       "step": 369
     },
     {
-      "epoch": 6.73,
+      "epoch": 6.85,
       "learning_rate": 0.0002,
-      "loss": 0.0233,
+      "loss": 0.0093,
       "step": 370
     },
     {
-      "epoch": 6.75,
+      "epoch": 6.87,
       "learning_rate": 0.0002,
-      "loss": 0.0111,
+      "loss": 0.0044,
       "step": 371
     },
     {
-      "epoch": 6.76,
+      "epoch": 6.89,
       "learning_rate": 0.0002,
-      "loss": 0.0116,
+      "loss": 0.007,
       "step": 372
     },
     {
-      "epoch": 6.78,
+      "epoch": 6.91,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.0061,
       "step": 373
     },
     {
-      "epoch": 6.8,
+      "epoch": 6.93,
       "learning_rate": 0.0002,
-      "loss": 0.0031,
+      "loss": 0.0057,
       "step": 374
     },
     {
-      "epoch": 6.82,
+      "epoch": 6.94,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0071,
       "step": 375
     },
     {
-      "epoch": 6.84,
+      "epoch": 6.96,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0084,
       "step": 376
     },
     {
-      "epoch": 6.85,
+      "epoch": 6.98,
       "learning_rate": 0.0002,
-      "loss": 0.0173,
+      "loss": 0.0073,
       "step": 377
     },
     {
-      "epoch": 6.87,
+      "epoch": 7.0,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0042,
       "step": 378
     },
     {
-      "epoch": 6.89,
+      "epoch": 7.0,
+      "eval_loss": 0.004907690919935703,
+      "eval_runtime": 138.0039,
+      "eval_samples_per_second": 12.398,
+      "eval_steps_per_second": 0.391,
+      "step": 378
+    },
+    {
+      "epoch": 7.02,
       "learning_rate": 0.0002,
-      "loss": 0.0099,
+      "loss": 0.0059,
       "step": 379
     },
     {
-      "epoch": 6.91,
+      "epoch": 7.04,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0041,
       "step": 380
     },
     {
-      "epoch": 6.93,
+      "epoch": 7.06,
       "learning_rate": 0.0002,
-      "loss": 0.0156,
+      "loss": 0.0052,
       "step": 381
     },
     {
-      "epoch": 6.95,
+      "epoch": 7.07,
       "learning_rate": 0.0002,
-      "loss": 0.0069,
+      "loss": 0.0067,
       "step": 382
     },
     {
-      "epoch": 6.96,
+      "epoch": 7.09,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0061,
       "step": 383
     },
     {
-      "epoch": 6.98,
+      "epoch": 7.11,
       "learning_rate": 0.0002,
-      "loss": 0.0074,
+      "loss": 0.0023,
       "step": 384
     },
     {
-      "epoch": 7.0,
+      "epoch": 7.13,
       "learning_rate": 0.0002,
-      "loss": 0.0074,
-      "step": 385
-    },
-    {
-      "epoch": 7.0,
-      "eval_loss": 0.007481382694095373,
-      "eval_runtime": 97.3225,
-      "eval_samples_per_second": 8.909,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0046,
       "step": 385
     },
     {
-      "epoch": 7.02,
+      "epoch": 7.15,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0043,
       "step": 386
     },
     {
-      "epoch": 7.04,
+      "epoch": 7.17,
       "learning_rate": 0.0002,
-      "loss": 0.0022,
+      "loss": 0.0071,
       "step": 387
     },
     {
-      "epoch": 7.05,
+      "epoch": 7.19,
       "learning_rate": 0.0002,
-      "loss": 0.0098,
+      "loss": 0.0068,
       "step": 388
     },
     {
-      "epoch": 7.07,
+      "epoch": 7.2,
       "learning_rate": 0.0002,
-      "loss": 0.0066,
+      "loss": 0.0039,
       "step": 389
     },
     {
-      "epoch": 7.09,
+      "epoch": 7.22,
       "learning_rate": 0.0002,
-      "loss": 0.0286,
+      "loss": 0.0041,
       "step": 390
     },
     {
-      "epoch": 7.11,
+      "epoch": 7.24,
       "learning_rate": 0.0002,
-      "loss": 0.0045,
+      "loss": 0.0027,
       "step": 391
     },
     {
-      "epoch": 7.13,
+      "epoch": 7.26,
       "learning_rate": 0.0002,
-      "loss": 0.0031,
+      "loss": 0.006,
       "step": 392
     },
     {
-      "epoch": 7.15,
+      "epoch": 7.28,
       "learning_rate": 0.0002,
-      "loss": 0.0088,
+      "loss": 0.006,
       "step": 393
     },
     {
-      "epoch": 7.16,
+      "epoch": 7.3,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.004,
       "step": 394
     },
     {
-      "epoch": 7.18,
+      "epoch": 7.31,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0041,
       "step": 395
     },
     {
-      "epoch": 7.2,
+      "epoch": 7.33,
       "learning_rate": 0.0002,
-      "loss": 0.0033,
+      "loss": 0.0045,
       "step": 396
     },
     {
-      "epoch": 7.22,
+      "epoch": 7.35,
       "learning_rate": 0.0002,
-      "loss": 0.0106,
+      "loss": 0.0048,
       "step": 397
     },
     {
-      "epoch": 7.24,
+      "epoch": 7.37,
       "learning_rate": 0.0002,
-      "loss": 0.0142,
+      "loss": 0.0062,
       "step": 398
     },
     {
-      "epoch": 7.25,
+      "epoch": 7.39,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.0055,
       "step": 399
     },
     {
-      "epoch": 7.27,
+      "epoch": 7.41,
       "learning_rate": 0.0002,
-      "loss": 0.0058,
+      "loss": 0.0061,
       "step": 400
     },
     {
-      "epoch": 7.29,
+      "epoch": 7.43,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0057,
       "step": 401
     },
     {
-      "epoch": 7.31,
+      "epoch": 7.44,
       "learning_rate": 0.0002,
-      "loss": 0.0026,
+      "loss": 0.006,
       "step": 402
     },
     {
-      "epoch": 7.33,
+      "epoch": 7.46,
       "learning_rate": 0.0002,
-      "loss": 0.0048,
+      "loss": 0.0051,
       "step": 403
     },
     {
-      "epoch": 7.35,
+      "epoch": 7.48,
       "learning_rate": 0.0002,
-      "loss": 0.006,
+      "loss": 0.0081,
       "step": 404
     },
     {
-      "epoch": 7.36,
+      "epoch": 7.5,
       "learning_rate": 0.0002,
-      "loss": 0.0111,
+      "loss": 0.0055,
       "step": 405
     },
     {
-      "epoch": 7.38,
+      "epoch": 7.52,
       "learning_rate": 0.0002,
-      "loss": 0.0028,
+      "loss": 0.0058,
       "step": 406
     },
     {
-      "epoch": 7.4,
+      "epoch": 7.54,
       "learning_rate": 0.0002,
-      "loss": 0.0122,
+      "loss": 0.0084,
       "step": 407
     },
     {
-      "epoch": 7.42,
+      "epoch": 7.56,
       "learning_rate": 0.0002,
-      "loss": 0.0089,
+      "loss": 0.0034,
       "step": 408
     },
     {
-      "epoch": 7.44,
+      "epoch": 7.57,
       "learning_rate": 0.0002,
-      "loss": 0.0158,
+      "loss": 0.0043,
       "step": 409
     },
     {
-      "epoch": 7.45,
+      "epoch": 7.59,
       "learning_rate": 0.0002,
-      "loss": 0.0048,
+      "loss": 0.0071,
       "step": 410
     },
     {
-      "epoch": 7.47,
+      "epoch": 7.61,
       "learning_rate": 0.0002,
-      "loss": 0.0045,
+      "loss": 0.0078,
       "step": 411
     },
     {
-      "epoch": 7.49,
+      "epoch": 7.63,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0059,
       "step": 412
     },
     {
-      "epoch": 7.51,
+      "epoch": 7.65,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0063,
       "step": 413
     },
     {
-      "epoch": 7.53,
+      "epoch": 7.67,
       "learning_rate": 0.0002,
-      "loss": 0.0048,
+      "loss": 0.0054,
       "step": 414
     },
     {
-      "epoch": 7.55,
+      "epoch": 7.69,
       "learning_rate": 0.0002,
-      "loss": 0.0055,
+      "loss": 0.0063,
       "step": 415
     },
     {
-      "epoch": 7.56,
+      "epoch": 7.7,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0078,
       "step": 416
     },
     {
-      "epoch": 7.58,
+      "epoch": 7.72,
       "learning_rate": 0.0002,
-      "loss": 0.006,
+      "loss": 0.0088,
       "step": 417
     },
     {
-      "epoch": 7.6,
+      "epoch": 7.74,
       "learning_rate": 0.0002,
-      "loss": 0.0122,
+      "loss": 0.0107,
       "step": 418
     },
     {
-      "epoch": 7.62,
+      "epoch": 7.76,
       "learning_rate": 0.0002,
-      "loss": 0.0064,
+      "loss": 0.0037,
       "step": 419
     },
     {
-      "epoch": 7.64,
+      "epoch": 7.78,
       "learning_rate": 0.0002,
-      "loss": 0.0099,
+      "loss": 0.0067,
       "step": 420
     },
     {
-      "epoch": 7.65,
+      "epoch": 7.8,
       "learning_rate": 0.0002,
-      "loss": 0.0158,
+      "loss": 0.0085,
       "step": 421
     },
     {
-      "epoch": 7.67,
+      "epoch": 7.81,
       "learning_rate": 0.0002,
-      "loss": 0.0042,
+      "loss": 0.0055,
       "step": 422
     },
     {
-      "epoch": 7.69,
+      "epoch": 7.83,
       "learning_rate": 0.0002,
-      "loss": 0.0113,
+      "loss": 0.0108,
       "step": 423
     },
     {
-      "epoch": 7.71,
+      "epoch": 7.85,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0103,
       "step": 424
     },
     {
-      "epoch": 7.73,
+      "epoch": 7.87,
       "learning_rate": 0.0002,
-      "loss": 0.0064,
+      "loss": 0.0119,
       "step": 425
     },
     {
-      "epoch": 7.75,
+      "epoch": 7.89,
       "learning_rate": 0.0002,
-      "loss": 0.0131,
+      "loss": 0.0058,
       "step": 426
     },
     {
-      "epoch": 7.76,
+      "epoch": 7.91,
       "learning_rate": 0.0002,
-      "loss": 0.0147,
+      "loss": 0.0043,
       "step": 427
     },
     {
-      "epoch": 7.78,
+      "epoch": 7.93,
       "learning_rate": 0.0002,
-      "loss": 0.0099,
+      "loss": 0.0106,
       "step": 428
     },
     {
-      "epoch": 7.8,
+      "epoch": 7.94,
       "learning_rate": 0.0002,
-      "loss": 0.0079,
+      "loss": 0.006,
       "step": 429
     },
     {
-      "epoch": 7.82,
+      "epoch": 7.96,
       "learning_rate": 0.0002,
-      "loss": 0.0036,
+      "loss": 0.0084,
       "step": 430
     },
     {
-      "epoch": 7.84,
+      "epoch": 7.98,
       "learning_rate": 0.0002,
-      "loss": 0.0056,
+      "loss": 0.0121,
       "step": 431
     },
     {
-      "epoch": 7.85,
+      "epoch": 8.0,
       "learning_rate": 0.0002,
-      "loss": 0.0084,
+      "loss": 0.0054,
       "step": 432
     },
     {
-      "epoch": 7.87,
+      "epoch": 8.0,
+      "eval_loss": 0.005581296514719725,
+      "eval_runtime": 138.2508,
+      "eval_samples_per_second": 12.376,
+      "eval_steps_per_second": 0.391,
+      "step": 432
+    },
+    {
+      "epoch": 8.02,
       "learning_rate": 0.0002,
-      "loss": 0.0083,
+      "loss": 0.008,
       "step": 433
     },
     {
-      "epoch": 7.89,
+      "epoch": 8.04,
       "learning_rate": 0.0002,
-      "loss": 0.019,
+      "loss": 0.0079,
       "step": 434
     },
     {
-      "epoch": 7.91,
+      "epoch": 8.06,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.0049,
       "step": 435
     },
     {
-      "epoch": 7.93,
+      "epoch": 8.07,
       "learning_rate": 0.0002,
-      "loss": 0.0083,
+      "loss": 0.0057,
       "step": 436
     },
     {
-      "epoch": 7.95,
+      "epoch": 8.09,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
+      "loss": 0.0085,
       "step": 437
     },
     {
-      "epoch": 7.96,
+      "epoch": 8.11,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0055,
       "step": 438
     },
     {
-      "epoch": 7.98,
+      "epoch": 8.13,
       "learning_rate": 0.0002,
-      "loss": 0.0097,
+      "loss": 0.0067,
       "step": 439
     },
     {
-      "epoch": 8.0,
+      "epoch": 8.15,
       "learning_rate": 0.0002,
-      "loss": 0.0205,
-      "step": 440
-    },
-    {
-      "epoch": 8.0,
-      "eval_loss": 0.009691957384347916,
-      "eval_runtime": 97.3628,
-      "eval_samples_per_second": 8.905,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0069,
       "step": 440
     },
     {
-      "epoch": 8.02,
+      "epoch": 8.17,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0055,
       "step": 441
     },
     {
-      "epoch": 8.04,
+      "epoch": 8.19,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
+      "loss": 0.0081,
       "step": 442
     },
     {
-      "epoch": 8.05,
+      "epoch": 8.2,
       "learning_rate": 0.0002,
-      "loss": 0.0055,
+      "loss": 0.0064,
       "step": 443
     },
     {
-      "epoch": 8.07,
+      "epoch": 8.22,
       "learning_rate": 0.0002,
       "loss": 0.0105,
       "step": 444
     },
     {
-      "epoch": 8.09,
+      "epoch": 8.24,
       "learning_rate": 0.0002,
-      "loss": 0.0172,
+      "loss": 0.0023,
       "step": 445
     },
     {
-      "epoch": 8.11,
+      "epoch": 8.26,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0071,
       "step": 446
     },
     {
-      "epoch": 8.13,
+      "epoch": 8.28,
       "learning_rate": 0.0002,
-      "loss": 0.0057,
+      "loss": 0.0125,
       "step": 447
     },
     {
-      "epoch": 8.15,
+      "epoch": 8.3,
       "learning_rate": 0.0002,
-      "loss": 0.0239,
+      "loss": 0.0076,
       "step": 448
     },
     {
-      "epoch": 8.16,
+      "epoch": 8.31,
       "learning_rate": 0.0002,
-      "loss": 0.0048,
+      "loss": 0.0047,
       "step": 449
     },
     {
-      "epoch": 8.18,
+      "epoch": 8.33,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.0104,
       "step": 450
     },
     {
-      "epoch": 8.2,
+      "epoch": 8.35,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0039,
       "step": 451
     },
     {
-      "epoch": 8.22,
+      "epoch": 8.37,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0083,
       "step": 452
     },
     {
-      "epoch": 8.24,
+      "epoch": 8.39,
       "learning_rate": 0.0002,
-      "loss": 0.0043,
+      "loss": 0.0063,
       "step": 453
     },
     {
-      "epoch": 8.25,
+      "epoch": 8.41,
       "learning_rate": 0.0002,
-      "loss": 0.006,
+      "loss": 0.0046,
       "step": 454
     },
     {
-      "epoch": 8.27,
+      "epoch": 8.43,
       "learning_rate": 0.0002,
-      "loss": 0.0079,
+      "loss": 0.0093,
       "step": 455
     },
     {
-      "epoch": 8.29,
+      "epoch": 8.44,
       "learning_rate": 0.0002,
-      "loss": 0.0031,
+      "loss": 0.0087,
       "step": 456
     },
     {
-      "epoch": 8.31,
+      "epoch": 8.46,
       "learning_rate": 0.0002,
-      "loss": 0.0069,
+      "loss": 0.0095,
       "step": 457
     },
     {
-      "epoch": 8.33,
+      "epoch": 8.48,
       "learning_rate": 0.0002,
-      "loss": 0.0119,
+      "loss": 0.0061,
       "step": 458
     },
     {
-      "epoch": 8.35,
+      "epoch": 8.5,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0074,
       "step": 459
     },
     {
-      "epoch": 8.36,
+      "epoch": 8.52,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
+      "loss": 0.0047,
       "step": 460
     },
     {
-      "epoch": 8.38,
+      "epoch": 8.54,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0135,
       "step": 461
     },
     {
-      "epoch": 8.4,
+      "epoch": 8.56,
       "learning_rate": 0.0002,
-      "loss": 0.0103,
+      "loss": 0.0033,
       "step": 462
     },
     {
-      "epoch": 8.42,
+      "epoch": 8.57,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0062,
       "step": 463
     },
     {
-      "epoch": 8.44,
+      "epoch": 8.59,
       "learning_rate": 0.0002,
-      "loss": 0.0038,
+      "loss": 0.0071,
       "step": 464
     },
     {
-      "epoch": 8.45,
+      "epoch": 8.61,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.0049,
       "step": 465
     },
     {
-      "epoch": 8.47,
+      "epoch": 8.63,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.0062,
       "step": 466
     },
     {
-      "epoch": 8.49,
+      "epoch": 8.65,
       "learning_rate": 0.0002,
-      "loss": 0.0058,
+      "loss": 0.006,
       "step": 467
     },
     {
-      "epoch": 8.51,
+      "epoch": 8.67,
       "learning_rate": 0.0002,
-      "loss": 0.0127,
+      "loss": 0.0035,
       "step": 468
     },
     {
-      "epoch": 8.53,
+      "epoch": 8.69,
       "learning_rate": 0.0002,
-      "loss": 0.0138,
+      "loss": 0.0068,
       "step": 469
     },
     {
-      "epoch": 8.55,
+      "epoch": 8.7,
       "learning_rate": 0.0002,
-      "loss": 0.009,
+      "loss": 0.0074,
       "step": 470
     },
     {
-      "epoch": 8.56,
+      "epoch": 8.72,
       "learning_rate": 0.0002,
-      "loss": 0.0277,
+      "loss": 0.0065,
       "step": 471
     },
     {
-      "epoch": 8.58,
+      "epoch": 8.74,
       "learning_rate": 0.0002,
-      "loss": 0.0042,
+      "loss": 0.0071,
       "step": 472
     },
     {
-      "epoch": 8.6,
+      "epoch": 8.76,
       "learning_rate": 0.0002,
-      "loss": 0.0184,
+      "loss": 0.0096,
       "step": 473
     },
     {
-      "epoch": 8.62,
+      "epoch": 8.78,
       "learning_rate": 0.0002,
-      "loss": 0.013,
+      "loss": 0.0076,
       "step": 474
     },
     {
-      "epoch": 8.64,
+      "epoch": 8.8,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0063,
       "step": 475
     },
     {
-      "epoch": 8.65,
+      "epoch": 8.81,
       "learning_rate": 0.0002,
-      "loss": 0.0207,
+      "loss": 0.005,
       "step": 476
     },
     {
-      "epoch": 8.67,
+      "epoch": 8.83,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.0065,
       "step": 477
     },
     {
-      "epoch": 8.69,
+      "epoch": 8.85,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.0048,
       "step": 478
     },
     {
-      "epoch": 8.71,
+      "epoch": 8.87,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0094,
       "step": 479
     },
     {
-      "epoch": 8.73,
+      "epoch": 8.89,
       "learning_rate": 0.0002,
-      "loss": 0.009,
+      "loss": 0.0093,
       "step": 480
     },
     {
-      "epoch": 8.75,
+      "epoch": 8.91,
       "learning_rate": 0.0002,
-      "loss": 0.0043,
+      "loss": 0.0086,
       "step": 481
     },
     {
-      "epoch": 8.76,
+      "epoch": 8.93,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0042,
       "step": 482
     },
     {
-      "epoch": 8.78,
+      "epoch": 8.94,
       "learning_rate": 0.0002,
-      "loss": 0.012,
+      "loss": 0.0049,
       "step": 483
     },
     {
-      "epoch": 8.8,
+      "epoch": 8.96,
       "learning_rate": 0.0002,
-      "loss": 0.0098,
+      "loss": 0.0081,
       "step": 484
     },
     {
-      "epoch": 8.82,
+      "epoch": 8.98,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0116,
       "step": 485
     },
     {
-      "epoch": 8.84,
+      "epoch": 9.0,
       "learning_rate": 0.0002,
-      "loss": 0.0112,
+      "loss": 0.0036,
       "step": 486
     },
     {
-      "epoch": 8.85,
+      "epoch": 9.0,
+      "eval_loss": 0.004996659234166145,
+      "eval_runtime": 138.2117,
+      "eval_samples_per_second": 12.38,
+      "eval_steps_per_second": 0.391,
+      "step": 486
+    },
+    {
+      "epoch": 9.02,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.0045,
       "step": 487
     },
     {
-      "epoch": 8.87,
+      "epoch": 9.04,
       "learning_rate": 0.0002,
-      "loss": 0.0243,
+      "loss": 0.0036,
       "step": 488
     },
     {
-      "epoch": 8.89,
+      "epoch": 9.06,
       "learning_rate": 0.0002,
-      "loss": 0.0097,
+      "loss": 0.0038,
       "step": 489
     },
     {
-      "epoch": 8.91,
+      "epoch": 9.07,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.003,
       "step": 490
     },
     {
-      "epoch": 8.93,
+      "epoch": 9.09,
       "learning_rate": 0.0002,
-      "loss": 0.0097,
+      "loss": 0.0045,
       "step": 491
     },
     {
-      "epoch": 8.95,
+      "epoch": 9.11,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0079,
       "step": 492
     },
     {
-      "epoch": 8.96,
+      "epoch": 9.13,
       "learning_rate": 0.0002,
-      "loss": 0.0088,
+      "loss": 0.006,
       "step": 493
     },
     {
-      "epoch": 8.98,
+      "epoch": 9.15,
       "learning_rate": 0.0002,
-      "loss": 0.0185,
+      "loss": 0.0066,
       "step": 494
     },
     {
-      "epoch": 9.0,
+      "epoch": 9.17,
       "learning_rate": 0.0002,
-      "loss": 0.009,
-      "step": 495
-    },
-    {
-      "epoch": 9.0,
-      "eval_loss": 0.006804489530622959,
-      "eval_runtime": 97.3228,
-      "eval_samples_per_second": 8.909,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0041,
       "step": 495
     },
     {
-      "epoch": 9.02,
+      "epoch": 9.19,
       "learning_rate": 0.0002,
-      "loss": 0.0014,
+      "loss": 0.0057,
       "step": 496
     },
     {
-      "epoch": 9.04,
+      "epoch": 9.2,
       "learning_rate": 0.0002,
-      "loss": 0.0144,
+      "loss": 0.0038,
       "step": 497
     },
     {
-      "epoch": 9.05,
+      "epoch": 9.22,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0061,
       "step": 498
     },
     {
-      "epoch": 9.07,
+      "epoch": 9.24,
       "learning_rate": 0.0002,
-      "loss": 0.006,
+      "loss": 0.007,
       "step": 499
     },
     {
-      "epoch": 9.09,
+      "epoch": 9.26,
       "learning_rate": 0.0002,
-      "loss": 0.0015,
+      "loss": 0.0101,
       "step": 500
     },
     {
-      "epoch": 9.11,
+      "epoch": 9.28,
       "learning_rate": 0.0002,
-      "loss": 0.0053,
+      "loss": 0.0036,
       "step": 501
     },
     {
-      "epoch": 9.13,
+      "epoch": 9.3,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0061,
       "step": 502
     },
     {
-      "epoch": 9.15,
+      "epoch": 9.31,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.0066,
       "step": 503
     },
     {
-      "epoch": 9.16,
+      "epoch": 9.33,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0057,
       "step": 504
     },
     {
-      "epoch": 9.18,
+      "epoch": 9.35,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.0078,
       "step": 505
     },
     {
-      "epoch": 9.2,
+      "epoch": 9.37,
       "learning_rate": 0.0002,
-      "loss": 0.0125,
+      "loss": 0.0052,
       "step": 506
     },
     {
-      "epoch": 9.22,
+      "epoch": 9.39,
       "learning_rate": 0.0002,
-      "loss": 0.0106,
+      "loss": 0.0091,
       "step": 507
     },
     {
-      "epoch": 9.24,
+      "epoch": 9.41,
       "learning_rate": 0.0002,
-      "loss": 0.017,
+      "loss": 0.0054,
       "step": 508
     },
     {
-      "epoch": 9.25,
+      "epoch": 9.43,
       "learning_rate": 0.0002,
-      "loss": 0.0023,
+      "loss": 0.0128,
       "step": 509
     },
     {
-      "epoch": 9.27,
+      "epoch": 9.44,
       "learning_rate": 0.0002,
-      "loss": 0.0131,
+      "loss": 0.0084,
       "step": 510
     },
     {
-      "epoch": 9.29,
+      "epoch": 9.46,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0043,
       "step": 511
     },
     {
-      "epoch": 9.31,
+      "epoch": 9.48,
       "learning_rate": 0.0002,
-      "loss": 0.007,
+      "loss": 0.0045,
       "step": 512
     },
     {
-      "epoch": 9.33,
+      "epoch": 9.5,
       "learning_rate": 0.0002,
-      "loss": 0.0098,
+      "loss": 0.0075,
       "step": 513
     },
     {
-      "epoch": 9.35,
+      "epoch": 9.52,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0089,
       "step": 514
     },
     {
-      "epoch": 9.36,
+      "epoch": 9.54,
       "learning_rate": 0.0002,
-      "loss": 0.0141,
+      "loss": 0.006,
       "step": 515
     },
     {
-      "epoch": 9.38,
+      "epoch": 9.56,
       "learning_rate": 0.0002,
-      "loss": 0.008,
+      "loss": 0.009,
       "step": 516
     },
     {
-      "epoch": 9.4,
+      "epoch": 9.57,
       "learning_rate": 0.0002,
-      "loss": 0.0045,
+      "loss": 0.0082,
       "step": 517
     },
     {
-      "epoch": 9.42,
+      "epoch": 9.59,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.0094,
       "step": 518
     },
     {
-      "epoch": 9.44,
+      "epoch": 9.61,
       "learning_rate": 0.0002,
-      "loss": 0.0037,
+      "loss": 0.0049,
       "step": 519
     },
     {
-      "epoch": 9.45,
+      "epoch": 9.63,
       "learning_rate": 0.0002,
-      "loss": 0.0095,
+      "loss": 0.0093,
       "step": 520
     },
     {
-      "epoch": 9.47,
+      "epoch": 9.65,
       "learning_rate": 0.0002,
-      "loss": 0.0108,
+      "loss": 0.0087,
       "step": 521
     },
     {
-      "epoch": 9.49,
+      "epoch": 9.67,
       "learning_rate": 0.0002,
-      "loss": 0.027,
+      "loss": 0.0056,
       "step": 522
     },
     {
-      "epoch": 9.51,
+      "epoch": 9.69,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0122,
       "step": 523
     },
     {
-      "epoch": 9.53,
+      "epoch": 9.7,
       "learning_rate": 0.0002,
-      "loss": 0.0052,
+      "loss": 0.0054,
       "step": 524
     },
     {
-      "epoch": 9.55,
+      "epoch": 9.72,
       "learning_rate": 0.0002,
-      "loss": 0.012,
+      "loss": 0.0043,
       "step": 525
     },
     {
-      "epoch": 9.56,
+      "epoch": 9.74,
       "learning_rate": 0.0002,
-      "loss": 0.0215,
+      "loss": 0.0055,
       "step": 526
     },
     {
-      "epoch": 9.58,
+      "epoch": 9.76,
       "learning_rate": 0.0002,
-      "loss": 0.0145,
+      "loss": 0.0047,
       "step": 527
     },
     {
-      "epoch": 9.6,
+      "epoch": 9.78,
       "learning_rate": 0.0002,
-      "loss": 0.0081,
+      "loss": 0.0084,
       "step": 528
     },
     {
-      "epoch": 9.62,
+      "epoch": 9.8,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0051,
       "step": 529
     },
     {
-      "epoch": 9.64,
+      "epoch": 9.81,
       "learning_rate": 0.0002,
-      "loss": 0.0116,
+      "loss": 0.0048,
       "step": 530
     },
     {
-      "epoch": 9.65,
+      "epoch": 9.83,
       "learning_rate": 0.0002,
-      "loss": 0.0078,
+      "loss": 0.0061,
       "step": 531
     },
     {
-      "epoch": 9.67,
+      "epoch": 9.85,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.0046,
       "step": 532
     },
     {
-      "epoch": 9.69,
+      "epoch": 9.87,
       "learning_rate": 0.0002,
-      "loss": 0.0138,
+      "loss": 0.0068,
       "step": 533
     },
     {
-      "epoch": 9.71,
+      "epoch": 9.89,
       "learning_rate": 0.0002,
-      "loss": 0.0141,
+      "loss": 0.0104,
       "step": 534
     },
     {
-      "epoch": 9.73,
+      "epoch": 9.91,
       "learning_rate": 0.0002,
-      "loss": 0.0115,
+      "loss": 0.0035,
       "step": 535
     },
     {
-      "epoch": 9.75,
+      "epoch": 9.93,
       "learning_rate": 0.0002,
-      "loss": 0.0144,
+      "loss": 0.0047,
       "step": 536
     },
     {
-      "epoch": 9.76,
+      "epoch": 9.94,
       "learning_rate": 0.0002,
-      "loss": 0.0069,
+      "loss": 0.0056,
       "step": 537
     },
     {
-      "epoch": 9.78,
+      "epoch": 9.96,
       "learning_rate": 0.0002,
-      "loss": 0.0113,
+      "loss": 0.0036,
       "step": 538
     },
     {
-      "epoch": 9.8,
+      "epoch": 9.98,
       "learning_rate": 0.0002,
-      "loss": 0.008,
+      "loss": 0.0081,
       "step": 539
     },
     {
-      "epoch": 9.82,
+      "epoch": 10.0,
       "learning_rate": 0.0002,
-      "loss": 0.0116,
+      "loss": 0.0073,
+      "step": 540
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 0.0059353057295084,
+      "eval_runtime": 138.1046,
+      "eval_samples_per_second": 12.389,
+      "eval_steps_per_second": 0.391,
       "step": 540
     },
     {
-      "epoch": 9.84,
+      "epoch": 10.02,
       "learning_rate": 0.0002,
-      "loss": 0.0214,
+      "loss": 0.0059,
       "step": 541
     },
     {
-      "epoch": 9.85,
+      "epoch": 10.04,
       "learning_rate": 0.0002,
-      "loss": 0.0128,
+      "loss": 0.0063,
       "step": 542
     },
     {
-      "epoch": 9.87,
+      "epoch": 10.06,
       "learning_rate": 0.0002,
-      "loss": 0.0102,
+      "loss": 0.0073,
       "step": 543
     },
     {
-      "epoch": 9.89,
+      "epoch": 10.07,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0078,
       "step": 544
     },
     {
-      "epoch": 9.91,
+      "epoch": 10.09,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0049,
       "step": 545
     },
     {
-      "epoch": 9.93,
+      "epoch": 10.11,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
+      "loss": 0.009,
       "step": 546
     },
     {
-      "epoch": 9.95,
+      "epoch": 10.13,
       "learning_rate": 0.0002,
-      "loss": 0.0089,
+      "loss": 0.0029,
       "step": 547
     },
     {
-      "epoch": 9.96,
+      "epoch": 10.15,
       "learning_rate": 0.0002,
-      "loss": 0.0107,
+      "loss": 0.0067,
       "step": 548
     },
     {
-      "epoch": 9.98,
+      "epoch": 10.17,
       "learning_rate": 0.0002,
-      "loss": 0.0047,
+      "loss": 0.0032,
       "step": 549
     },
     {
-      "epoch": 10.0,
+      "epoch": 10.19,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
-      "step": 550
-    },
-    {
-      "epoch": 10.0,
-      "eval_loss": 0.007513745687901974,
-      "eval_runtime": 97.3175,
-      "eval_samples_per_second": 8.909,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0109,
       "step": 550
     },
     {
-      "epoch": 10.02,
+      "epoch": 10.2,
       "learning_rate": 0.0002,
-      "loss": 0.0226,
+      "loss": 0.0028,
       "step": 551
     },
     {
-      "epoch": 10.04,
+      "epoch": 10.22,
       "learning_rate": 0.0002,
-      "loss": 0.0014,
+      "loss": 0.0042,
       "step": 552
     },
     {
-      "epoch": 10.05,
+      "epoch": 10.24,
       "learning_rate": 0.0002,
-      "loss": 0.0133,
+      "loss": 0.0087,
       "step": 553
     },
     {
-      "epoch": 10.07,
+      "epoch": 10.26,
       "learning_rate": 0.0002,
-      "loss": 0.0073,
+      "loss": 0.0036,
       "step": 554
     },
     {
-      "epoch": 10.09,
+      "epoch": 10.28,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.0061,
       "step": 555
     },
     {
-      "epoch": 10.11,
+      "epoch": 10.3,
       "learning_rate": 0.0002,
-      "loss": 0.0127,
+      "loss": 0.0085,
       "step": 556
     },
     {
-      "epoch": 10.13,
+      "epoch": 10.31,
       "learning_rate": 0.0002,
       "loss": 0.0057,
       "step": 557
     },
     {
-      "epoch": 10.15,
+      "epoch": 10.33,
       "learning_rate": 0.0002,
-      "loss": 0.0055,
+      "loss": 0.0093,
       "step": 558
     },
     {
-      "epoch": 10.16,
+      "epoch": 10.35,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0061,
       "step": 559
     },
     {
-      "epoch": 10.18,
+      "epoch": 10.37,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
+      "loss": 0.0117,
       "step": 560
     },
     {
-      "epoch": 10.2,
+      "epoch": 10.39,
       "learning_rate": 0.0002,
-      "loss": 0.0185,
+      "loss": 0.0088,
       "step": 561
     },
     {
-      "epoch": 10.22,
+      "epoch": 10.41,
       "learning_rate": 0.0002,
-      "loss": 0.0115,
+      "loss": 0.0124,
       "step": 562
     },
     {
-      "epoch": 10.24,
+      "epoch": 10.43,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.0098,
       "step": 563
     },
     {
-      "epoch": 10.25,
+      "epoch": 10.44,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.006,
       "step": 564
     },
     {
-      "epoch": 10.27,
+      "epoch": 10.46,
       "learning_rate": 0.0002,
-      "loss": 0.0086,
+      "loss": 0.0078,
       "step": 565
     },
     {
-      "epoch": 10.29,
+      "epoch": 10.48,
       "learning_rate": 0.0002,
-      "loss": 0.0086,
+      "loss": 0.0103,
       "step": 566
     },
     {
-      "epoch": 10.31,
+      "epoch": 10.5,
       "learning_rate": 0.0002,
-      "loss": 0.0074,
+      "loss": 0.0048,
       "step": 567
     },
     {
-      "epoch": 10.33,
+      "epoch": 10.52,
       "learning_rate": 0.0002,
-      "loss": 0.0056,
+      "loss": 0.0081,
       "step": 568
     },
     {
-      "epoch": 10.35,
+      "epoch": 10.54,
       "learning_rate": 0.0002,
-      "loss": 0.0096,
+      "loss": 0.0076,
       "step": 569
     },
     {
-      "epoch": 10.36,
+      "epoch": 10.56,
       "learning_rate": 0.0002,
-      "loss": 0.0053,
+      "loss": 0.006,
       "step": 570
     },
     {
-      "epoch": 10.38,
+      "epoch": 10.57,
       "learning_rate": 0.0002,
-      "loss": 0.0037,
+      "loss": 0.0053,
       "step": 571
     },
     {
-      "epoch": 10.4,
+      "epoch": 10.59,
       "learning_rate": 0.0002,
-      "loss": 0.0042,
+      "loss": 0.0078,
       "step": 572
     },
     {
-      "epoch": 10.42,
+      "epoch": 10.61,
       "learning_rate": 0.0002,
-      "loss": 0.0126,
+      "loss": 0.0062,
       "step": 573
     },
     {
-      "epoch": 10.44,
+      "epoch": 10.63,
       "learning_rate": 0.0002,
-      "loss": 0.0085,
+      "loss": 0.005,
       "step": 574
     },
     {
-      "epoch": 10.45,
+      "epoch": 10.65,
       "learning_rate": 0.0002,
-      "loss": 0.0123,
+      "loss": 0.0047,
       "step": 575
     },
     {
-      "epoch": 10.47,
+      "epoch": 10.67,
       "learning_rate": 0.0002,
-      "loss": 0.0113,
+      "loss": 0.0057,
       "step": 576
     },
     {
-      "epoch": 10.49,
+      "epoch": 10.69,
       "learning_rate": 0.0002,
-      "loss": 0.0078,
+      "loss": 0.0041,
       "step": 577
     },
     {
-      "epoch": 10.51,
+      "epoch": 10.7,
       "learning_rate": 0.0002,
-      "loss": 0.0056,
+      "loss": 0.0055,
       "step": 578
     },
     {
-      "epoch": 10.53,
+      "epoch": 10.72,
       "learning_rate": 0.0002,
-      "loss": 0.0114,
+      "loss": 0.0082,
       "step": 579
     },
     {
-      "epoch": 10.55,
+      "epoch": 10.74,
       "learning_rate": 0.0002,
-      "loss": 0.0128,
+      "loss": 0.004,
       "step": 580
     },
     {
-      "epoch": 10.56,
+      "epoch": 10.76,
       "learning_rate": 0.0002,
-      "loss": 0.012,
+      "loss": 0.0055,
       "step": 581
     },
     {
-      "epoch": 10.58,
+      "epoch": 10.78,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0098,
       "step": 582
     },
     {
-      "epoch": 10.6,
+      "epoch": 10.8,
       "learning_rate": 0.0002,
-      "loss": 0.0192,
+      "loss": 0.0048,
       "step": 583
     },
     {
-      "epoch": 10.62,
+      "epoch": 10.81,
       "learning_rate": 0.0002,
-      "loss": 0.0307,
+      "loss": 0.006,
       "step": 584
     },
     {
-      "epoch": 10.64,
+      "epoch": 10.83,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0058,
       "step": 585
     },
     {
-      "epoch": 10.65,
+      "epoch": 10.85,
       "learning_rate": 0.0002,
-      "loss": 0.0093,
+      "loss": 0.0087,
       "step": 586
     },
     {
-      "epoch": 10.67,
+      "epoch": 10.87,
       "learning_rate": 0.0002,
-      "loss": 0.0103,
+      "loss": 0.0045,
       "step": 587
     },
     {
-      "epoch": 10.69,
+      "epoch": 10.89,
       "learning_rate": 0.0002,
-      "loss": 0.014,
+      "loss": 0.0054,
       "step": 588
     },
     {
-      "epoch": 10.71,
+      "epoch": 10.91,
       "learning_rate": 0.0002,
-      "loss": 0.0086,
+      "loss": 0.0079,
       "step": 589
     },
     {
-      "epoch": 10.73,
+      "epoch": 10.93,
       "learning_rate": 0.0002,
-      "loss": 0.0076,
+      "loss": 0.0064,
       "step": 590
     },
     {
-      "epoch": 10.75,
+      "epoch": 10.94,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0066,
       "step": 591
     },
     {
-      "epoch": 10.76,
+      "epoch": 10.96,
       "learning_rate": 0.0002,
-      "loss": 0.0126,
+      "loss": 0.0063,
       "step": 592
     },
     {
-      "epoch": 10.78,
+      "epoch": 10.98,
       "learning_rate": 0.0002,
-      "loss": 0.0132,
+      "loss": 0.0106,
       "step": 593
     },
     {
-      "epoch": 10.8,
+      "epoch": 11.0,
       "learning_rate": 0.0002,
-      "loss": 0.0012,
+      "loss": 0.0029,
       "step": 594
     },
     {
-      "epoch": 10.82,
+      "epoch": 11.0,
+      "eval_loss": 0.005342724267393351,
+      "eval_runtime": 138.1544,
+      "eval_samples_per_second": 12.385,
+      "eval_steps_per_second": 0.391,
+      "step": 594
+    },
+    {
+      "epoch": 11.02,
       "learning_rate": 0.0002,
-      "loss": 0.0044,
+      "loss": 0.0047,
       "step": 595
     },
     {
-      "epoch": 10.84,
+      "epoch": 11.04,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0067,
       "step": 596
     },
     {
-      "epoch": 10.85,
+      "epoch": 11.06,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0051,
       "step": 597
     },
     {
-      "epoch": 10.87,
+      "epoch": 11.07,
       "learning_rate": 0.0002,
-      "loss": 0.008,
+      "loss": 0.0038,
       "step": 598
     },
     {
-      "epoch": 10.89,
+      "epoch": 11.09,
       "learning_rate": 0.0002,
       "loss": 0.0078,
       "step": 599
     },
     {
-      "epoch": 10.91,
+      "epoch": 11.11,
       "learning_rate": 0.0002,
-      "loss": 0.0097,
+      "loss": 0.0083,
       "step": 600
     },
     {
-      "epoch": 10.93,
+      "epoch": 11.13,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.0043,
       "step": 601
     },
     {
-      "epoch": 10.95,
+      "epoch": 11.15,
       "learning_rate": 0.0002,
-      "loss": 0.0114,
+      "loss": 0.007,
       "step": 602
     },
     {
-      "epoch": 10.96,
+      "epoch": 11.17,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0037,
       "step": 603
     },
     {
-      "epoch": 10.98,
+      "epoch": 11.19,
       "learning_rate": 0.0002,
-      "loss": 0.0249,
+      "loss": 0.0053,
       "step": 604
     },
     {
-      "epoch": 11.0,
+      "epoch": 11.2,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
-      "step": 605
-    },
-    {
-      "epoch": 11.0,
-      "eval_loss": 0.006923212669789791,
-      "eval_runtime": 97.4302,
-      "eval_samples_per_second": 8.899,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.005,
       "step": 605
     },
     {
-      "epoch": 11.02,
+      "epoch": 11.22,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.0054,
       "step": 606
     },
     {
-      "epoch": 11.04,
+      "epoch": 11.24,
       "learning_rate": 0.0002,
-      "loss": 0.0047,
+      "loss": 0.0069,
       "step": 607
     },
     {
-      "epoch": 11.05,
+      "epoch": 11.26,
       "learning_rate": 0.0002,
-      "loss": 0.006,
+      "loss": 0.0022,
       "step": 608
     },
     {
-      "epoch": 11.07,
+      "epoch": 11.28,
       "learning_rate": 0.0002,
-      "loss": 0.0164,
+      "loss": 0.0072,
       "step": 609
     },
     {
-      "epoch": 11.09,
+      "epoch": 11.3,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
+      "loss": 0.0141,
       "step": 610
     },
     {
-      "epoch": 11.11,
+      "epoch": 11.31,
       "learning_rate": 0.0002,
-      "loss": 0.0145,
+      "loss": 0.0058,
       "step": 611
     },
     {
-      "epoch": 11.13,
+      "epoch": 11.33,
       "learning_rate": 0.0002,
-      "loss": 0.0108,
+      "loss": 0.0081,
       "step": 612
     },
     {
-      "epoch": 11.15,
+      "epoch": 11.35,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.0078,
       "step": 613
     },
     {
-      "epoch": 11.16,
+      "epoch": 11.37,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.0131,
       "step": 614
     },
     {
-      "epoch": 11.18,
+      "epoch": 11.39,
       "learning_rate": 0.0002,
-      "loss": 0.0123,
+      "loss": 0.0078,
       "step": 615
     },
     {
-      "epoch": 11.2,
+      "epoch": 11.41,
       "learning_rate": 0.0002,
-      "loss": 0.0053,
+      "loss": 0.01,
       "step": 616
     },
     {
-      "epoch": 11.22,
+      "epoch": 11.43,
       "learning_rate": 0.0002,
-      "loss": 0.0081,
+      "loss": 0.0066,
       "step": 617
     },
     {
-      "epoch": 11.24,
+      "epoch": 11.44,
       "learning_rate": 0.0002,
-      "loss": 0.0172,
+      "loss": 0.0047,
       "step": 618
     },
     {
-      "epoch": 11.25,
+      "epoch": 11.46,
       "learning_rate": 0.0002,
-      "loss": 0.0133,
+      "loss": 0.0099,
       "step": 619
     },
     {
-      "epoch": 11.27,
+      "epoch": 11.48,
       "learning_rate": 0.0002,
-      "loss": 0.013,
+      "loss": 0.0138,
       "step": 620
     },
     {
-      "epoch": 11.29,
+      "epoch": 11.5,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0057,
       "step": 621
     },
     {
-      "epoch": 11.31,
+      "epoch": 11.52,
       "learning_rate": 0.0002,
-      "loss": 0.0097,
+      "loss": 0.005,
       "step": 622
     },
     {
-      "epoch": 11.33,
+      "epoch": 11.54,
       "learning_rate": 0.0002,
-      "loss": 0.0152,
+      "loss": 0.0058,
       "step": 623
     },
     {
-      "epoch": 11.35,
+      "epoch": 11.56,
       "learning_rate": 0.0002,
-      "loss": 0.0066,
+      "loss": 0.0079,
       "step": 624
     },
     {
-      "epoch": 11.36,
+      "epoch": 11.57,
       "learning_rate": 0.0002,
-      "loss": 0.0116,
+      "loss": 0.0056,
       "step": 625
     },
     {
-      "epoch": 11.38,
+      "epoch": 11.59,
       "learning_rate": 0.0002,
-      "loss": 0.0073,
+      "loss": 0.0136,
       "step": 626
     },
     {
-      "epoch": 11.4,
+      "epoch": 11.61,
       "learning_rate": 0.0002,
-      "loss": 0.0024,
+      "loss": 0.0098,
       "step": 627
     },
     {
-      "epoch": 11.42,
+      "epoch": 11.63,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
+      "loss": 0.005,
       "step": 628
     },
     {
-      "epoch": 11.44,
+      "epoch": 11.65,
       "learning_rate": 0.0002,
-      "loss": 0.0097,
+      "loss": 0.0091,
       "step": 629
     },
     {
-      "epoch": 11.45,
+      "epoch": 11.67,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0081,
       "step": 630
     },
     {
-      "epoch": 11.47,
+      "epoch": 11.69,
       "learning_rate": 0.0002,
-      "loss": 0.0119,
+      "loss": 0.0072,
       "step": 631
     },
     {
-      "epoch": 11.49,
+      "epoch": 11.7,
       "learning_rate": 0.0002,
-      "loss": 0.0139,
+      "loss": 0.0074,
       "step": 632
     },
     {
-      "epoch": 11.51,
+      "epoch": 11.72,
       "learning_rate": 0.0002,
-      "loss": 0.0023,
+      "loss": 0.0082,
       "step": 633
     },
     {
-      "epoch": 11.53,
+      "epoch": 11.74,
       "learning_rate": 0.0002,
-      "loss": 0.0114,
+      "loss": 0.009,
       "step": 634
     },
     {
-      "epoch": 11.55,
+      "epoch": 11.76,
       "learning_rate": 0.0002,
-      "loss": 0.0034,
+      "loss": 0.0128,
       "step": 635
     },
     {
-      "epoch": 11.56,
+      "epoch": 11.78,
       "learning_rate": 0.0002,
-      "loss": 0.0083,
+      "loss": 0.0093,
       "step": 636
     },
     {
-      "epoch": 11.58,
+      "epoch": 11.8,
       "learning_rate": 0.0002,
-      "loss": 0.0139,
+      "loss": 0.0072,
       "step": 637
     },
     {
-      "epoch": 11.6,
+      "epoch": 11.81,
       "learning_rate": 0.0002,
-      "loss": 0.0084,
+      "loss": 0.0067,
       "step": 638
     },
     {
-      "epoch": 11.62,
+      "epoch": 11.83,
       "learning_rate": 0.0002,
-      "loss": 0.0055,
+      "loss": 0.0121,
       "step": 639
     },
     {
-      "epoch": 11.64,
+      "epoch": 11.85,
       "learning_rate": 0.0002,
-      "loss": 0.0064,
+      "loss": 0.0136,
       "step": 640
     },
     {
-      "epoch": 11.65,
+      "epoch": 11.87,
       "learning_rate": 0.0002,
-      "loss": 0.0289,
+      "loss": 0.004,
       "step": 641
     },
     {
-      "epoch": 11.67,
+      "epoch": 11.89,
       "learning_rate": 0.0002,
-      "loss": 0.0093,
+      "loss": 0.0043,
       "step": 642
     },
     {
-      "epoch": 11.69,
+      "epoch": 11.91,
       "learning_rate": 0.0002,
-      "loss": 0.0113,
+      "loss": 0.0067,
       "step": 643
     },
     {
-      "epoch": 11.71,
+      "epoch": 11.93,
       "learning_rate": 0.0002,
-      "loss": 0.0114,
+      "loss": 0.0097,
       "step": 644
     },
     {
-      "epoch": 11.73,
+      "epoch": 11.94,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0089,
       "step": 645
     },
     {
-      "epoch": 11.75,
+      "epoch": 11.96,
       "learning_rate": 0.0002,
-      "loss": 0.0089,
+      "loss": 0.009,
       "step": 646
     },
     {
-      "epoch": 11.76,
+      "epoch": 11.98,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0058,
       "step": 647
     },
     {
-      "epoch": 11.78,
+      "epoch": 12.0,
       "learning_rate": 0.0002,
-      "loss": 0.0191,
+      "loss": 0.0076,
       "step": 648
     },
     {
-      "epoch": 11.8,
+      "epoch": 12.0,
+      "eval_loss": 0.006620803847908974,
+      "eval_runtime": 138.0688,
+      "eval_samples_per_second": 12.392,
+      "eval_steps_per_second": 0.391,
+      "step": 648
+    },
+    {
+      "epoch": 12.02,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0034,
       "step": 649
     },
     {
-      "epoch": 11.82,
+      "epoch": 12.04,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.0048,
       "step": 650
     },
     {
-      "epoch": 11.84,
+      "epoch": 12.06,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0054,
       "step": 651
     },
     {
-      "epoch": 11.85,
+      "epoch": 12.07,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0164,
       "step": 652
     },
     {
-      "epoch": 11.87,
+      "epoch": 12.09,
       "learning_rate": 0.0002,
-      "loss": 0.0128,
+      "loss": 0.0075,
       "step": 653
     },
     {
-      "epoch": 11.89,
+      "epoch": 12.11,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
+      "loss": 0.0056,
       "step": 654
     },
     {
-      "epoch": 11.91,
+      "epoch": 12.13,
       "learning_rate": 0.0002,
-      "loss": 0.0056,
+      "loss": 0.0069,
       "step": 655
     },
     {
-      "epoch": 11.93,
+      "epoch": 12.15,
       "learning_rate": 0.0002,
-      "loss": 0.0161,
+      "loss": 0.0046,
       "step": 656
     },
     {
-      "epoch": 11.95,
+      "epoch": 12.17,
       "learning_rate": 0.0002,
-      "loss": 0.0175,
+      "loss": 0.0063,
       "step": 657
     },
     {
-      "epoch": 11.96,
+      "epoch": 12.19,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0068,
       "step": 658
     },
     {
-      "epoch": 11.98,
+      "epoch": 12.2,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0082,
       "step": 659
     },
     {
-      "epoch": 12.0,
+      "epoch": 12.22,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
-      "step": 660
-    },
-    {
-      "epoch": 12.0,
-      "eval_loss": 0.005593516398221254,
-      "eval_runtime": 97.5106,
-      "eval_samples_per_second": 8.891,
-      "eval_steps_per_second": 0.564,
+      "loss": 0.0114,
       "step": 660
     },
     {
-      "epoch": 12.02,
+      "epoch": 12.24,
       "learning_rate": 0.0002,
-      "loss": 0.007,
+      "loss": 0.0046,
       "step": 661
     },
     {
-      "epoch": 12.04,
+      "epoch": 12.26,
       "learning_rate": 0.0002,
       "loss": 0.0082,
       "step": 662
     },
     {
-      "epoch": 12.05,
+      "epoch": 12.28,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.0046,
       "step": 663
     },
     {
-      "epoch": 12.07,
+      "epoch": 12.3,
       "learning_rate": 0.0002,
-      "loss": 0.0058,
+      "loss": 0.0081,
       "step": 664
     },
     {
-      "epoch": 12.09,
+      "epoch": 12.31,
       "learning_rate": 0.0002,
-      "loss": 0.0037,
+      "loss": 0.0115,
       "step": 665
     },
     {
-      "epoch": 12.11,
+      "epoch": 12.33,
       "learning_rate": 0.0002,
-      "loss": 0.021,
+      "loss": 0.0143,
       "step": 666
     },
     {
-      "epoch": 12.13,
+      "epoch": 12.35,
       "learning_rate": 0.0002,
-      "loss": 0.0027,
+      "loss": 0.0075,
       "step": 667
     },
     {
-      "epoch": 12.15,
+      "epoch": 12.37,
       "learning_rate": 0.0002,
-      "loss": 0.0051,
+      "loss": 0.0076,
       "step": 668
     },
     {
-      "epoch": 12.16,
+      "epoch": 12.39,
       "learning_rate": 0.0002,
-      "loss": 0.0179,
+      "loss": 0.0083,
       "step": 669
     },
     {
-      "epoch": 12.18,
+      "epoch": 12.41,
       "learning_rate": 0.0002,
-      "loss": 0.0029,
+      "loss": 0.009,
       "step": 670
     },
     {
-      "epoch": 12.2,
+      "epoch": 12.43,
       "learning_rate": 0.0002,
-      "loss": 0.0014,
+      "loss": 0.0055,
       "step": 671
     },
     {
-      "epoch": 12.22,
+      "epoch": 12.44,
       "learning_rate": 0.0002,
-      "loss": 0.008,
+      "loss": 0.0071,
       "step": 672
     },
     {
-      "epoch": 12.24,
+      "epoch": 12.46,
       "learning_rate": 0.0002,
-      "loss": 0.0039,
+      "loss": 0.007,
       "step": 673
     },
     {
-      "epoch": 12.25,
+      "epoch": 12.48,
       "learning_rate": 0.0002,
-      "loss": 0.0335,
+      "loss": 0.0115,
       "step": 674
     },
     {
-      "epoch": 12.27,
+      "epoch": 12.5,
       "learning_rate": 0.0002,
-      "loss": 0.0034,
+      "loss": 0.0062,
       "step": 675
     },
     {
-      "epoch": 12.29,
+      "epoch": 12.52,
       "learning_rate": 0.0002,
-      "loss": 0.0073,
+      "loss": 0.0089,
       "step": 676
     },
     {
-      "epoch": 12.31,
+      "epoch": 12.54,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0093,
       "step": 677
     },
     {
-      "epoch": 12.33,
+      "epoch": 12.56,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.0065,
       "step": 678
     },
     {
-      "epoch": 12.35,
+      "epoch": 12.57,
       "learning_rate": 0.0002,
-      "loss": 0.0036,
+      "loss": 0.0044,
       "step": 679
     },
     {
-      "epoch": 12.36,
+      "epoch": 12.59,
       "learning_rate": 0.0002,
-      "loss": 0.0081,
+      "loss": 0.0124,
       "step": 680
     },
     {
-      "epoch": 12.38,
+      "epoch": 12.61,
       "learning_rate": 0.0002,
-      "loss": 0.0079,
+      "loss": 0.0134,
       "step": 681
     },
     {
-      "epoch": 12.4,
+      "epoch": 12.63,
       "learning_rate": 0.0002,
-      "loss": 0.0555,
+      "loss": 0.0073,
       "step": 682
     },
     {
-      "epoch": 12.42,
+      "epoch": 12.65,
       "learning_rate": 0.0002,
-      "loss": 0.022,
+      "loss": 0.0081,
       "step": 683
     },
     {
-      "epoch": 12.44,
+      "epoch": 12.67,
       "learning_rate": 0.0002,
-      "loss": 0.0026,
+      "loss": 0.006,
       "step": 684
     },
     {
-      "epoch": 12.45,
+      "epoch": 12.69,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0135,
       "step": 685
     },
     {
-      "epoch": 12.47,
+      "epoch": 12.7,
       "learning_rate": 0.0002,
-      "loss": 0.0122,
+      "loss": 0.007,
       "step": 686
     },
     {
-      "epoch": 12.49,
+      "epoch": 12.72,
       "learning_rate": 0.0002,
-      "loss": 0.012,
+      "loss": 0.0175,
       "step": 687
     },
     {
-      "epoch": 12.51,
+      "epoch": 12.74,
       "learning_rate": 0.0002,
-      "loss": 0.0085,
+      "loss": 0.0066,
       "step": 688
     },
     {
-      "epoch": 12.53,
+      "epoch": 12.76,
       "learning_rate": 0.0002,
-      "loss": 0.0176,
+      "loss": 0.0114,
       "step": 689
     },
     {
-      "epoch": 12.55,
+      "epoch": 12.78,
       "learning_rate": 0.0002,
-      "loss": 0.0217,
+      "loss": 0.0084,
       "step": 690
     },
     {
-      "epoch": 12.56,
+      "epoch": 12.8,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.0116,
       "step": 691
     },
     {
-      "epoch": 12.58,
+      "epoch": 12.81,
       "learning_rate": 0.0002,
-      "loss": 0.0139,
+      "loss": 0.0053,
       "step": 692
     },
     {
-      "epoch": 12.6,
+      "epoch": 12.83,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0098,
       "step": 693
     },
     {
-      "epoch": 12.62,
+      "epoch": 12.85,
       "learning_rate": 0.0002,
-      "loss": 0.0098,
+      "loss": 0.0055,
       "step": 694
     },
     {
-      "epoch": 12.64,
+      "epoch": 12.87,
       "learning_rate": 0.0002,
-      "loss": 0.0124,
+      "loss": 0.007,
       "step": 695
     },
     {
-      "epoch": 12.65,
+      "epoch": 12.89,
       "learning_rate": 0.0002,
-      "loss": 0.0083,
+      "loss": 0.0109,
       "step": 696
     },
     {
-      "epoch": 12.67,
+      "epoch": 12.91,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.0142,
       "step": 697
     },
     {
-      "epoch": 12.69,
+      "epoch": 12.93,
       "learning_rate": 0.0002,
-      "loss": 0.0109,
+      "loss": 0.0049,
       "step": 698
     },
     {
-      "epoch": 12.71,
+      "epoch": 12.94,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0075,
       "step": 699
     },
     {
-      "epoch": 12.73,
+      "epoch": 12.96,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.0062,
       "step": 700
     },
     {
-      "epoch": 12.75,
+      "epoch": 12.98,
       "learning_rate": 0.0002,
-      "loss": 0.0215,
+      "loss": 0.0076,
       "step": 701
     },
     {
-      "epoch": 12.76,
+      "epoch": 13.0,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0054,
       "step": 702
     },
     {
-      "epoch": 12.78,
+      "epoch": 13.0,
+      "eval_loss": 0.005915054120123386,
+      "eval_runtime": 138.1179,
+      "eval_samples_per_second": 12.388,
+      "eval_steps_per_second": 0.391,
+      "step": 702
+    },
+    {
+      "epoch": 13.02,
       "learning_rate": 0.0002,
-      "loss": 0.016,
+      "loss": 0.0043,
       "step": 703
     },
     {
-      "epoch": 12.8,
+      "epoch": 13.04,
       "learning_rate": 0.0002,
-      "loss": 0.0207,
+      "loss": 0.0068,
       "step": 704
     },
     {
-      "epoch": 12.82,
+      "epoch": 13.06,
       "learning_rate": 0.0002,
-      "loss": 0.01,
+      "loss": 0.0047,
       "step": 705
     },
     {
-      "epoch": 12.84,
+      "epoch": 13.07,
       "learning_rate": 0.0002,
-      "loss": 0.0126,
+      "loss": 0.0054,
       "step": 706
     },
     {
-      "epoch": 12.85,
+      "epoch": 13.09,
       "learning_rate": 0.0002,
-      "loss": 0.0039,
+      "loss": 0.0055,
       "step": 707
     },
     {
-      "epoch": 12.87,
+      "epoch": 13.11,
       "learning_rate": 0.0002,
-      "loss": 0.0137,
+      "loss": 0.0048,
       "step": 708
     },
     {
-      "epoch": 12.89,
+      "epoch": 13.13,
       "learning_rate": 0.0002,
-      "loss": 0.0205,
+      "loss": 0.0065,
       "step": 709
     },
     {
-      "epoch": 12.91,
+      "epoch": 13.15,
       "learning_rate": 0.0002,
-      "loss": 0.006,
+      "loss": 0.0037,
       "step": 710
     },
     {
-      "epoch": 12.93,
+      "epoch": 13.17,
       "learning_rate": 0.0002,
-      "loss": 0.0115,
+      "loss": 0.0063,
       "step": 711
     },
     {
-      "epoch": 12.95,
+      "epoch": 13.19,
       "learning_rate": 0.0002,
-      "loss": 0.0176,
+      "loss": 0.013,
       "step": 712
     },
     {
-      "epoch": 12.96,
+      "epoch": 13.2,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
+      "loss": 0.007,
       "step": 713
     },
     {
-      "epoch": 12.98,
+      "epoch": 13.22,
       "learning_rate": 0.0002,
-      "loss": 0.0188,
+      "loss": 0.0047,
       "step": 714
     },
     {
-      "epoch": 13.0,
+      "epoch": 13.24,
       "learning_rate": 0.0002,
-      "loss": 0.0133,
-      "step": 715
-    },
-    {
-      "epoch": 13.0,
-      "eval_loss": 0.016621621325612068,
-      "eval_runtime": 97.3476,
-      "eval_samples_per_second": 8.906,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0048,
       "step": 715
     },
     {
-      "epoch": 13.02,
+      "epoch": 13.26,
       "learning_rate": 0.0002,
-      "loss": 0.0096,
+      "loss": 0.0038,
       "step": 716
     },
     {
-      "epoch": 13.04,
+      "epoch": 13.28,
       "learning_rate": 0.0002,
-      "loss": 0.025,
+      "loss": 0.0044,
       "step": 717
     },
     {
-      "epoch": 13.05,
+      "epoch": 13.3,
       "learning_rate": 0.0002,
-      "loss": 0.01,
+      "loss": 0.0052,
       "step": 718
     },
     {
-      "epoch": 13.07,
+      "epoch": 13.31,
       "learning_rate": 0.0002,
-      "loss": 0.0285,
+      "loss": 0.0097,
       "step": 719
     },
     {
-      "epoch": 13.09,
+      "epoch": 13.33,
       "learning_rate": 0.0002,
-      "loss": 0.0109,
+      "loss": 0.0142,
       "step": 720
     },
     {
-      "epoch": 13.11,
+      "epoch": 13.35,
       "learning_rate": 0.0002,
-      "loss": 0.0167,
+      "loss": 0.0125,
       "step": 721
     },
     {
-      "epoch": 13.13,
+      "epoch": 13.37,
       "learning_rate": 0.0002,
-      "loss": 0.0159,
+      "loss": 0.0053,
       "step": 722
     },
     {
-      "epoch": 13.15,
+      "epoch": 13.39,
       "learning_rate": 0.0002,
-      "loss": 0.0107,
+      "loss": 0.0081,
       "step": 723
     },
     {
-      "epoch": 13.16,
+      "epoch": 13.41,
       "learning_rate": 0.0002,
-      "loss": 0.0134,
+      "loss": 0.0092,
       "step": 724
     },
     {
-      "epoch": 13.18,
+      "epoch": 13.43,
       "learning_rate": 0.0002,
-      "loss": 0.0952,
+      "loss": 0.0131,
       "step": 725
     },
     {
-      "epoch": 13.2,
+      "epoch": 13.44,
       "learning_rate": 0.0002,
-      "loss": 0.0185,
+      "loss": 0.0061,
       "step": 726
     },
     {
-      "epoch": 13.22,
+      "epoch": 13.46,
       "learning_rate": 0.0002,
-      "loss": 0.0115,
+      "loss": 0.0054,
       "step": 727
     },
     {
-      "epoch": 13.24,
+      "epoch": 13.48,
       "learning_rate": 0.0002,
-      "loss": 0.0205,
+      "loss": 0.0077,
       "step": 728
     },
     {
-      "epoch": 13.25,
+      "epoch": 13.5,
       "learning_rate": 0.0002,
-      "loss": 0.0095,
+      "loss": 0.0114,
       "step": 729
     },
     {
-      "epoch": 13.27,
+      "epoch": 13.52,
       "learning_rate": 0.0002,
-      "loss": 0.0145,
+      "loss": 0.0039,
       "step": 730
     },
     {
-      "epoch": 13.29,
+      "epoch": 13.54,
       "learning_rate": 0.0002,
-      "loss": 0.0162,
+      "loss": 0.009,
       "step": 731
     },
     {
-      "epoch": 13.31,
+      "epoch": 13.56,
       "learning_rate": 0.0002,
-      "loss": 0.0111,
+      "loss": 0.0074,
       "step": 732
     },
     {
-      "epoch": 13.33,
+      "epoch": 13.57,
       "learning_rate": 0.0002,
-      "loss": 0.0212,
+      "loss": 0.0033,
       "step": 733
     },
     {
-      "epoch": 13.35,
+      "epoch": 13.59,
       "learning_rate": 0.0002,
-      "loss": 0.0363,
+      "loss": 0.0061,
       "step": 734
     },
     {
-      "epoch": 13.36,
+      "epoch": 13.61,
       "learning_rate": 0.0002,
-      "loss": 0.0346,
+      "loss": 0.0079,
       "step": 735
     },
     {
-      "epoch": 13.38,
+      "epoch": 13.63,
       "learning_rate": 0.0002,
-      "loss": 0.0379,
+      "loss": 0.0087,
       "step": 736
     },
     {
-      "epoch": 13.4,
+      "epoch": 13.65,
       "learning_rate": 0.0002,
-      "loss": 0.0169,
+      "loss": 0.0113,
       "step": 737
     },
     {
-      "epoch": 13.42,
+      "epoch": 13.67,
       "learning_rate": 0.0002,
-      "loss": 0.0247,
+      "loss": 0.0075,
       "step": 738
     },
     {
-      "epoch": 13.44,
+      "epoch": 13.69,
       "learning_rate": 0.0002,
-      "loss": 0.0314,
+      "loss": 0.0056,
       "step": 739
     },
     {
-      "epoch": 13.45,
+      "epoch": 13.7,
       "learning_rate": 0.0002,
-      "loss": 0.0718,
+      "loss": 0.007,
       "step": 740
     },
     {
-      "epoch": 13.47,
+      "epoch": 13.72,
       "learning_rate": 0.0002,
-      "loss": 0.0277,
+      "loss": 0.01,
       "step": 741
     },
     {
-      "epoch": 13.49,
+      "epoch": 13.74,
       "learning_rate": 0.0002,
-      "loss": 0.0183,
+      "loss": 0.0039,
       "step": 742
     },
     {
-      "epoch": 13.51,
+      "epoch": 13.76,
       "learning_rate": 0.0002,
-      "loss": 0.0181,
+      "loss": 0.0095,
       "step": 743
     },
     {
-      "epoch": 13.53,
+      "epoch": 13.78,
       "learning_rate": 0.0002,
-      "loss": 0.0177,
+      "loss": 0.0075,
       "step": 744
     },
     {
-      "epoch": 13.55,
+      "epoch": 13.8,
       "learning_rate": 0.0002,
-      "loss": 0.0478,
+      "loss": 0.0069,
       "step": 745
     },
     {
-      "epoch": 13.56,
+      "epoch": 13.81,
       "learning_rate": 0.0002,
-      "loss": 0.0219,
+      "loss": 0.0085,
       "step": 746
     },
     {
-      "epoch": 13.58,
+      "epoch": 13.83,
       "learning_rate": 0.0002,
-      "loss": 0.0309,
+      "loss": 0.0066,
       "step": 747
     },
     {
-      "epoch": 13.6,
+      "epoch": 13.85,
       "learning_rate": 0.0002,
-      "loss": 0.0185,
+      "loss": 0.0095,
       "step": 748
     },
     {
-      "epoch": 13.62,
+      "epoch": 13.87,
       "learning_rate": 0.0002,
-      "loss": 0.0139,
+      "loss": 0.0124,
       "step": 749
     },
     {
-      "epoch": 13.64,
+      "epoch": 13.89,
       "learning_rate": 0.0002,
-      "loss": 0.0274,
+      "loss": 0.0053,
       "step": 750
     },
     {
-      "epoch": 13.65,
+      "epoch": 13.91,
       "learning_rate": 0.0002,
-      "loss": 0.0167,
+      "loss": 0.007,
       "step": 751
     },
     {
-      "epoch": 13.67,
+      "epoch": 13.93,
       "learning_rate": 0.0002,
-      "loss": 0.0283,
+      "loss": 0.0062,
       "step": 752
     },
     {
-      "epoch": 13.69,
+      "epoch": 13.94,
       "learning_rate": 0.0002,
-      "loss": 0.017,
+      "loss": 0.0074,
       "step": 753
     },
     {
-      "epoch": 13.71,
+      "epoch": 13.96,
       "learning_rate": 0.0002,
-      "loss": 0.0236,
+      "loss": 0.0095,
       "step": 754
     },
     {
-      "epoch": 13.73,
+      "epoch": 13.98,
       "learning_rate": 0.0002,
-      "loss": 0.0207,
+      "loss": 0.0061,
       "step": 755
     },
     {
-      "epoch": 13.75,
+      "epoch": 14.0,
       "learning_rate": 0.0002,
-      "loss": 0.0385,
+      "loss": 0.0108,
       "step": 756
     },
     {
-      "epoch": 13.76,
+      "epoch": 14.0,
+      "eval_loss": 0.006522215902805328,
+      "eval_runtime": 138.2196,
+      "eval_samples_per_second": 12.379,
+      "eval_steps_per_second": 0.391,
+      "step": 756
+    },
+    {
+      "epoch": 14.02,
       "learning_rate": 0.0002,
-      "loss": 0.0311,
+      "loss": 0.0036,
       "step": 757
     },
     {
-      "epoch": 13.78,
+      "epoch": 14.04,
       "learning_rate": 0.0002,
-      "loss": 0.0332,
+      "loss": 0.005,
       "step": 758
     },
     {
-      "epoch": 13.8,
+      "epoch": 14.06,
       "learning_rate": 0.0002,
-      "loss": 0.0392,
+      "loss": 0.0114,
       "step": 759
     },
     {
-      "epoch": 13.82,
+      "epoch": 14.07,
       "learning_rate": 0.0002,
-      "loss": 0.0386,
+      "loss": 0.0062,
       "step": 760
     },
     {
-      "epoch": 13.84,
+      "epoch": 14.09,
       "learning_rate": 0.0002,
-      "loss": 0.018,
+      "loss": 0.006,
       "step": 761
     },
     {
-      "epoch": 13.85,
+      "epoch": 14.11,
       "learning_rate": 0.0002,
-      "loss": 0.0224,
+      "loss": 0.0071,
       "step": 762
     },
     {
-      "epoch": 13.87,
+      "epoch": 14.13,
       "learning_rate": 0.0002,
-      "loss": 0.046,
+      "loss": 0.0065,
       "step": 763
     },
     {
-      "epoch": 13.89,
+      "epoch": 14.15,
       "learning_rate": 0.0002,
-      "loss": 0.0415,
+      "loss": 0.0106,
       "step": 764
     },
     {
-      "epoch": 13.91,
+      "epoch": 14.17,
       "learning_rate": 0.0002,
-      "loss": 0.0429,
+      "loss": 0.0124,
       "step": 765
     },
     {
-      "epoch": 13.93,
+      "epoch": 14.19,
       "learning_rate": 0.0002,
-      "loss": 0.0279,
+      "loss": 0.0068,
       "step": 766
     },
     {
-      "epoch": 13.95,
+      "epoch": 14.2,
       "learning_rate": 0.0002,
-      "loss": 0.0219,
+      "loss": 0.0076,
       "step": 767
     },
     {
-      "epoch": 13.96,
+      "epoch": 14.22,
       "learning_rate": 0.0002,
-      "loss": 0.0225,
+      "loss": 0.0073,
       "step": 768
     },
     {
-      "epoch": 13.98,
+      "epoch": 14.24,
       "learning_rate": 0.0002,
-      "loss": 0.0361,
+      "loss": 0.0131,
       "step": 769
     },
     {
-      "epoch": 14.0,
+      "epoch": 14.26,
       "learning_rate": 0.0002,
-      "loss": 0.0231,
-      "step": 770
-    },
-    {
-      "epoch": 14.0,
-      "eval_loss": 0.02342851273715496,
-      "eval_runtime": 97.3497,
-      "eval_samples_per_second": 8.906,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0078,
       "step": 770
     },
     {
-      "epoch": 14.02,
+      "epoch": 14.28,
       "learning_rate": 0.0002,
-      "loss": 0.0142,
+      "loss": 0.0094,
       "step": 771
     },
     {
-      "epoch": 14.04,
+      "epoch": 14.3,
       "learning_rate": 0.0002,
-      "loss": 0.0162,
+      "loss": 0.0093,
       "step": 772
     },
     {
-      "epoch": 14.05,
+      "epoch": 14.31,
       "learning_rate": 0.0002,
-      "loss": 0.0422,
+      "loss": 0.0032,
       "step": 773
     },
     {
-      "epoch": 14.07,
+      "epoch": 14.33,
       "learning_rate": 0.0002,
-      "loss": 0.0495,
+      "loss": 0.0096,
       "step": 774
     },
     {
-      "epoch": 14.09,
+      "epoch": 14.35,
       "learning_rate": 0.0002,
-      "loss": 0.0556,
+      "loss": 0.0079,
       "step": 775
     },
     {
-      "epoch": 14.11,
+      "epoch": 14.37,
       "learning_rate": 0.0002,
-      "loss": 0.0392,
+      "loss": 0.0049,
       "step": 776
     },
     {
-      "epoch": 14.13,
+      "epoch": 14.39,
       "learning_rate": 0.0002,
-      "loss": 0.0427,
+      "loss": 0.0049,
       "step": 777
     },
     {
-      "epoch": 14.15,
+      "epoch": 14.41,
       "learning_rate": 0.0002,
-      "loss": 0.0225,
+      "loss": 0.0045,
       "step": 778
     },
     {
-      "epoch": 14.16,
+      "epoch": 14.43,
       "learning_rate": 0.0002,
-      "loss": 0.0244,
+      "loss": 0.0056,
       "step": 779
     },
     {
-      "epoch": 14.18,
+      "epoch": 14.44,
       "learning_rate": 0.0002,
-      "loss": 0.0144,
+      "loss": 0.0084,
       "step": 780
     },
     {
-      "epoch": 14.2,
+      "epoch": 14.46,
       "learning_rate": 0.0002,
-      "loss": 0.0224,
+      "loss": 0.006,
       "step": 781
     },
     {
-      "epoch": 14.22,
+      "epoch": 14.48,
       "learning_rate": 0.0002,
-      "loss": 0.0284,
+      "loss": 0.0146,
       "step": 782
     },
     {
-      "epoch": 14.24,
+      "epoch": 14.5,
       "learning_rate": 0.0002,
-      "loss": 0.0183,
+      "loss": 0.0046,
       "step": 783
     },
     {
-      "epoch": 14.25,
+      "epoch": 14.52,
       "learning_rate": 0.0002,
-      "loss": 0.0162,
+      "loss": 0.0065,
       "step": 784
     },
     {
-      "epoch": 14.27,
+      "epoch": 14.54,
       "learning_rate": 0.0002,
-      "loss": 0.0158,
+      "loss": 0.0095,
       "step": 785
     },
     {
-      "epoch": 14.29,
+      "epoch": 14.56,
       "learning_rate": 0.0002,
-      "loss": 0.0239,
+      "loss": 0.0089,
       "step": 786
     },
     {
-      "epoch": 14.31,
+      "epoch": 14.57,
       "learning_rate": 0.0002,
-      "loss": 0.0124,
+      "loss": 0.0084,
       "step": 787
     },
     {
-      "epoch": 14.33,
+      "epoch": 14.59,
       "learning_rate": 0.0002,
-      "loss": 0.031,
+      "loss": 0.0074,
       "step": 788
     },
     {
-      "epoch": 14.35,
+      "epoch": 14.61,
       "learning_rate": 0.0002,
-      "loss": 0.0164,
+      "loss": 0.0101,
       "step": 789
     },
     {
-      "epoch": 14.36,
+      "epoch": 14.63,
       "learning_rate": 0.0002,
-      "loss": 0.0433,
+      "loss": 0.0057,
       "step": 790
     },
     {
-      "epoch": 14.38,
+      "epoch": 14.65,
       "learning_rate": 0.0002,
-      "loss": 0.0194,
+      "loss": 0.01,
       "step": 791
     },
     {
-      "epoch": 14.4,
+      "epoch": 14.67,
       "learning_rate": 0.0002,
-      "loss": 0.0179,
+      "loss": 0.0072,
       "step": 792
     },
     {
-      "epoch": 14.42,
+      "epoch": 14.69,
       "learning_rate": 0.0002,
-      "loss": 0.0302,
+      "loss": 0.0108,
       "step": 793
     },
     {
-      "epoch": 14.44,
+      "epoch": 14.7,
       "learning_rate": 0.0002,
-      "loss": 0.0181,
+      "loss": 0.0106,
       "step": 794
     },
     {
-      "epoch": 14.45,
+      "epoch": 14.72,
       "learning_rate": 0.0002,
-      "loss": 0.0204,
+      "loss": 0.0099,
       "step": 795
     },
     {
-      "epoch": 14.47,
+      "epoch": 14.74,
       "learning_rate": 0.0002,
-      "loss": 0.031,
+      "loss": 0.0115,
       "step": 796
     },
     {
-      "epoch": 14.49,
+      "epoch": 14.76,
       "learning_rate": 0.0002,
-      "loss": 0.0441,
+      "loss": 0.008,
       "step": 797
     },
     {
-      "epoch": 14.51,
+      "epoch": 14.78,
       "learning_rate": 0.0002,
-      "loss": 0.0296,
+      "loss": 0.0151,
       "step": 798
     },
     {
-      "epoch": 14.53,
+      "epoch": 14.8,
       "learning_rate": 0.0002,
-      "loss": 0.0213,
+      "loss": 0.007,
       "step": 799
     },
     {
-      "epoch": 14.55,
+      "epoch": 14.81,
       "learning_rate": 0.0002,
-      "loss": 0.0363,
+      "loss": 0.0106,
       "step": 800
     },
     {
-      "epoch": 14.56,
+      "epoch": 14.83,
       "learning_rate": 0.0002,
-      "loss": 0.0261,
+      "loss": 0.0122,
       "step": 801
     },
     {
-      "epoch": 14.58,
+      "epoch": 14.85,
       "learning_rate": 0.0002,
-      "loss": 0.0314,
+      "loss": 0.0081,
       "step": 802
     },
     {
-      "epoch": 14.6,
+      "epoch": 14.87,
       "learning_rate": 0.0002,
-      "loss": 0.0196,
+      "loss": 0.0098,
       "step": 803
     },
     {
-      "epoch": 14.62,
+      "epoch": 14.89,
       "learning_rate": 0.0002,
-      "loss": 0.034,
+      "loss": 0.0074,
       "step": 804
     },
     {
-      "epoch": 14.64,
+      "epoch": 14.91,
       "learning_rate": 0.0002,
-      "loss": 0.0277,
+      "loss": 0.0073,
       "step": 805
     },
     {
-      "epoch": 14.65,
+      "epoch": 14.93,
       "learning_rate": 0.0002,
-      "loss": 0.026,
+      "loss": 0.0057,
       "step": 806
     },
     {
-      "epoch": 14.67,
+      "epoch": 14.94,
       "learning_rate": 0.0002,
-      "loss": 0.0386,
+      "loss": 0.0065,
       "step": 807
     },
     {
-      "epoch": 14.69,
+      "epoch": 14.96,
       "learning_rate": 0.0002,
-      "loss": 0.0247,
+      "loss": 0.0093,
       "step": 808
     },
     {
-      "epoch": 14.71,
+      "epoch": 14.98,
       "learning_rate": 0.0002,
-      "loss": 0.0305,
+      "loss": 0.0087,
       "step": 809
     },
     {
-      "epoch": 14.73,
+      "epoch": 15.0,
       "learning_rate": 0.0002,
-      "loss": 0.0199,
+      "loss": 0.0076,
       "step": 810
     },
     {
-      "epoch": 14.75,
+      "epoch": 15.0,
+      "eval_loss": 0.005220985040068626,
+      "eval_runtime": 138.2674,
+      "eval_samples_per_second": 12.375,
+      "eval_steps_per_second": 0.391,
+      "step": 810
+    },
+    {
+      "epoch": 15.02,
       "learning_rate": 0.0002,
-      "loss": 0.0333,
+      "loss": 0.0059,
       "step": 811
     },
     {
-      "epoch": 14.76,
+      "epoch": 15.04,
       "learning_rate": 0.0002,
-      "loss": 0.0416,
+      "loss": 0.0075,
       "step": 812
     },
     {
-      "epoch": 14.78,
+      "epoch": 15.06,
       "learning_rate": 0.0002,
-      "loss": 0.0394,
+      "loss": 0.0029,
       "step": 813
     },
     {
-      "epoch": 14.8,
+      "epoch": 15.07,
       "learning_rate": 0.0002,
-      "loss": 0.0874,
+      "loss": 0.0031,
       "step": 814
     },
     {
-      "epoch": 14.82,
+      "epoch": 15.09,
       "learning_rate": 0.0002,
-      "loss": 0.033,
+      "loss": 0.008,
       "step": 815
     },
     {
-      "epoch": 14.84,
+      "epoch": 15.11,
       "learning_rate": 0.0002,
-      "loss": 0.0401,
+      "loss": 0.0046,
       "step": 816
     },
     {
-      "epoch": 14.85,
+      "epoch": 15.13,
       "learning_rate": 0.0002,
-      "loss": 0.0202,
+      "loss": 0.0042,
       "step": 817
     },
     {
-      "epoch": 14.87,
+      "epoch": 15.15,
       "learning_rate": 0.0002,
-      "loss": 0.0366,
+      "loss": 0.0113,
       "step": 818
     },
     {
-      "epoch": 14.89,
+      "epoch": 15.17,
       "learning_rate": 0.0002,
-      "loss": 0.024,
+      "loss": 0.0077,
       "step": 819
     },
     {
-      "epoch": 14.91,
+      "epoch": 15.19,
       "learning_rate": 0.0002,
-      "loss": 0.0241,
+      "loss": 0.0064,
       "step": 820
     },
     {
-      "epoch": 14.93,
+      "epoch": 15.2,
       "learning_rate": 0.0002,
-      "loss": 0.0455,
+      "loss": 0.0053,
       "step": 821
     },
     {
-      "epoch": 14.95,
+      "epoch": 15.22,
       "learning_rate": 0.0002,
-      "loss": 0.0399,
+      "loss": 0.003,
       "step": 822
     },
     {
-      "epoch": 14.96,
+      "epoch": 15.24,
       "learning_rate": 0.0002,
-      "loss": 0.0209,
+      "loss": 0.0096,
       "step": 823
     },
     {
-      "epoch": 14.98,
+      "epoch": 15.26,
       "learning_rate": 0.0002,
-      "loss": 0.0352,
+      "loss": 0.0067,
       "step": 824
     },
     {
-      "epoch": 15.0,
+      "epoch": 15.28,
       "learning_rate": 0.0002,
-      "loss": 0.0483,
-      "step": 825
-    },
-    {
-      "epoch": 15.0,
-      "eval_loss": 0.031725089997053146,
-      "eval_runtime": 97.4788,
-      "eval_samples_per_second": 8.894,
-      "eval_steps_per_second": 0.564,
+      "loss": 0.0058,
       "step": 825
     },
     {
-      "epoch": 15.02,
+      "epoch": 15.3,
       "learning_rate": 0.0002,
-      "loss": 0.0219,
+      "loss": 0.0068,
       "step": 826
     },
     {
-      "epoch": 15.04,
+      "epoch": 15.31,
       "learning_rate": 0.0002,
-      "loss": 0.0393,
+      "loss": 0.0054,
       "step": 827
     },
     {
-      "epoch": 15.05,
+      "epoch": 15.33,
       "learning_rate": 0.0002,
-      "loss": 0.0225,
+      "loss": 0.005,
       "step": 828
     },
     {
-      "epoch": 15.07,
+      "epoch": 15.35,
       "learning_rate": 0.0002,
-      "loss": 0.029,
+      "loss": 0.0083,
       "step": 829
     },
     {
-      "epoch": 15.09,
+      "epoch": 15.37,
       "learning_rate": 0.0002,
-      "loss": 0.0331,
+      "loss": 0.0054,
       "step": 830
     },
     {
-      "epoch": 15.11,
+      "epoch": 15.39,
       "learning_rate": 0.0002,
-      "loss": 0.0398,
+      "loss": 0.0086,
       "step": 831
     },
     {
-      "epoch": 15.13,
+      "epoch": 15.41,
       "learning_rate": 0.0002,
-      "loss": 0.0312,
+      "loss": 0.0088,
       "step": 832
     },
     {
-      "epoch": 15.15,
+      "epoch": 15.43,
       "learning_rate": 0.0002,
-      "loss": 0.0169,
+      "loss": 0.0073,
       "step": 833
     },
     {
-      "epoch": 15.16,
+      "epoch": 15.44,
       "learning_rate": 0.0002,
-      "loss": 0.0259,
+      "loss": 0.0077,
       "step": 834
     },
     {
-      "epoch": 15.18,
+      "epoch": 15.46,
       "learning_rate": 0.0002,
-      "loss": 0.0309,
+      "loss": 0.009,
       "step": 835
     },
     {
-      "epoch": 15.2,
+      "epoch": 15.48,
       "learning_rate": 0.0002,
-      "loss": 0.0296,
+      "loss": 0.0058,
       "step": 836
     },
     {
-      "epoch": 15.22,
+      "epoch": 15.5,
       "learning_rate": 0.0002,
-      "loss": 0.0476,
+      "loss": 0.0063,
       "step": 837
     },
     {
-      "epoch": 15.24,
+      "epoch": 15.52,
       "learning_rate": 0.0002,
-      "loss": 0.0245,
+      "loss": 0.0131,
       "step": 838
     },
     {
-      "epoch": 15.25,
+      "epoch": 15.54,
       "learning_rate": 0.0002,
-      "loss": 0.0266,
+      "loss": 0.0081,
       "step": 839
     },
     {
-      "epoch": 15.27,
+      "epoch": 15.56,
       "learning_rate": 0.0002,
-      "loss": 0.0202,
+      "loss": 0.0072,
       "step": 840
     },
     {
-      "epoch": 15.29,
+      "epoch": 15.57,
       "learning_rate": 0.0002,
-      "loss": 0.029,
+      "loss": 0.0062,
       "step": 841
     },
     {
-      "epoch": 15.31,
+      "epoch": 15.59,
       "learning_rate": 0.0002,
-      "loss": 0.0174,
+      "loss": 0.0093,
       "step": 842
     },
     {
-      "epoch": 15.33,
+      "epoch": 15.61,
       "learning_rate": 0.0002,
-      "loss": 0.0485,
+      "loss": 0.0038,
       "step": 843
     },
     {
-      "epoch": 15.35,
+      "epoch": 15.63,
       "learning_rate": 0.0002,
-      "loss": 0.0292,
+      "loss": 0.0053,
       "step": 844
     },
     {
-      "epoch": 15.36,
+      "epoch": 15.65,
       "learning_rate": 0.0002,
-      "loss": 0.027,
+      "loss": 0.0066,
       "step": 845
     },
     {
-      "epoch": 15.38,
+      "epoch": 15.67,
       "learning_rate": 0.0002,
-      "loss": 0.0134,
+      "loss": 0.0064,
       "step": 846
     },
     {
-      "epoch": 15.4,
+      "epoch": 15.69,
       "learning_rate": 0.0002,
-      "loss": 0.0261,
+      "loss": 0.0143,
       "step": 847
     },
     {
-      "epoch": 15.42,
+      "epoch": 15.7,
       "learning_rate": 0.0002,
-      "loss": 0.0156,
+      "loss": 0.01,
       "step": 848
     },
     {
-      "epoch": 15.44,
+      "epoch": 15.72,
       "learning_rate": 0.0002,
-      "loss": 0.0224,
+      "loss": 0.0069,
       "step": 849
     },
     {
-      "epoch": 15.45,
+      "epoch": 15.74,
       "learning_rate": 0.0002,
-      "loss": 0.0264,
+      "loss": 0.0078,
       "step": 850
     },
     {
-      "epoch": 15.47,
+      "epoch": 15.76,
       "learning_rate": 0.0002,
-      "loss": 0.0184,
+      "loss": 0.0142,
       "step": 851
     },
     {
-      "epoch": 15.49,
+      "epoch": 15.78,
       "learning_rate": 0.0002,
-      "loss": 0.011,
+      "loss": 0.0146,
       "step": 852
     },
     {
-      "epoch": 15.51,
+      "epoch": 15.8,
       "learning_rate": 0.0002,
-      "loss": 0.0221,
+      "loss": 0.0108,
       "step": 853
     },
     {
-      "epoch": 15.53,
+      "epoch": 15.81,
       "learning_rate": 0.0002,
-      "loss": 0.0224,
+      "loss": 0.0062,
       "step": 854
     },
     {
-      "epoch": 15.55,
+      "epoch": 15.83,
       "learning_rate": 0.0002,
-      "loss": 0.0128,
+      "loss": 0.0136,
       "step": 855
     },
     {
-      "epoch": 15.56,
+      "epoch": 15.85,
       "learning_rate": 0.0002,
-      "loss": 0.0199,
+      "loss": 0.008,
       "step": 856
     },
     {
-      "epoch": 15.58,
+      "epoch": 15.87,
       "learning_rate": 0.0002,
-      "loss": 0.018,
+      "loss": 0.0116,
       "step": 857
     },
     {
-      "epoch": 15.6,
+      "epoch": 15.89,
       "learning_rate": 0.0002,
-      "loss": 0.026,
+      "loss": 0.0127,
       "step": 858
     },
     {
-      "epoch": 15.62,
+      "epoch": 15.91,
       "learning_rate": 0.0002,
-      "loss": 0.0216,
+      "loss": 0.0082,
       "step": 859
     },
     {
-      "epoch": 15.64,
+      "epoch": 15.93,
       "learning_rate": 0.0002,
-      "loss": 0.0052,
+      "loss": 0.0079,
       "step": 860
     },
     {
-      "epoch": 15.65,
+      "epoch": 15.94,
       "learning_rate": 0.0002,
-      "loss": 0.0215,
+      "loss": 0.0102,
       "step": 861
     },
     {
-      "epoch": 15.67,
+      "epoch": 15.96,
       "learning_rate": 0.0002,
-      "loss": 0.0118,
+      "loss": 0.0082,
       "step": 862
     },
     {
-      "epoch": 15.69,
+      "epoch": 15.98,
       "learning_rate": 0.0002,
-      "loss": 0.0166,
+      "loss": 0.0068,
       "step": 863
     },
     {
-      "epoch": 15.71,
+      "epoch": 16.0,
       "learning_rate": 0.0002,
-      "loss": 0.0157,
+      "loss": 0.0069,
+      "step": 864
+    },
+    {
+      "epoch": 16.0,
+      "eval_loss": 0.006789057515561581,
+      "eval_runtime": 138.1348,
+      "eval_samples_per_second": 12.386,
+      "eval_steps_per_second": 0.391,
       "step": 864
     },
     {
-      "epoch": 15.73,
+      "epoch": 16.02,
       "learning_rate": 0.0002,
-      "loss": 0.0244,
+      "loss": 0.0072,
       "step": 865
     },
     {
-      "epoch": 15.75,
+      "epoch": 16.04,
       "learning_rate": 0.0002,
-      "loss": 0.0098,
+      "loss": 0.0065,
       "step": 866
     },
     {
-      "epoch": 15.76,
+      "epoch": 16.06,
       "learning_rate": 0.0002,
-      "loss": 0.0171,
+      "loss": 0.0064,
       "step": 867
     },
     {
-      "epoch": 15.78,
+      "epoch": 16.07,
       "learning_rate": 0.0002,
-      "loss": 0.0156,
+      "loss": 0.0111,
       "step": 868
     },
     {
-      "epoch": 15.8,
+      "epoch": 16.09,
       "learning_rate": 0.0002,
-      "loss": 0.0142,
+      "loss": 0.0091,
       "step": 869
     },
     {
-      "epoch": 15.82,
+      "epoch": 16.11,
       "learning_rate": 0.0002,
-      "loss": 0.0144,
+      "loss": 0.0093,
       "step": 870
     },
     {
-      "epoch": 15.84,
+      "epoch": 16.13,
       "learning_rate": 0.0002,
-      "loss": 0.0134,
+      "loss": 0.0081,
       "step": 871
     },
     {
-      "epoch": 15.85,
+      "epoch": 16.15,
       "learning_rate": 0.0002,
-      "loss": 0.0252,
+      "loss": 0.0058,
       "step": 872
     },
     {
-      "epoch": 15.87,
+      "epoch": 16.17,
       "learning_rate": 0.0002,
-      "loss": 0.0293,
+      "loss": 0.0096,
       "step": 873
     },
     {
-      "epoch": 15.89,
+      "epoch": 16.19,
       "learning_rate": 0.0002,
-      "loss": 0.0324,
+      "loss": 0.0063,
       "step": 874
     },
     {
-      "epoch": 15.91,
+      "epoch": 16.2,
       "learning_rate": 0.0002,
-      "loss": 0.0112,
+      "loss": 0.0034,
       "step": 875
     },
     {
-      "epoch": 15.93,
+      "epoch": 16.22,
       "learning_rate": 0.0002,
-      "loss": 0.0285,
+      "loss": 0.0125,
       "step": 876
     },
     {
-      "epoch": 15.95,
+      "epoch": 16.24,
       "learning_rate": 0.0002,
-      "loss": 0.036,
+      "loss": 0.0082,
       "step": 877
     },
     {
-      "epoch": 15.96,
+      "epoch": 16.26,
       "learning_rate": 0.0002,
-      "loss": 0.0246,
+      "loss": 0.009,
       "step": 878
     },
     {
-      "epoch": 15.98,
+      "epoch": 16.28,
       "learning_rate": 0.0002,
-      "loss": 0.0143,
+      "loss": 0.0043,
       "step": 879
     },
     {
-      "epoch": 16.0,
+      "epoch": 16.3,
       "learning_rate": 0.0002,
-      "loss": 0.0094,
-      "step": 880
-    },
-    {
-      "epoch": 16.0,
-      "eval_loss": 0.014622457325458527,
-      "eval_runtime": 97.3077,
-      "eval_samples_per_second": 8.91,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0073,
       "step": 880
     },
     {
-      "epoch": 16.02,
+      "epoch": 16.31,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0101,
       "step": 881
     },
     {
-      "epoch": 16.04,
+      "epoch": 16.33,
       "learning_rate": 0.0002,
-      "loss": 0.0142,
+      "loss": 0.005,
       "step": 882
     },
     {
-      "epoch": 16.05,
+      "epoch": 16.35,
       "learning_rate": 0.0002,
-      "loss": 0.0118,
+      "loss": 0.0062,
       "step": 883
     },
     {
-      "epoch": 16.07,
+      "epoch": 16.37,
       "learning_rate": 0.0002,
-      "loss": 0.0183,
+      "loss": 0.008,
       "step": 884
     },
     {
-      "epoch": 16.09,
+      "epoch": 16.39,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0034,
       "step": 885
     },
     {
-      "epoch": 16.11,
+      "epoch": 16.41,
       "learning_rate": 0.0002,
-      "loss": 0.0134,
+      "loss": 0.009,
       "step": 886
     },
     {
-      "epoch": 16.13,
+      "epoch": 16.43,
       "learning_rate": 0.0002,
-      "loss": 0.0199,
+      "loss": 0.0055,
       "step": 887
     },
     {
-      "epoch": 16.15,
+      "epoch": 16.44,
       "learning_rate": 0.0002,
-      "loss": 0.0106,
+      "loss": 0.0102,
       "step": 888
     },
     {
-      "epoch": 16.16,
+      "epoch": 16.46,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.0084,
       "step": 889
     },
     {
-      "epoch": 16.18,
+      "epoch": 16.48,
       "learning_rate": 0.0002,
-      "loss": 0.0155,
+      "loss": 0.0055,
       "step": 890
     },
     {
-      "epoch": 16.2,
+      "epoch": 16.5,
       "learning_rate": 0.0002,
-      "loss": 0.0138,
+      "loss": 0.0109,
       "step": 891
     },
     {
-      "epoch": 16.22,
+      "epoch": 16.52,
       "learning_rate": 0.0002,
-      "loss": 0.0123,
+      "loss": 0.0061,
       "step": 892
     },
     {
-      "epoch": 16.24,
+      "epoch": 16.54,
       "learning_rate": 0.0002,
-      "loss": 0.0144,
+      "loss": 0.0101,
       "step": 893
     },
     {
-      "epoch": 16.25,
+      "epoch": 16.56,
       "learning_rate": 0.0002,
-      "loss": 0.0088,
+      "loss": 0.0125,
       "step": 894
     },
     {
-      "epoch": 16.27,
+      "epoch": 16.57,
       "learning_rate": 0.0002,
-      "loss": 0.0312,
+      "loss": 0.0069,
       "step": 895
     },
     {
-      "epoch": 16.29,
+      "epoch": 16.59,
       "learning_rate": 0.0002,
-      "loss": 0.0058,
+      "loss": 0.0111,
       "step": 896
     },
     {
-      "epoch": 16.31,
+      "epoch": 16.61,
       "learning_rate": 0.0002,
-      "loss": 0.0137,
+      "loss": 0.0026,
       "step": 897
     },
     {
-      "epoch": 16.33,
+      "epoch": 16.63,
       "learning_rate": 0.0002,
-      "loss": 0.0147,
+      "loss": 0.0085,
       "step": 898
     },
     {
-      "epoch": 16.35,
+      "epoch": 16.65,
       "learning_rate": 0.0002,
-      "loss": 0.0291,
+      "loss": 0.0065,
       "step": 899
     },
     {
-      "epoch": 16.36,
+      "epoch": 16.67,
       "learning_rate": 0.0002,
-      "loss": 0.0145,
+      "loss": 0.005,
       "step": 900
     },
     {
-      "epoch": 16.38,
+      "epoch": 16.69,
       "learning_rate": 0.0002,
-      "loss": 0.0163,
+      "loss": 0.0077,
       "step": 901
     },
     {
-      "epoch": 16.4,
+      "epoch": 16.7,
       "learning_rate": 0.0002,
-      "loss": 0.0136,
+      "loss": 0.0135,
       "step": 902
     },
     {
-      "epoch": 16.42,
+      "epoch": 16.72,
       "learning_rate": 0.0002,
-      "loss": 0.0187,
+      "loss": 0.0119,
       "step": 903
     },
     {
-      "epoch": 16.44,
+      "epoch": 16.74,
       "learning_rate": 0.0002,
-      "loss": 0.0057,
+      "loss": 0.0073,
       "step": 904
     },
     {
-      "epoch": 16.45,
+      "epoch": 16.76,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0069,
       "step": 905
     },
     {
-      "epoch": 16.47,
+      "epoch": 16.78,
       "learning_rate": 0.0002,
-      "loss": 0.0106,
+      "loss": 0.0075,
       "step": 906
     },
     {
-      "epoch": 16.49,
+      "epoch": 16.8,
       "learning_rate": 0.0002,
-      "loss": 0.0058,
+      "loss": 0.0111,
       "step": 907
     },
     {
-      "epoch": 16.51,
+      "epoch": 16.81,
       "learning_rate": 0.0002,
-      "loss": 0.0138,
+      "loss": 0.0126,
       "step": 908
     },
     {
-      "epoch": 16.53,
+      "epoch": 16.83,
       "learning_rate": 0.0002,
-      "loss": 0.0228,
+      "loss": 0.0072,
       "step": 909
     },
     {
-      "epoch": 16.55,
+      "epoch": 16.85,
       "learning_rate": 0.0002,
-      "loss": 0.0127,
+      "loss": 0.0036,
       "step": 910
     },
     {
-      "epoch": 16.56,
+      "epoch": 16.87,
       "learning_rate": 0.0002,
-      "loss": 0.0183,
+      "loss": 0.0048,
       "step": 911
     },
     {
-      "epoch": 16.58,
+      "epoch": 16.89,
       "learning_rate": 0.0002,
-      "loss": 0.0086,
+      "loss": 0.0097,
       "step": 912
     },
     {
-      "epoch": 16.6,
+      "epoch": 16.91,
       "learning_rate": 0.0002,
-      "loss": 0.0106,
+      "loss": 0.0043,
       "step": 913
     },
     {
-      "epoch": 16.62,
+      "epoch": 16.93,
       "learning_rate": 0.0002,
-      "loss": 0.0262,
+      "loss": 0.012,
       "step": 914
     },
     {
-      "epoch": 16.64,
+      "epoch": 16.94,
       "learning_rate": 0.0002,
-      "loss": 0.0124,
+      "loss": 0.0089,
       "step": 915
     },
     {
-      "epoch": 16.65,
+      "epoch": 16.96,
       "learning_rate": 0.0002,
-      "loss": 0.0129,
+      "loss": 0.0094,
       "step": 916
     },
     {
-      "epoch": 16.67,
+      "epoch": 16.98,
       "learning_rate": 0.0002,
-      "loss": 0.0175,
+      "loss": 0.0236,
       "step": 917
     },
     {
-      "epoch": 16.69,
+      "epoch": 17.0,
       "learning_rate": 0.0002,
-      "loss": 0.0228,
+      "loss": 0.0123,
+      "step": 918
+    },
+    {
+      "epoch": 17.0,
+      "eval_loss": 0.0074133859016001225,
+      "eval_runtime": 138.1405,
+      "eval_samples_per_second": 12.386,
+      "eval_steps_per_second": 0.391,
       "step": 918
     },
     {
-      "epoch": 16.71,
+      "epoch": 17.02,
       "learning_rate": 0.0002,
-      "loss": 0.0147,
+      "loss": 0.0044,
       "step": 919
     },
     {
-      "epoch": 16.73,
+      "epoch": 17.04,
       "learning_rate": 0.0002,
-      "loss": 0.021,
+      "loss": 0.0162,
       "step": 920
     },
     {
-      "epoch": 16.75,
+      "epoch": 17.06,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0095,
       "step": 921
     },
     {
-      "epoch": 16.76,
+      "epoch": 17.07,
       "learning_rate": 0.0002,
-      "loss": 0.0166,
+      "loss": 0.0156,
       "step": 922
     },
     {
-      "epoch": 16.78,
+      "epoch": 17.09,
       "learning_rate": 0.0002,
-      "loss": 0.0139,
+      "loss": 0.0075,
       "step": 923
     },
     {
-      "epoch": 16.8,
+      "epoch": 17.11,
       "learning_rate": 0.0002,
-      "loss": 0.0209,
+      "loss": 0.0082,
       "step": 924
     },
     {
-      "epoch": 16.82,
+      "epoch": 17.13,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0106,
       "step": 925
     },
     {
-      "epoch": 16.84,
+      "epoch": 17.15,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
+      "loss": 0.0049,
       "step": 926
     },
     {
-      "epoch": 16.85,
+      "epoch": 17.17,
       "learning_rate": 0.0002,
-      "loss": 0.0129,
+      "loss": 0.0049,
       "step": 927
     },
     {
-      "epoch": 16.87,
+      "epoch": 17.19,
       "learning_rate": 0.0002,
-      "loss": 0.0134,
+      "loss": 0.0031,
       "step": 928
     },
     {
-      "epoch": 16.89,
+      "epoch": 17.2,
       "learning_rate": 0.0002,
-      "loss": 0.0181,
+      "loss": 0.0105,
       "step": 929
     },
     {
-      "epoch": 16.91,
+      "epoch": 17.22,
       "learning_rate": 0.0002,
-      "loss": 0.021,
+      "loss": 0.0143,
       "step": 930
     },
     {
-      "epoch": 16.93,
+      "epoch": 17.24,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.0085,
       "step": 931
     },
     {
-      "epoch": 16.95,
+      "epoch": 17.26,
       "learning_rate": 0.0002,
-      "loss": 0.0092,
+      "loss": 0.01,
       "step": 932
     },
     {
-      "epoch": 16.96,
+      "epoch": 17.28,
       "learning_rate": 0.0002,
-      "loss": 0.012,
+      "loss": 0.01,
       "step": 933
     },
     {
-      "epoch": 16.98,
+      "epoch": 17.3,
       "learning_rate": 0.0002,
-      "loss": 0.0237,
+      "loss": 0.0094,
       "step": 934
     },
     {
-      "epoch": 17.0,
+      "epoch": 17.31,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
-      "step": 935
-    },
-    {
-      "epoch": 17.0,
-      "eval_loss": 0.009773855097591877,
-      "eval_runtime": 97.5075,
-      "eval_samples_per_second": 8.892,
-      "eval_steps_per_second": 0.564,
+      "loss": 0.0097,
       "step": 935
     },
     {
-      "epoch": 17.02,
+      "epoch": 17.33,
       "learning_rate": 0.0002,
-      "loss": 0.0066,
+      "loss": 0.0075,
       "step": 936
     },
     {
-      "epoch": 17.04,
+      "epoch": 17.35,
       "learning_rate": 0.0002,
-      "loss": 0.0122,
+      "loss": 0.0068,
       "step": 937
     },
     {
-      "epoch": 17.05,
+      "epoch": 17.37,
       "learning_rate": 0.0002,
-      "loss": 0.0115,
+      "loss": 0.0112,
       "step": 938
     },
     {
-      "epoch": 17.07,
+      "epoch": 17.39,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0176,
       "step": 939
     },
     {
-      "epoch": 17.09,
+      "epoch": 17.41,
       "learning_rate": 0.0002,
-      "loss": 0.0104,
+      "loss": 0.0093,
       "step": 940
     },
     {
-      "epoch": 17.11,
+      "epoch": 17.43,
       "learning_rate": 0.0002,
-      "loss": 0.0175,
+      "loss": 0.008,
       "step": 941
     },
     {
-      "epoch": 17.13,
+      "epoch": 17.44,
       "learning_rate": 0.0002,
-      "loss": 0.0087,
+      "loss": 0.0057,
       "step": 942
     },
     {
-      "epoch": 17.15,
+      "epoch": 17.46,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0113,
       "step": 943
     },
     {
-      "epoch": 17.16,
+      "epoch": 17.48,
       "learning_rate": 0.0002,
-      "loss": 0.0159,
+      "loss": 0.0098,
       "step": 944
     },
     {
-      "epoch": 17.18,
+      "epoch": 17.5,
       "learning_rate": 0.0002,
-      "loss": 0.0054,
+      "loss": 0.0098,
       "step": 945
     },
     {
-      "epoch": 17.2,
+      "epoch": 17.52,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
+      "loss": 0.01,
       "step": 946
     },
     {
-      "epoch": 17.22,
+      "epoch": 17.54,
       "learning_rate": 0.0002,
-      "loss": 0.0062,
+      "loss": 0.0071,
       "step": 947
     },
     {
-      "epoch": 17.24,
+      "epoch": 17.56,
       "learning_rate": 0.0002,
-      "loss": 0.0101,
+      "loss": 0.01,
       "step": 948
     },
     {
-      "epoch": 17.25,
+      "epoch": 17.57,
       "learning_rate": 0.0002,
-      "loss": 0.0133,
+      "loss": 0.0041,
       "step": 949
     },
     {
-      "epoch": 17.27,
+      "epoch": 17.59,
       "learning_rate": 0.0002,
-      "loss": 0.0081,
+      "loss": 0.0263,
       "step": 950
     },
     {
-      "epoch": 17.29,
+      "epoch": 17.61,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0083,
       "step": 951
     },
     {
-      "epoch": 17.31,
+      "epoch": 17.63,
       "learning_rate": 0.0002,
-      "loss": 0.0008,
+      "loss": 0.0028,
       "step": 952
     },
     {
-      "epoch": 17.33,
+      "epoch": 17.65,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0037,
       "step": 953
     },
     {
-      "epoch": 17.35,
+      "epoch": 17.67,
       "learning_rate": 0.0002,
-      "loss": 0.008,
+      "loss": 0.0028,
       "step": 954
     },
     {
-      "epoch": 17.36,
+      "epoch": 17.69,
       "learning_rate": 0.0002,
-      "loss": 0.0055,
+      "loss": 0.0043,
       "step": 955
     },
     {
-      "epoch": 17.38,
+      "epoch": 17.7,
       "learning_rate": 0.0002,
-      "loss": 0.0129,
+      "loss": 0.0077,
       "step": 956
     },
     {
-      "epoch": 17.4,
+      "epoch": 17.72,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.0102,
       "step": 957
     },
     {
-      "epoch": 17.42,
+      "epoch": 17.74,
       "learning_rate": 0.0002,
-      "loss": 0.0082,
+      "loss": 0.0061,
       "step": 958
     },
     {
-      "epoch": 17.44,
+      "epoch": 17.76,
       "learning_rate": 0.0002,
-      "loss": 0.0131,
+      "loss": 0.0061,
       "step": 959
     },
     {
-      "epoch": 17.45,
+      "epoch": 17.78,
       "learning_rate": 0.0002,
-      "loss": 0.0022,
+      "loss": 0.0062,
       "step": 960
     },
     {
-      "epoch": 17.47,
+      "epoch": 17.8,
       "learning_rate": 0.0002,
-      "loss": 0.0111,
+      "loss": 0.0109,
       "step": 961
     },
     {
-      "epoch": 17.49,
+      "epoch": 17.81,
       "learning_rate": 0.0002,
-      "loss": 0.0278,
+      "loss": 0.0197,
       "step": 962
     },
     {
-      "epoch": 17.51,
+      "epoch": 17.83,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0111,
       "step": 963
     },
     {
-      "epoch": 17.53,
+      "epoch": 17.85,
       "learning_rate": 0.0002,
-      "loss": 0.0118,
+      "loss": 0.0074,
       "step": 964
     },
     {
-      "epoch": 17.55,
+      "epoch": 17.87,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.0057,
       "step": 965
     },
     {
-      "epoch": 17.56,
+      "epoch": 17.89,
       "learning_rate": 0.0002,
-      "loss": 0.0058,
+      "loss": 0.0049,
       "step": 966
     },
     {
-      "epoch": 17.58,
+      "epoch": 17.91,
       "learning_rate": 0.0002,
-      "loss": 0.0077,
+      "loss": 0.0098,
       "step": 967
     },
     {
-      "epoch": 17.6,
+      "epoch": 17.93,
       "learning_rate": 0.0002,
-      "loss": 0.0063,
+      "loss": 0.0139,
       "step": 968
     },
     {
-      "epoch": 17.62,
+      "epoch": 17.94,
       "learning_rate": 0.0002,
-      "loss": 0.0038,
+      "loss": 0.0117,
       "step": 969
     },
     {
-      "epoch": 17.64,
+      "epoch": 17.96,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.0061,
       "step": 970
     },
     {
-      "epoch": 17.65,
+      "epoch": 17.98,
       "learning_rate": 0.0002,
-      "loss": 0.0033,
+      "loss": 0.0059,
       "step": 971
     },
     {
-      "epoch": 17.67,
+      "epoch": 18.0,
       "learning_rate": 0.0002,
-      "loss": 0.002,
+      "loss": 0.0041,
       "step": 972
     },
     {
-      "epoch": 17.69,
+      "epoch": 18.0,
+      "eval_loss": 0.006209332030266523,
+      "eval_runtime": 138.2325,
+      "eval_samples_per_second": 12.378,
+      "eval_steps_per_second": 0.391,
+      "step": 972
+    },
+    {
+      "epoch": 18.02,
       "learning_rate": 0.0002,
-      "loss": 0.0039,
+      "loss": 0.0088,
       "step": 973
     },
     {
-      "epoch": 17.71,
+      "epoch": 18.04,
       "learning_rate": 0.0002,
-      "loss": 0.0074,
+      "loss": 0.0015,
       "step": 974
     },
     {
-      "epoch": 17.73,
+      "epoch": 18.06,
       "learning_rate": 0.0002,
-      "loss": 0.0223,
+      "loss": 0.0084,
       "step": 975
     },
     {
-      "epoch": 17.75,
+      "epoch": 18.07,
       "learning_rate": 0.0002,
-      "loss": 0.0068,
+      "loss": 0.0081,
       "step": 976
     },
     {
-      "epoch": 17.76,
+      "epoch": 18.09,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.0063,
       "step": 977
     },
     {
-      "epoch": 17.78,
+      "epoch": 18.11,
       "learning_rate": 0.0002,
-      "loss": 0.0029,
+      "loss": 0.0053,
       "step": 978
     },
     {
-      "epoch": 17.8,
+      "epoch": 18.13,
       "learning_rate": 0.0002,
-      "loss": 0.0037,
+      "loss": 0.0048,
       "step": 979
     },
     {
-      "epoch": 17.82,
+      "epoch": 18.15,
       "learning_rate": 0.0002,
-      "loss": 0.0178,
+      "loss": 0.0046,
       "step": 980
     },
     {
-      "epoch": 17.84,
+      "epoch": 18.17,
       "learning_rate": 0.0002,
-      "loss": 0.0095,
+      "loss": 0.0024,
       "step": 981
     },
     {
-      "epoch": 17.85,
+      "epoch": 18.19,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0058,
       "step": 982
     },
     {
-      "epoch": 17.87,
+      "epoch": 18.2,
       "learning_rate": 0.0002,
-      "loss": 0.0066,
+      "loss": 0.0067,
       "step": 983
     },
     {
-      "epoch": 17.89,
+      "epoch": 18.22,
       "learning_rate": 0.0002,
-      "loss": 0.0053,
+      "loss": 0.0135,
       "step": 984
     },
     {
-      "epoch": 17.91,
+      "epoch": 18.24,
       "learning_rate": 0.0002,
-      "loss": 0.0084,
+      "loss": 0.0049,
       "step": 985
     },
     {
-      "epoch": 17.93,
+      "epoch": 18.26,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.0029,
       "step": 986
     },
     {
-      "epoch": 17.95,
+      "epoch": 18.28,
       "learning_rate": 0.0002,
-      "loss": 0.0026,
+      "loss": 0.0086,
       "step": 987
     },
     {
-      "epoch": 17.96,
+      "epoch": 18.3,
       "learning_rate": 0.0002,
-      "loss": 0.0072,
+      "loss": 0.0079,
       "step": 988
     },
     {
-      "epoch": 17.98,
+      "epoch": 18.31,
       "learning_rate": 0.0002,
-      "loss": 0.0084,
+      "loss": 0.0061,
       "step": 989
     },
     {
-      "epoch": 18.0,
+      "epoch": 18.33,
       "learning_rate": 0.0002,
-      "loss": 0.0084,
-      "step": 990
-    },
-    {
-      "epoch": 18.0,
-      "eval_loss": 0.005390956066548824,
-      "eval_runtime": 97.3845,
-      "eval_samples_per_second": 8.903,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.011,
       "step": 990
     },
     {
-      "epoch": 18.02,
+      "epoch": 18.35,
       "learning_rate": 0.0002,
-      "loss": 0.0029,
+      "loss": 0.006,
       "step": 991
     },
     {
-      "epoch": 18.04,
+      "epoch": 18.37,
       "learning_rate": 0.0002,
-      "loss": 0.0026,
+      "loss": 0.0049,
       "step": 992
     },
     {
-      "epoch": 18.05,
+      "epoch": 18.39,
       "learning_rate": 0.0002,
-      "loss": 0.0036,
+      "loss": 0.0043,
       "step": 993
     },
     {
-      "epoch": 18.07,
+      "epoch": 18.41,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.0074,
       "step": 994
     },
     {
-      "epoch": 18.09,
+      "epoch": 18.43,
       "learning_rate": 0.0002,
-      "loss": 0.0042,
+      "loss": 0.0099,
       "step": 995
     },
     {
-      "epoch": 18.11,
+      "epoch": 18.44,
       "learning_rate": 0.0002,
-      "loss": 0.0158,
+      "loss": 0.0153,
       "step": 996
     },
     {
-      "epoch": 18.13,
+      "epoch": 18.46,
       "learning_rate": 0.0002,
-      "loss": 0.0043,
+      "loss": 0.0077,
       "step": 997
     },
     {
-      "epoch": 18.15,
+      "epoch": 18.48,
       "learning_rate": 0.0002,
-      "loss": 0.0022,
+      "loss": 0.0065,
       "step": 998
     },
     {
-      "epoch": 18.16,
+      "epoch": 18.5,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.006,
       "step": 999
     },
     {
-      "epoch": 18.18,
+      "epoch": 18.52,
       "learning_rate": 0.0002,
-      "loss": 0.0027,
+      "loss": 0.0091,
       "step": 1000
     },
     {
-      "epoch": 18.2,
+      "epoch": 18.54,
       "learning_rate": 0.0002,
-      "loss": 0.0028,
+      "loss": 0.0059,
       "step": 1001
     },
     {
-      "epoch": 18.22,
+      "epoch": 18.56,
       "learning_rate": 0.0002,
-      "loss": 0.0006,
+      "loss": 0.0081,
       "step": 1002
     },
     {
-      "epoch": 18.24,
+      "epoch": 18.57,
       "learning_rate": 0.0002,
-      "loss": 0.0132,
+      "loss": 0.0142,
       "step": 1003
     },
     {
-      "epoch": 18.25,
+      "epoch": 18.59,
       "learning_rate": 0.0002,
-      "loss": 0.0034,
+      "loss": 0.0067,
       "step": 1004
     },
     {
-      "epoch": 18.27,
+      "epoch": 18.61,
       "learning_rate": 0.0002,
-      "loss": 0.0088,
+      "loss": 0.0073,
       "step": 1005
     },
     {
-      "epoch": 18.29,
+      "epoch": 18.63,
       "learning_rate": 0.0002,
-      "loss": 0.0085,
+      "loss": 0.0061,
       "step": 1006
     },
     {
-      "epoch": 18.31,
+      "epoch": 18.65,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0101,
       "step": 1007
     },
     {
-      "epoch": 18.33,
+      "epoch": 18.67,
       "learning_rate": 0.0002,
-      "loss": 0.0011,
+      "loss": 0.0054,
       "step": 1008
     },
     {
-      "epoch": 18.35,
+      "epoch": 18.69,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.0044,
       "step": 1009
     },
     {
-      "epoch": 18.36,
+      "epoch": 18.7,
       "learning_rate": 0.0002,
-      "loss": 0.0045,
+      "loss": 0.0062,
       "step": 1010
     },
     {
-      "epoch": 18.38,
+      "epoch": 18.72,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0037,
       "step": 1011
     },
     {
-      "epoch": 18.4,
+      "epoch": 18.74,
       "learning_rate": 0.0002,
-      "loss": 0.0041,
+      "loss": 0.0107,
       "step": 1012
     },
     {
-      "epoch": 18.42,
+      "epoch": 18.76,
       "learning_rate": 0.0002,
-      "loss": 0.01,
+      "loss": 0.0082,
       "step": 1013
     },
     {
-      "epoch": 18.44,
+      "epoch": 18.78,
       "learning_rate": 0.0002,
-      "loss": 0.0057,
+      "loss": 0.009,
       "step": 1014
     },
     {
-      "epoch": 18.45,
+      "epoch": 18.8,
       "learning_rate": 0.0002,
-      "loss": 0.0022,
+      "loss": 0.0083,
       "step": 1015
     },
     {
-      "epoch": 18.47,
+      "epoch": 18.81,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0044,
       "step": 1016
     },
     {
-      "epoch": 18.49,
+      "epoch": 18.83,
       "learning_rate": 0.0002,
-      "loss": 0.0012,
+      "loss": 0.0054,
       "step": 1017
     },
     {
-      "epoch": 18.51,
+      "epoch": 18.85,
       "learning_rate": 0.0002,
-      "loss": 0.0076,
+      "loss": 0.0062,
       "step": 1018
     },
     {
-      "epoch": 18.53,
+      "epoch": 18.87,
       "learning_rate": 0.0002,
-      "loss": 0.0061,
+      "loss": 0.0046,
       "step": 1019
     },
     {
-      "epoch": 18.55,
+      "epoch": 18.89,
       "learning_rate": 0.0002,
-      "loss": 0.0016,
+      "loss": 0.01,
       "step": 1020
     },
     {
-      "epoch": 18.56,
+      "epoch": 18.91,
       "learning_rate": 0.0002,
-      "loss": 0.0129,
+      "loss": 0.0052,
       "step": 1021
     },
     {
-      "epoch": 18.58,
+      "epoch": 18.93,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.0055,
       "step": 1022
     },
     {
-      "epoch": 18.6,
+      "epoch": 18.94,
       "learning_rate": 0.0002,
-      "loss": 0.0026,
+      "loss": 0.0028,
       "step": 1023
     },
     {
-      "epoch": 18.62,
+      "epoch": 18.96,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.0085,
       "step": 1024
     },
     {
-      "epoch": 18.64,
+      "epoch": 18.98,
       "learning_rate": 0.0002,
-      "loss": 0.005,
+      "loss": 0.0047,
       "step": 1025
     },
     {
-      "epoch": 18.65,
+      "epoch": 19.0,
       "learning_rate": 0.0002,
-      "loss": 0.0044,
+      "loss": 0.009,
       "step": 1026
     },
     {
-      "epoch": 18.67,
+      "epoch": 19.0,
+      "eval_loss": 0.005724901333451271,
+      "eval_runtime": 138.3063,
+      "eval_samples_per_second": 12.371,
+      "eval_steps_per_second": 0.39,
+      "step": 1026
+    },
+    {
+      "epoch": 19.02,
       "learning_rate": 0.0002,
-      "loss": 0.0025,
+      "loss": 0.0059,
       "step": 1027
     },
     {
-      "epoch": 18.69,
+      "epoch": 19.04,
       "learning_rate": 0.0002,
-      "loss": 0.0042,
+      "loss": 0.0021,
       "step": 1028
     },
     {
-      "epoch": 18.71,
+      "epoch": 19.06,
       "learning_rate": 0.0002,
-      "loss": 0.0038,
+      "loss": 0.0042,
       "step": 1029
     },
     {
-      "epoch": 18.73,
+      "epoch": 19.07,
       "learning_rate": 0.0002,
-      "loss": 0.0046,
+      "loss": 0.0022,
       "step": 1030
     },
     {
-      "epoch": 18.75,
+      "epoch": 19.09,
       "learning_rate": 0.0002,
-      "loss": 0.0043,
+      "loss": 0.0067,
       "step": 1031
     },
     {
-      "epoch": 18.76,
+      "epoch": 19.11,
       "learning_rate": 0.0002,
-      "loss": 0.0121,
+      "loss": 0.0066,
       "step": 1032
     },
     {
-      "epoch": 18.78,
+      "epoch": 19.13,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0067,
       "step": 1033
     },
     {
-      "epoch": 18.8,
+      "epoch": 19.15,
       "learning_rate": 0.0002,
-      "loss": 0.0114,
+      "loss": 0.0058,
       "step": 1034
     },
     {
-      "epoch": 18.82,
+      "epoch": 19.17,
       "learning_rate": 0.0002,
-      "loss": 0.0038,
+      "loss": 0.0072,
       "step": 1035
     },
     {
-      "epoch": 18.84,
+      "epoch": 19.19,
       "learning_rate": 0.0002,
-      "loss": 0.0028,
+      "loss": 0.0043,
       "step": 1036
     },
     {
-      "epoch": 18.85,
+      "epoch": 19.2,
       "learning_rate": 0.0002,
-      "loss": 0.0075,
+      "loss": 0.0058,
       "step": 1037
     },
     {
-      "epoch": 18.87,
+      "epoch": 19.22,
       "learning_rate": 0.0002,
-      "loss": 0.0069,
+      "loss": 0.005,
       "step": 1038
     },
     {
-      "epoch": 18.89,
+      "epoch": 19.24,
       "learning_rate": 0.0002,
-      "loss": 0.0039,
+      "loss": 0.0037,
       "step": 1039
     },
     {
-      "epoch": 18.91,
+      "epoch": 19.26,
       "learning_rate": 0.0002,
-      "loss": 0.0012,
+      "loss": 0.0201,
       "step": 1040
     },
     {
-      "epoch": 18.93,
+      "epoch": 19.28,
       "learning_rate": 0.0002,
-      "loss": 0.0081,
+      "loss": 0.0089,
       "step": 1041
     },
     {
-      "epoch": 18.95,
+      "epoch": 19.3,
       "learning_rate": 0.0002,
-      "loss": 0.0021,
+      "loss": 0.005,
       "step": 1042
     },
     {
-      "epoch": 18.96,
+      "epoch": 19.31,
       "learning_rate": 0.0002,
-      "loss": 0.0091,
+      "loss": 0.0043,
       "step": 1043
     },
     {
-      "epoch": 18.98,
+      "epoch": 19.33,
       "learning_rate": 0.0002,
-      "loss": 0.0056,
+      "loss": 0.0029,
       "step": 1044
     },
     {
-      "epoch": 19.0,
+      "epoch": 19.35,
       "learning_rate": 0.0002,
-      "loss": 0.0039,
-      "step": 1045
-    },
-    {
-      "epoch": 19.0,
-      "eval_loss": 0.004796092864125967,
-      "eval_runtime": 97.366,
-      "eval_samples_per_second": 8.905,
-      "eval_steps_per_second": 0.565,
+      "loss": 0.0082,
       "step": 1045
     },
     {
-      "epoch": 19.02,
+      "epoch": 19.37,
       "learning_rate": 0.0002,
-      "loss": 0.0024,
+      "loss": 0.0026,
       "step": 1046
     },
     {
-      "epoch": 19.04,
+      "epoch": 19.39,
       "learning_rate": 0.0002,
-      "loss": 0.0059,
+      "loss": 0.0099,
       "step": 1047
     },
     {
-      "epoch": 19.05,
+      "epoch": 19.41,
       "learning_rate": 0.0002,
-      "loss": 0.0146,
+      "loss": 0.0062,
       "step": 1048
     },
     {
-      "epoch": 19.07,
+      "epoch": 19.43,
       "learning_rate": 0.0002,
-      "loss": 0.0027,
+      "loss": 0.0052,
       "step": 1049
     },
     {
-      "epoch": 19.09,
+      "epoch": 19.44,
       "learning_rate": 0.0002,
-      "loss": 0.0015,
+      "loss": 0.0049,
       "step": 1050
     },
     {
-      "epoch": 19.11,
+      "epoch": 19.46,
       "learning_rate": 0.0002,
-      "loss": 0.0071,
+      "loss": 0.0028,
       "step": 1051
     },
     {
-      "epoch": 19.13,
+      "epoch": 19.48,
       "learning_rate": 0.0002,
-      "loss": 0.0019,
+      "loss": 0.0063,
       "step": 1052
     },
     {
-      "epoch": 19.15,
+      "epoch": 19.5,
       "learning_rate": 0.0002,
-      "loss": 0.001,
+      "loss": 0.0072,
       "step": 1053
     },
     {
-      "epoch": 19.16,
+      "epoch": 19.52,
       "learning_rate": 0.0002,
-      "loss": 0.0113,
+      "loss": 0.0064,
       "step": 1054
     },
     {
-      "epoch": 19.18,
+      "epoch": 19.54,
       "learning_rate": 0.0002,
-      "loss": 0.0074,
+      "loss": 0.0047,
       "step": 1055
     },
     {
-      "epoch": 19.2,
+      "epoch": 19.56,
       "learning_rate": 0.0002,
-      "loss": 0.0039,
+      "loss": 0.0107,
       "step": 1056
     },
     {
-      "epoch": 19.22,
+      "epoch": 19.57,
       "learning_rate": 0.0002,
-      "loss": 0.0044,
+      "loss": 0.0062,
       "step": 1057
     },
     {
-      "epoch": 19.24,
+      "epoch": 19.59,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.005,
       "step": 1058
     },
     {
-      "epoch": 19.25,
+      "epoch": 19.61,
       "learning_rate": 0.0002,
-      "loss": 0.0026,
+      "loss": 0.0072,
       "step": 1059
     },
     {
-      "epoch": 19.27,
+      "epoch": 19.63,
       "learning_rate": 0.0002,
-      "loss": 0.0036,
+      "loss": 0.0037,
       "step": 1060
     },
     {
-      "epoch": 19.29,
+      "epoch": 19.65,
       "learning_rate": 0.0002,
-      "loss": 0.001,
+      "loss": 0.004,
       "step": 1061
     },
     {
-      "epoch": 19.31,
+      "epoch": 19.67,
       "learning_rate": 0.0002,
-      "loss": 0.0014,
+      "loss": 0.0083,
       "step": 1062
     },
     {
-      "epoch": 19.33,
+      "epoch": 19.69,
       "learning_rate": 0.0002,
-      "loss": 0.0065,
+      "loss": 0.0041,
       "step": 1063
     },
     {
-      "epoch": 19.35,
+      "epoch": 19.7,
       "learning_rate": 0.0002,
-      "loss": 0.0151,
+      "loss": 0.0056,
       "step": 1064
     },
     {
-      "epoch": 19.36,
+      "epoch": 19.72,
       "learning_rate": 0.0002,
-      "loss": 0.0049,
+      "loss": 0.0037,
       "step": 1065
     },
     {
-      "epoch": 19.38,
+      "epoch": 19.74,
       "learning_rate": 0.0002,
-      "loss": 0.0043,
+      "loss": 0.0093,
       "step": 1066
     },
     {
-      "epoch": 19.4,
+      "epoch": 19.76,
       "learning_rate": 0.0002,
-      "loss": 0.0064,
+      "loss": 0.0067,
       "step": 1067
     },
     {
-      "epoch": 19.42,
+      "epoch": 19.78,
       "learning_rate": 0.0002,
-      "loss": 0.0035,
+      "loss": 0.005,
       "step": 1068
     },
     {
-      "epoch": 19.44,
+      "epoch": 19.8,
       "learning_rate": 0.0002,
-      "loss": 0.0025,
+      "loss": 0.0085,
       "step": 1069
     },
     {
-      "epoch": 19.45,
+      "epoch": 19.81,
       "learning_rate": 0.0002,
-      "loss": 0.0117,
+      "loss": 0.0057,
       "step": 1070
     },
     {
-      "epoch": 19.47,
+      "epoch": 19.83,
       "learning_rate": 0.0002,
-      "loss": 0.0025,
+      "loss": 0.0083,
       "step": 1071
     },
     {
-      "epoch": 19.49,
+      "epoch": 19.85,
       "learning_rate": 0.0002,
-      "loss": 0.0008,
+      "loss": 0.0054,
       "step": 1072
     },
     {
-      "epoch": 19.51,
+      "epoch": 19.87,
       "learning_rate": 0.0002,
-      "loss": 0.0024,
+      "loss": 0.0087,
       "step": 1073
     },
     {
-      "epoch": 19.53,
+      "epoch": 19.89,
       "learning_rate": 0.0002,
-      "loss": 0.0023,
+      "loss": 0.0082,
       "step": 1074
     },
     {
-      "epoch": 19.55,
+      "epoch": 19.91,
       "learning_rate": 0.0002,
-      "loss": 0.0009,
+      "loss": 0.0085,
       "step": 1075
     },
     {
-      "epoch": 19.56,
+      "epoch": 19.93,
       "learning_rate": 0.0002,
-      "loss": 0.004,
+      "loss": 0.0042,
       "step": 1076
     },
     {
-      "epoch": 19.58,
-      "learning_rate": 0.0002,
-      "loss": 0.0006,
-      "step": 1077
-    },
-    {
-      "epoch": 19.6,
-      "learning_rate": 0.0002,
-      "loss": 0.0054,
-      "step": 1078
-    },
-    {
-      "epoch": 19.62,
-      "learning_rate": 0.0002,
-      "loss": 0.0008,
-      "step": 1079
-    },
-    {
-      "epoch": 19.64,
-      "learning_rate": 0.0002,
-      "loss": 0.0007,
-      "step": 1080
-    },
-    {
-      "epoch": 19.65,
-      "learning_rate": 0.0002,
-      "loss": 0.0012,
-      "step": 1081
-    },
-    {
-      "epoch": 19.67,
-      "learning_rate": 0.0002,
-      "loss": 0.0063,
-      "step": 1082
-    },
-    {
-      "epoch": 19.69,
-      "learning_rate": 0.0002,
-      "loss": 0.0023,
-      "step": 1083
-    },
-    {
-      "epoch": 19.71,
-      "learning_rate": 0.0002,
-      "loss": 0.0005,
-      "step": 1084
-    },
-    {
-      "epoch": 19.73,
-      "learning_rate": 0.0002,
-      "loss": 0.0017,
-      "step": 1085
-    },
-    {
-      "epoch": 19.75,
-      "learning_rate": 0.0002,
-      "loss": 0.0028,
-      "step": 1086
-    },
-    {
-      "epoch": 19.76,
-      "learning_rate": 0.0002,
-      "loss": 0.0003,
-      "step": 1087
-    },
-    {
-      "epoch": 19.78,
-      "learning_rate": 0.0002,
-      "loss": 0.0004,
-      "step": 1088
-    },
-    {
-      "epoch": 19.8,
-      "learning_rate": 0.0002,
-      "loss": 0.0021,
-      "step": 1089
-    },
-    {
-      "epoch": 19.82,
-      "learning_rate": 0.0002,
-      "loss": 0.0004,
-      "step": 1090
-    },
-    {
-      "epoch": 19.84,
-      "learning_rate": 0.0002,
-      "loss": 0.0027,
-      "step": 1091
-    },
-    {
-      "epoch": 19.85,
+      "epoch": 19.94,
       "learning_rate": 0.0002,
       "loss": 0.0059,
-      "step": 1092
-    },
-    {
-      "epoch": 19.87,
-      "learning_rate": 0.0002,
-      "loss": 0.0086,
-      "step": 1093
-    },
-    {
-      "epoch": 19.89,
-      "learning_rate": 0.0002,
-      "loss": 0.0003,
-      "step": 1094
-    },
-    {
-      "epoch": 19.91,
-      "learning_rate": 0.0002,
-      "loss": 0.001,
-      "step": 1095
-    },
-    {
-      "epoch": 19.93,
-      "learning_rate": 0.0002,
-      "loss": 0.0025,
-      "step": 1096
-    },
-    {
-      "epoch": 19.95,
-      "learning_rate": 0.0002,
-      "loss": 0.0033,
-      "step": 1097
+      "step": 1077
     },
     {
       "epoch": 19.96,
       "learning_rate": 0.0002,
-      "loss": 0.0027,
-      "step": 1098
+      "loss": 0.0036,
+      "step": 1078
     },
     {
       "epoch": 19.98,
       "learning_rate": 0.0002,
-      "loss": 0.0014,
-      "step": 1099
+      "loss": 0.0093,
+      "step": 1079
     },
     {
       "epoch": 20.0,
       "learning_rate": 0.0002,
-      "loss": 0.0021,
-      "step": 1100
+      "loss": 0.0074,
+      "step": 1080
     },
     {
       "epoch": 20.0,
-      "eval_loss": 0.002584549132734537,
-      "eval_runtime": 97.4365,
-      "eval_samples_per_second": 8.898,
-      "eval_steps_per_second": 0.564,
-      "step": 1100
+      "eval_loss": 0.005008201580494642,
+      "eval_runtime": 138.3235,
+      "eval_samples_per_second": 12.37,
+      "eval_steps_per_second": 0.39,
+      "step": 1080
     },
     {
       "epoch": 20.0,
-      "step": 1100,
-      "total_flos": 1.6053819377430036e+18,
-      "train_loss": 0.0495865726133872,
-      "train_runtime": 11409.4652,
-      "train_samples_per_second": 1.52,
-      "train_steps_per_second": 0.096
+      "step": 1080,
+      "total_flos": 3.913384658510807e+18,
+      "train_loss": 0.05248785450024082,
+      "train_runtime": 18010.5365,
+      "train_samples_per_second": 1.9,
+      "train_steps_per_second": 0.06
     }
   ],
   "logging_steps": 1.0,
-  "max_steps": 1100,
+  "max_steps": 1080,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 20,
   "save_steps": 50000,
-  "total_flos": 1.6053819377430036e+18,
+  "total_flos": 3.913384658510807e+18,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null