diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,7 +2,7 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,5375 +10,2690 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.0428,
+      "loss": 9.0656,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.829,
+      "loss": 8.6633,
       "step": 2
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 2e-08,
-      "loss": 8.6694,
+      "loss": 8.9119,
       "step": 3
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 4e-08,
-      "loss": 8.9229,
+      "epoch": 0.01,
+      "learning_rate": 2e-08,
+      "loss": 8.8743,
       "step": 4
     },
     {
       "epoch": 0.01,
       "learning_rate": 4e-08,
-      "loss": 8.8281,
+      "loss": 8.7322,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.000000000000001e-08,
-      "loss": 8.6038,
+      "loss": 8.8432,
       "step": 6
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 8e-08,
-      "loss": 8.9877,
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 8.7598,
       "step": 7
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 8e-08,
-      "loss": 8.6319,
+      "loss": 8.7179,
       "step": 8
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.0000000000000001e-07,
-      "loss": 8.7545,
+      "loss": 8.7009,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 8.8392,
+      "epoch": 0.02,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 8.8273,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4e-07,
-      "loss": 8.7453,
+      "epoch": 0.02,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 8.6765,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4e-07,
-      "loss": 8.8195,
+      "epoch": 0.03,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 8.7786,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6e-07,
-      "loss": 8.7024,
+      "epoch": 0.03,
+      "learning_rate": 1.4e-07,
+      "loss": 8.6871,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.8e-07,
-      "loss": 9.0251,
+      "epoch": 0.03,
+      "learning_rate": 1.6e-07,
+      "loss": 8.8664,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 8.6294,
+      "epoch": 0.03,
+      "learning_rate": 1.8e-07,
+      "loss": 8.658,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.2e-07,
-      "loss": 8.5382,
+      "epoch": 0.04,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 8.8457,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 8.8118,
+      "epoch": 0.04,
+      "learning_rate": 2.2e-07,
+      "loss": 8.5633,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.6e-07,
-      "loss": 8.4245,
+      "epoch": 0.04,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 8.6972,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.8e-07,
-      "loss": 8.4836,
+      "epoch": 0.04,
+      "learning_rate": 2.6e-07,
+      "loss": 8.7013,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 8.712,
+      "epoch": 0.04,
+      "learning_rate": 2.8e-07,
+      "loss": 8.6702,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2e-07,
-      "loss": 8.4121,
+      "epoch": 0.05,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 8.9022,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 8.6381,
+      "epoch": 0.05,
+      "learning_rate": 3.2e-07,
+      "loss": 8.8576,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.6e-07,
-      "loss": 8.5455,
+      "epoch": 0.05,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 8.7779,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.8e-07,
-      "loss": 8.5124,
+      "epoch": 0.05,
+      "learning_rate": 3.6e-07,
+      "loss": 8.834,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 8.6074,
+      "epoch": 0.06,
+      "learning_rate": 3.8e-07,
+      "loss": 8.5806,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 8.9795,
+      "epoch": 0.06,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 8.4857,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.4e-07,
-      "loss": 8.5878,
+      "epoch": 0.06,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 8.2719,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.6000000000000004e-07,
-      "loss": 8.5223,
+      "epoch": 0.06,
+      "learning_rate": 4.4e-07,
+      "loss": 8.3136,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 8.467,
+      "epoch": 0.07,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 8.5229,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 8.405,
+      "epoch": 0.07,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 8.2262,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.2e-07,
-      "loss": 8.5238,
+      "epoch": 0.07,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 8.3328,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 5.4e-07,
-      "loss": 8.301,
+      "epoch": 0.07,
+      "learning_rate": 5.2e-07,
+      "loss": 8.2745,
       "step": 32
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 5.4e-07,
-      "loss": 8.3524,
+      "loss": 8.1359,
       "step": 33
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 5.6e-07,
-      "loss": 8.3574,
+      "loss": 8.2812,
       "step": 34
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 5.800000000000001e-07,
-      "loss": 8.3977,
+      "loss": 8.2066,
       "step": 35
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 6.000000000000001e-07,
-      "loss": 8.2797,
+      "loss": 8.2059,
       "step": 36
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 6.200000000000001e-07,
-      "loss": 8.1361,
+      "loss": 8.1422,
       "step": 37
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 6.4e-07,
-      "loss": 8.5122,
+      "loss": 8.1902,
       "step": 38
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 6.6e-07,
-      "loss": 8.3448,
+      "loss": 8.1267,
       "step": 39
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 6.800000000000001e-07,
-      "loss": 8.0017,
+      "loss": 8.2181,
       "step": 40
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 7.000000000000001e-07,
-      "loss": 8.1624,
+      "loss": 8.0245,
       "step": 41
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 7.2e-07,
-      "loss": 8.1326,
+      "loss": 8.1449,
       "step": 42
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 7.4e-07,
-      "loss": 7.9557,
+      "loss": 7.998,
       "step": 43
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 7.6e-07,
-      "loss": 8.3076,
+      "loss": 8.1898,
       "step": 44
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 7.8e-07,
-      "loss": 8.324,
+      "loss": 8.2973,
       "step": 45
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 8.000000000000001e-07,
-      "loss": 8.5414,
+      "loss": 8.1341,
       "step": 46
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 8.200000000000001e-07,
-      "loss": 8.2702,
+      "loss": 8.1824,
       "step": 47
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 8.400000000000001e-07,
-      "loss": 8.3125,
+      "loss": 8.4063,
       "step": 48
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 8.6e-07,
-      "loss": 8.3917,
+      "loss": 8.26,
       "step": 49
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 8.8e-07,
-      "loss": 8.2945,
+      "loss": 8.3837,
       "step": 50
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 9.000000000000001e-07,
-      "loss": 7.7748,
+      "loss": 7.823,
       "step": 51
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 9.200000000000001e-07,
-      "loss": 7.9108,
+      "loss": 7.7067,
       "step": 52
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 9.400000000000001e-07,
-      "loss": 7.6341,
+      "loss": 7.8324,
       "step": 53
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 9.600000000000001e-07,
-      "loss": 7.9437,
+      "loss": 7.7619,
       "step": 54
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 9.800000000000001e-07,
-      "loss": 7.7416,
+      "loss": 7.5628,
       "step": 55
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.13,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 7.8138,
+      "loss": 7.6529,
       "step": 56
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.13,
       "learning_rate": 1.02e-06,
-      "loss": 7.7159,
+      "loss": 7.6872,
       "step": 57
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 1.04e-06,
-      "loss": 7.7735,
+      "loss": 7.7812,
       "step": 58
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 1.06e-06,
-      "loss": 7.6901,
+      "loss": 7.6103,
       "step": 59
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 1.08e-06,
-      "loss": 7.6372,
+      "loss": 7.5827,
       "step": 60
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 1.1e-06,
-      "loss": 7.7964,
+      "loss": 7.5597,
       "step": 61
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 1.12e-06,
-      "loss": 7.6406,
+      "loss": 7.5961,
       "step": 62
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 1.14e-06,
-      "loss": 7.6627,
+      "loss": 7.521,
       "step": 63
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 1.1600000000000001e-06,
-      "loss": 7.7208,
+      "loss": 7.5574,
       "step": 64
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 1.1800000000000001e-06,
-      "loss": 7.4625,
+      "loss": 7.5383,
       "step": 65
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 1.2000000000000002e-06,
-      "loss": 7.9479,
+      "loss": 7.5139,
       "step": 66
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 1.2200000000000002e-06,
-      "loss": 7.5426,
+      "loss": 7.4725,
       "step": 67
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 1.2400000000000002e-06,
-      "loss": 7.4711,
+      "loss": 7.4597,
       "step": 68
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 1.26e-06,
-      "loss": 7.4953,
+      "loss": 7.5113,
       "step": 69
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 1.28e-06,
-      "loss": 7.4501,
+      "loss": 7.4046,
       "step": 70
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 1.3e-06,
-      "loss": 7.5951,
+      "loss": 7.5338,
       "step": 71
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 1.32e-06,
-      "loss": 7.5967,
+      "loss": 7.513,
       "step": 72
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 1.34e-06,
-      "loss": 7.3933,
+      "loss": 7.4987,
       "step": 73
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 1.3600000000000001e-06,
-      "loss": 7.5341,
+      "loss": 7.5031,
       "step": 74
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 1.3800000000000001e-06,
-      "loss": 7.4092,
+      "loss": 7.4272,
       "step": 75
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 7.4695,
+      "loss": 7.4681,
       "step": 76
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 1.42e-06,
-      "loss": 7.3926,
+      "loss": 7.4724,
       "step": 77
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 1.44e-06,
-      "loss": 7.3551,
+      "loss": 7.4617,
       "step": 78
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 1.46e-06,
-      "loss": 7.3655,
+      "loss": 7.3632,
       "step": 79
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 1.48e-06,
-      "loss": 7.4154,
+      "loss": 7.4406,
       "step": 80
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 1.5e-06,
-      "loss": 7.4007,
+      "loss": 7.3949,
       "step": 81
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 1.52e-06,
-      "loss": 7.3681,
+      "loss": 7.4337,
       "step": 82
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 1.54e-06,
-      "loss": 7.4672,
+      "loss": 7.4132,
       "step": 83
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 1.56e-06,
-      "loss": 7.289,
+      "loss": 7.3994,
       "step": 84
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.19,
       "learning_rate": 1.5800000000000001e-06,
-      "loss": 7.3618,
+      "loss": 7.355,
       "step": 85
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.19,
       "learning_rate": 1.6000000000000001e-06,
-      "loss": 7.3641,
+      "loss": 7.4217,
       "step": 86
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 1.6200000000000002e-06,
-      "loss": 7.3236,
+      "loss": 7.3948,
       "step": 87
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 1.6400000000000002e-06,
-      "loss": 7.3401,
+      "loss": 7.3221,
       "step": 88
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 1.6600000000000002e-06,
-      "loss": 7.3653,
+      "loss": 7.2875,
       "step": 89
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 1.6800000000000002e-06,
-      "loss": 7.4319,
+      "loss": 7.4339,
       "step": 90
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 1.7000000000000002e-06,
-      "loss": 7.5956,
+      "loss": 7.3627,
       "step": 91
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.21,
       "learning_rate": 1.72e-06,
-      "loss": 7.4795,
+      "loss": 7.5915,
       "step": 92
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.21,
       "learning_rate": 1.74e-06,
-      "loss": 7.2545,
+      "loss": 7.4667,
       "step": 93
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.21,
       "learning_rate": 1.76e-06,
-      "loss": 7.3914,
+      "loss": 7.3207,
       "step": 94
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.21,
       "learning_rate": 1.7800000000000001e-06,
-      "loss": 7.5556,
+      "loss": 7.282,
       "step": 95
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 1.8000000000000001e-06,
-      "loss": 7.688,
+      "loss": 7.5122,
       "step": 96
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 1.8200000000000002e-06,
-      "loss": 7.5064,
+      "loss": 7.4713,
       "step": 97
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 1.8400000000000002e-06,
-      "loss": 7.6114,
+      "loss": 7.2456,
       "step": 98
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 1.8600000000000002e-06,
-      "loss": 7.4397,
+      "loss": 7.5184,
       "step": 99
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 1.8800000000000002e-06,
-      "loss": 7.4834,
+      "loss": 7.3295,
       "step": 100
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.23,
       "learning_rate": 1.9000000000000002e-06,
-      "loss": 7.4016,
+      "loss": 7.3192,
       "step": 101
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.23,
       "learning_rate": 1.9200000000000003e-06,
-      "loss": 7.4267,
+      "loss": 7.2951,
       "step": 102
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.23,
       "learning_rate": 1.94e-06,
-      "loss": 7.3692,
+      "loss": 7.1776,
       "step": 103
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.23,
       "learning_rate": 1.9600000000000003e-06,
-      "loss": 7.2584,
+      "loss": 7.2282,
       "step": 104
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 1.98e-06,
-      "loss": 7.1761,
+      "loss": 7.2341,
       "step": 105
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 7.2418,
+      "loss": 7.2799,
       "step": 106
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 2.02e-06,
-      "loss": 7.3251,
+      "loss": 7.223,
       "step": 107
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 2.04e-06,
-      "loss": 7.3413,
+      "loss": 7.2384,
       "step": 108
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 2.06e-06,
-      "loss": 7.0807,
+      "loss": 7.126,
       "step": 109
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.25,
       "learning_rate": 2.08e-06,
-      "loss": 7.1881,
+      "loss": 7.2243,
       "step": 110
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.25,
       "learning_rate": 2.1000000000000002e-06,
-      "loss": 7.2618,
+      "loss": 7.1182,
       "step": 111
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.25,
       "learning_rate": 2.12e-06,
-      "loss": 7.3473,
+      "loss": 7.1275,
       "step": 112
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.25,
       "learning_rate": 2.1400000000000003e-06,
-      "loss": 7.221,
+      "loss": 7.2087,
       "step": 113
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 2.16e-06,
-      "loss": 7.1923,
+      "loss": 7.2448,
       "step": 114
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 2.1800000000000003e-06,
-      "loss": 7.2204,
+      "loss": 7.174,
       "step": 115
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 2.2e-06,
-      "loss": 7.1247,
+      "loss": 7.2965,
       "step": 116
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 2.2200000000000003e-06,
-      "loss": 7.2446,
+      "loss": 7.2543,
       "step": 117
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 2.24e-06,
-      "loss": 7.189,
+      "loss": 7.2287,
       "step": 118
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.27,
       "learning_rate": 2.2600000000000004e-06,
-      "loss": 7.2223,
+      "loss": 7.139,
       "step": 119
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.27,
       "learning_rate": 2.28e-06,
-      "loss": 7.2258,
+      "loss": 7.2388,
       "step": 120
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.27,
       "learning_rate": 2.3000000000000004e-06,
-      "loss": 7.1578,
+      "loss": 7.1634,
       "step": 121
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.27,
       "learning_rate": 2.3200000000000002e-06,
-      "loss": 7.2855,
+      "loss": 7.1059,
       "step": 122
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 2.3400000000000005e-06,
-      "loss": 7.076,
+      "loss": 7.1523,
       "step": 123
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 2.3600000000000003e-06,
-      "loss": 7.2075,
+      "loss": 7.1194,
       "step": 124
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 2.38e-06,
-      "loss": 7.2781,
+      "loss": 7.2067,
       "step": 125
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 2.4000000000000003e-06,
-      "loss": 7.3497,
+      "loss": 7.2303,
       "step": 126
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 2.42e-06,
-      "loss": 7.3957,
+      "loss": 7.2051,
       "step": 127
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.29,
       "learning_rate": 2.4400000000000004e-06,
-      "loss": 7.296,
+      "loss": 7.1689,
       "step": 128
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.29,
       "learning_rate": 2.46e-06,
-      "loss": 7.1256,
+      "loss": 7.2276,
       "step": 129
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 2.4800000000000004e-06,
-      "loss": 7.2404,
+      "loss": 7.2736,
       "step": 130
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 2.5e-06,
-      "loss": 7.2025,
+      "loss": 7.0741,
       "step": 131
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.52e-06,
-      "loss": 7.2939,
+      "loss": 7.2548,
       "step": 132
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.5400000000000002e-06,
-      "loss": 7.2065,
+      "loss": 7.1224,
       "step": 133
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.56e-06,
-      "loss": 7.3444,
+      "loss": 7.1548,
       "step": 134
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.5800000000000003e-06,
-      "loss": 7.0889,
+      "loss": 7.2653,
       "step": 135
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.6e-06,
-      "loss": 7.1157,
+      "loss": 7.0389,
       "step": 136
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 2.6200000000000003e-06,
-      "loss": 7.3834,
+      "loss": 7.2487,
       "step": 137
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 2.64e-06,
-      "loss": 7.2181,
+      "loss": 7.173,
       "step": 138
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 2.6600000000000004e-06,
-      "loss": 7.136,
+      "loss": 7.1693,
       "step": 139
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 2.68e-06,
-      "loss": 7.175,
+      "loss": 7.2814,
       "step": 140
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.7000000000000004e-06,
-      "loss": 7.3844,
+      "loss": 7.2482,
       "step": 141
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.7200000000000002e-06,
-      "loss": 7.3548,
+      "loss": 7.2497,
       "step": 142
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.7400000000000004e-06,
-      "loss": 7.2571,
+      "loss": 7.0239,
       "step": 143
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.7600000000000003e-06,
-      "loss": 7.0886,
+      "loss": 7.1039,
       "step": 144
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 2.7800000000000005e-06,
-      "loss": 7.1203,
+      "loss": 7.0526,
       "step": 145
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 2.8000000000000003e-06,
-      "loss": 7.0545,
+      "loss": 7.0991,
       "step": 146
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 2.82e-06,
-      "loss": 7.3739,
+      "loss": 7.1917,
       "step": 147
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 2.84e-06,
-      "loss": 6.8122,
+      "loss": 7.2096,
       "step": 148
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 2.86e-06,
-      "loss": 7.2062,
+      "loss": 6.9993,
       "step": 149
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 2.88e-06,
-      "loss": 6.794,
+      "loss": 6.9164,
       "step": 150
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 2.9e-06,
-      "loss": 7.1737,
+      "loss": 7.1531,
       "step": 151
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 2.92e-06,
-      "loss": 7.1242,
+      "loss": 7.1666,
       "step": 152
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 2.9400000000000002e-06,
-      "loss": 7.158,
+      "loss": 7.1133,
       "step": 153
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 2.96e-06,
-      "loss": 7.043,
+      "loss": 7.0933,
       "step": 154
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 2.9800000000000003e-06,
-      "loss": 7.0695,
+      "loss": 7.0639,
       "step": 155
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 3e-06,
-      "loss": 7.1494,
+      "loss": 7.1154,
       "step": 156
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 3.0200000000000003e-06,
-      "loss": 7.3168,
+      "loss": 7.1863,
       "step": 157
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 3.04e-06,
-      "loss": 7.276,
+      "loss": 7.1349,
       "step": 158
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.0600000000000003e-06,
-      "loss": 7.2506,
+      "loss": 7.1162,
       "step": 159
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.08e-06,
-      "loss": 7.2188,
+      "loss": 7.1441,
       "step": 160
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.1000000000000004e-06,
-      "loss": 7.0923,
+      "loss": 7.0902,
       "step": 161
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.12e-06,
-      "loss": 7.0838,
+      "loss": 7.1615,
       "step": 162
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 3.1400000000000004e-06,
-      "loss": 7.1382,
+      "loss": 7.1423,
       "step": 163
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 3.1600000000000002e-06,
-      "loss": 7.0863,
+      "loss": 7.1207,
       "step": 164
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 3.1800000000000005e-06,
-      "loss": 7.151,
+      "loss": 7.0631,
       "step": 165
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 3.2000000000000003e-06,
-      "loss": 7.098,
+      "loss": 7.1458,
       "step": 166
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 3.2200000000000005e-06,
-      "loss": 7.0416,
+      "loss": 7.0482,
       "step": 167
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.2400000000000003e-06,
-      "loss": 7.1658,
+      "loss": 7.0045,
       "step": 168
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.2600000000000006e-06,
-      "loss": 7.1738,
+      "loss": 7.0607,
       "step": 169
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.2800000000000004e-06,
-      "loss": 7.0728,
+      "loss": 7.1924,
       "step": 170
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.3000000000000006e-06,
-      "loss": 7.2656,
+      "loss": 7.1058,
       "step": 171
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 3.3200000000000004e-06,
-      "loss": 7.254,
+      "loss": 7.1291,
       "step": 172
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 3.3400000000000006e-06,
-      "loss": 7.2756,
+      "loss": 7.0849,
       "step": 173
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 3.3600000000000004e-06,
-      "loss": 7.138,
+      "loss": 7.1391,
       "step": 174
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 3.3800000000000007e-06,
-      "loss": 7.022,
+      "loss": 7.1101,
       "step": 175
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 3.4000000000000005e-06,
-      "loss": 7.1407,
+      "loss": 7.0369,
       "step": 176
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.4200000000000007e-06,
-      "loss": 7.1017,
+      "loss": 7.0363,
       "step": 177
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.44e-06,
-      "loss": 7.0626,
+      "loss": 7.1012,
       "step": 178
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.46e-06,
-      "loss": 7.1772,
+      "loss": 7.0995,
       "step": 179
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.48e-06,
-      "loss": 7.2003,
+      "loss": 7.1561,
       "step": 180
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 3.5e-06,
-      "loss": 7.1492,
+      "loss": 7.1402,
       "step": 181
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 3.52e-06,
-      "loss": 7.009,
+      "loss": 7.0451,
       "step": 182
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 3.54e-06,
-      "loss": 7.0999,
+      "loss": 7.148,
       "step": 183
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 3.5600000000000002e-06,
-      "loss": 7.2062,
+      "loss": 7.1851,
       "step": 184
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 3.58e-06,
-      "loss": 7.1359,
+      "loss": 7.1082,
       "step": 185
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.6000000000000003e-06,
-      "loss": 7.2602,
+      "loss": 6.9746,
       "step": 186
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.62e-06,
-      "loss": 7.1816,
+      "loss": 7.0704,
       "step": 187
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.6400000000000003e-06,
-      "loss": 6.9805,
+      "loss": 7.0528,
       "step": 188
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.66e-06,
-      "loss": 7.0477,
+      "loss": 7.1322,
       "step": 189
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 3.6800000000000003e-06,
-      "loss": 7.0994,
+      "loss": 7.1972,
       "step": 190
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 3.7e-06,
-      "loss": 7.0917,
+      "loss": 7.2024,
       "step": 191
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 3.7200000000000004e-06,
-      "loss": 7.2496,
+      "loss": 7.0408,
       "step": 192
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 3.74e-06,
-      "loss": 7.2584,
+      "loss": 7.1215,
       "step": 193
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 3.7600000000000004e-06,
-      "loss": 7.1031,
+      "loss": 7.0417,
       "step": 194
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.7800000000000002e-06,
-      "loss": 6.8404,
+      "loss": 7.01,
       "step": 195
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.8000000000000005e-06,
-      "loss": 7.251,
+      "loss": 6.9221,
       "step": 196
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.820000000000001e-06,
-      "loss": 7.0204,
+      "loss": 6.8631,
       "step": 197
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.8400000000000005e-06,
-      "loss": 6.958,
+      "loss": 7.045,
       "step": 198
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 3.86e-06,
-      "loss": 6.8229,
+      "loss": 6.8861,
       "step": 199
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 3.88e-06,
-      "loss": 6.6638,
+      "loss": 6.7337,
       "step": 200
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 3.900000000000001e-06,
-      "loss": 7.1504,
+      "loss": 7.2072,
       "step": 201
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 3.920000000000001e-06,
-      "loss": 7.1799,
+      "loss": 7.1981,
       "step": 202
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 3.94e-06,
-      "loss": 7.0748,
+      "loss": 7.1899,
       "step": 203
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 3.96e-06,
-      "loss": 7.1227,
+      "loss": 7.1453,
       "step": 204
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 3.980000000000001e-06,
-      "loss": 7.0821,
+      "loss": 7.1685,
       "step": 205
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 7.0723,
+      "loss": 7.1609,
       "step": 206
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 4.0200000000000005e-06,
-      "loss": 7.1149,
+      "loss": 7.0495,
       "step": 207
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 4.04e-06,
-      "loss": 7.0503,
+      "loss": 7.187,
       "step": 208
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 4.060000000000001e-06,
-      "loss": 7.038,
+      "loss": 7.1006,
       "step": 209
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 4.08e-06,
-      "loss": 7.0363,
+      "loss": 7.1509,
       "step": 210
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 4.1e-06,
-      "loss": 6.909,
+      "loss": 7.0645,
       "step": 211
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.12e-06,
-      "loss": 6.9881,
+      "loss": 7.1027,
       "step": 212
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.14e-06,
-      "loss": 7.0661,
+      "loss": 6.9497,
       "step": 213
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.16e-06,
-      "loss": 7.1025,
+      "loss": 7.0604,
       "step": 214
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.18e-06,
-      "loss": 7.0462,
+      "loss": 7.032,
       "step": 215
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.2000000000000004e-06,
-      "loss": 7.192,
+      "loss": 7.0713,
       "step": 216
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 4.22e-06,
-      "loss": 7.0445,
+      "loss": 7.059,
       "step": 217
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 4.24e-06,
-      "loss": 7.0708,
+      "loss": 7.0995,
       "step": 218
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 4.26e-06,
-      "loss": 6.9884,
+      "loss": 7.0596,
       "step": 219
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 4.2800000000000005e-06,
-      "loss": 7.1473,
+      "loss": 7.0901,
       "step": 220
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.3e-06,
-      "loss": 7.0784,
+      "loss": 7.1409,
       "step": 221
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.32e-06,
-      "loss": 6.9736,
+      "loss": 7.1144,
       "step": 222
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.34e-06,
-      "loss": 7.1086,
+      "loss": 7.0433,
       "step": 223
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.360000000000001e-06,
-      "loss": 7.0579,
+      "loss": 6.9968,
       "step": 224
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.38e-06,
-      "loss": 7.0152,
+      "loss": 7.1748,
       "step": 225
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 4.4e-06,
-      "loss": 7.2458,
+      "loss": 7.0851,
       "step": 226
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 4.42e-06,
-      "loss": 7.0439,
+      "loss": 7.178,
       "step": 227
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 4.440000000000001e-06,
-      "loss": 7.0338,
+      "loss": 7.1072,
       "step": 228
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 4.4600000000000005e-06,
-      "loss": 7.1478,
+      "loss": 7.0404,
       "step": 229
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.48e-06,
-      "loss": 7.1501,
+      "loss": 7.0697,
       "step": 230
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.5e-06,
-      "loss": 7.0003,
+      "loss": 7.1125,
       "step": 231
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.520000000000001e-06,
-      "loss": 7.1987,
+      "loss": 6.9526,
       "step": 232
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.540000000000001e-06,
-      "loss": 7.0871,
+      "loss": 7.0077,
       "step": 233
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.56e-06,
-      "loss": 7.1093,
+      "loss": 7.0224,
       "step": 234
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 4.58e-06,
-      "loss": 7.1929,
+      "loss": 7.1392,
       "step": 235
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 4.600000000000001e-06,
-      "loss": 6.943,
+      "loss": 7.0185,
       "step": 236
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 4.620000000000001e-06,
-      "loss": 7.0211,
+      "loss": 7.1857,
       "step": 237
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 4.6400000000000005e-06,
-      "loss": 7.2254,
+      "loss": 7.0496,
       "step": 238
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.66e-06,
-      "loss": 7.0068,
+      "loss": 6.9863,
       "step": 239
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.680000000000001e-06,
-      "loss": 7.0653,
+      "loss": 7.2381,
       "step": 240
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.7e-06,
-      "loss": 7.0161,
+      "loss": 7.0243,
       "step": 241
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.7200000000000005e-06,
-      "loss": 7.1127,
+      "loss": 7.0274,
       "step": 242
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.74e-06,
-      "loss": 6.9831,
+      "loss": 7.1441,
       "step": 243
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 4.76e-06,
-      "loss": 6.9056,
+      "loss": 6.9591,
       "step": 244
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 4.78e-06,
-      "loss": 7.0091,
+      "loss": 6.9841,
       "step": 245
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 4.800000000000001e-06,
-      "loss": 7.0381,
+      "loss": 6.9644,
       "step": 246
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 4.8200000000000004e-06,
-      "loss": 7.0298,
+      "loss": 6.7741,
       "step": 247
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.84e-06,
-      "loss": 7.1736,
+      "loss": 7.0463,
       "step": 248
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.86e-06,
-      "loss": 6.9504,
+      "loss": 6.9203,
       "step": 249
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.880000000000001e-06,
-      "loss": 6.6929,
+      "loss": 6.7171,
       "step": 250
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.9000000000000005e-06,
-      "loss": 7.2317,
+      "loss": 7.203,
       "step": 251
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 4.92e-06,
-      "loss": 7.1661,
+      "loss": 7.1808,
       "step": 252
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 4.94e-06,
-      "loss": 7.0993,
+      "loss": 7.1442,
       "step": 253
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 4.960000000000001e-06,
-      "loss": 7.1939,
+      "loss": 7.1738,
       "step": 254
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 4.980000000000001e-06,
-      "loss": 7.0019,
+      "loss": 7.2137,
       "step": 255
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 5e-06,
-      "loss": 7.1824,
+      "loss": 7.1239,
       "step": 256
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.02e-06,
-      "loss": 7.1135,
+      "loss": 7.0826,
       "step": 257
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.04e-06,
-      "loss": 7.099,
+      "loss": 7.1231,
       "step": 258
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.060000000000001e-06,
-      "loss": 7.0016,
+      "loss": 7.0327,
       "step": 259
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.0800000000000005e-06,
-      "loss": 7.1541,
+      "loss": 7.062,
       "step": 260
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 5.1e-06,
-      "loss": 7.0712,
+      "loss": 7.0684,
       "step": 261
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 5.12e-06,
-      "loss": 6.9414,
+      "loss": 7.0394,
       "step": 262
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 5.140000000000001e-06,
-      "loss": 7.1157,
+      "loss": 7.1074,
       "step": 263
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 5.1600000000000006e-06,
-      "loss": 7.0929,
+      "loss": 7.1088,
       "step": 264
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 5.18e-06,
-      "loss": 7.0189,
+      "loss": 7.1337,
       "step": 265
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.2e-06,
-      "loss": 7.1713,
+      "loss": 6.9456,
       "step": 266
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.220000000000001e-06,
-      "loss": 7.1973,
+      "loss": 7.0948,
       "step": 267
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.240000000000001e-06,
-      "loss": 7.1602,
+      "loss": 6.9807,
       "step": 268
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.2600000000000005e-06,
-      "loss": 7.0172,
+      "loss": 7.0879,
       "step": 269
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 5.28e-06,
-      "loss": 7.0955,
+      "loss": 7.1338,
       "step": 270
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 5.300000000000001e-06,
-      "loss": 6.9876,
+      "loss": 7.0202,
       "step": 271
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 5.320000000000001e-06,
-      "loss": 6.9961,
+      "loss": 7.0207,
       "step": 272
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 5.3400000000000005e-06,
-      "loss": 7.0105,
+      "loss": 6.9342,
       "step": 273
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 5.36e-06,
-      "loss": 6.899,
+      "loss": 6.9222,
       "step": 274
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.380000000000001e-06,
-      "loss": 7.1635,
+      "loss": 7.0924,
       "step": 275
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.400000000000001e-06,
-      "loss": 7.0067,
+      "loss": 7.0018,
       "step": 276
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.420000000000001e-06,
-      "loss": 7.1221,
+      "loss": 6.9829,
       "step": 277
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.4400000000000004e-06,
-      "loss": 7.0151,
+      "loss": 7.1071,
       "step": 278
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 5.460000000000001e-06,
-      "loss": 7.1027,
+      "loss": 7.1579,
       "step": 279
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 5.480000000000001e-06,
-      "loss": 7.1135,
+      "loss": 7.1568,
       "step": 280
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 6.9153,
+      "loss": 7.0662,
       "step": 281
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 5.5200000000000005e-06,
-      "loss": 7.1004,
+      "loss": 7.0609,
       "step": 282
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 5.540000000000001e-06,
-      "loss": 6.8542,
+      "loss": 6.9644,
       "step": 283
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.560000000000001e-06,
-      "loss": 7.1294,
+      "loss": 7.0189,
       "step": 284
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.580000000000001e-06,
-      "loss": 7.0684,
+      "loss": 7.059,
       "step": 285
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.600000000000001e-06,
-      "loss": 6.8944,
+      "loss": 7.1024,
       "step": 286
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.620000000000001e-06,
-      "loss": 7.1458,
+      "loss": 7.0941,
       "step": 287
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 5.64e-06,
-      "loss": 7.0512,
+      "loss": 7.0827,
       "step": 288
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 5.66e-06,
-      "loss": 7.1965,
+      "loss": 7.0403,
       "step": 289
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 5.68e-06,
-      "loss": 7.2709,
+      "loss": 7.0994,
       "step": 290
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 5.7e-06,
-      "loss": 7.132,
+      "loss": 7.1633,
       "step": 291
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 5.72e-06,
-      "loss": 7.0172,
+      "loss": 7.0665,
       "step": 292
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.74e-06,
-      "loss": 6.9473,
+      "loss": 6.9603,
       "step": 293
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.76e-06,
-      "loss": 6.9962,
+      "loss": 6.9473,
       "step": 294
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.78e-06,
-      "loss": 6.878,
+      "loss": 6.9004,
       "step": 295
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.8e-06,
-      "loss": 6.9316,
+      "loss": 6.866,
       "step": 296
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 5.82e-06,
-      "loss": 6.9463,
+      "loss": 6.9382,
       "step": 297
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 5.84e-06,
-      "loss": 6.7394,
+      "loss": 7.0233,
       "step": 298
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 5.86e-06,
-      "loss": 6.7593,
+      "loss": 6.6025,
       "step": 299
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 5.8800000000000005e-06,
-      "loss": 6.5637,
+      "loss": 6.6428,
       "step": 300
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 5.9e-06,
-      "loss": 7.1099,
+      "loss": 7.2423,
       "step": 301
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 5.92e-06,
-      "loss": 7.1065,
+      "loss": 7.1925,
       "step": 302
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 5.94e-06,
-      "loss": 7.0142,
+      "loss": 7.2078,
       "step": 303
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 5.9600000000000005e-06,
-      "loss": 7.1323,
+      "loss": 7.0933,
       "step": 304
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 5.98e-06,
-      "loss": 6.9193,
+      "loss": 7.255,
       "step": 305
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 6e-06,
-      "loss": 7.0063,
+      "loss": 7.0256,
       "step": 306
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 6.02e-06,
-      "loss": 7.0331,
+      "loss": 6.9909,
       "step": 307
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 6.040000000000001e-06,
-      "loss": 7.0819,
+      "loss": 7.1496,
       "step": 308
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 6.0600000000000004e-06,
-      "loss": 7.0249,
+      "loss": 6.9642,
       "step": 309
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.08e-06,
-      "loss": 7.1125,
+      "loss": 7.0763,
       "step": 310
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.1e-06,
-      "loss": 7.2098,
+      "loss": 7.0671,
       "step": 311
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.120000000000001e-06,
-      "loss": 7.0713,
+      "loss": 7.0533,
       "step": 312
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.1400000000000005e-06,
-      "loss": 7.1249,
+      "loss": 6.982,
       "step": 313
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.16e-06,
-      "loss": 7.0367,
+      "loss": 6.9866,
       "step": 314
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 6.18e-06,
-      "loss": 6.9228,
+      "loss": 7.1002,
       "step": 315
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 6.200000000000001e-06,
-      "loss": 7.1705,
+      "loss": 7.0481,
       "step": 316
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 6.220000000000001e-06,
-      "loss": 7.0058,
+      "loss": 7.014,
       "step": 317
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 6.24e-06,
-      "loss": 6.9554,
+      "loss": 6.9361,
       "step": 318
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.26e-06,
-      "loss": 6.9348,
+      "loss": 7.0155,
       "step": 319
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.280000000000001e-06,
-      "loss": 7.1256,
+      "loss": 7.0522,
       "step": 320
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.300000000000001e-06,
-      "loss": 7.0572,
+      "loss": 6.9896,
       "step": 321
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.3200000000000005e-06,
-      "loss": 6.939,
+      "loss": 7.0202,
       "step": 322
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.34e-06,
-      "loss": 7.0327,
+      "loss": 7.078,
       "step": 323
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 6.360000000000001e-06,
-      "loss": 7.1016,
+      "loss": 7.0054,
       "step": 324
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 6.380000000000001e-06,
-      "loss": 7.093,
+      "loss": 7.0511,
       "step": 325
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 6.4000000000000006e-06,
-      "loss": 7.0614,
+      "loss": 7.0643,
       "step": 326
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 6.42e-06,
-      "loss": 6.9429,
+      "loss": 7.047,
       "step": 327
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.440000000000001e-06,
-      "loss": 7.0674,
+      "loss": 6.9796,
       "step": 328
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.460000000000001e-06,
-      "loss": 7.1251,
+      "loss": 7.0924,
       "step": 329
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.480000000000001e-06,
-      "loss": 7.0794,
+      "loss": 7.0856,
       "step": 330
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 7.1103,
+      "loss": 7.0372,
       "step": 331
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.520000000000001e-06,
-      "loss": 6.8241,
+      "loss": 6.9479,
       "step": 332
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 6.540000000000001e-06,
-      "loss": 7.0326,
+      "loss": 7.0917,
       "step": 333
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 6.560000000000001e-06,
-      "loss": 7.063,
+      "loss": 7.0307,
       "step": 334
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 6.5800000000000005e-06,
-      "loss": 7.1207,
+      "loss": 7.124,
       "step": 335
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 6.600000000000001e-06,
-      "loss": 6.9549,
+      "loss": 7.0226,
       "step": 336
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.620000000000001e-06,
-      "loss": 6.9118,
+      "loss": 6.9063,
       "step": 337
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.640000000000001e-06,
-      "loss": 7.0833,
+      "loss": 7.0728,
       "step": 338
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.660000000000001e-06,
-      "loss": 7.0792,
+      "loss": 7.0572,
       "step": 339
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.680000000000001e-06,
-      "loss": 7.0288,
+      "loss": 7.0749,
       "step": 340
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.700000000000001e-06,
-      "loss": 7.1238,
+      "loss": 6.969,
       "step": 341
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 6.720000000000001e-06,
-      "loss": 6.9803,
+      "loss": 7.0759,
       "step": 342
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 6.740000000000001e-06,
-      "loss": 6.9616,
+      "loss": 6.9305,
       "step": 343
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 6.760000000000001e-06,
-      "loss": 6.9533,
+      "loss": 6.9073,
       "step": 344
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 6.780000000000001e-06,
-      "loss": 6.8297,
+      "loss": 6.881,
       "step": 345
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.800000000000001e-06,
-      "loss": 6.9362,
+      "loss": 6.9599,
       "step": 346
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.820000000000001e-06,
-      "loss": 6.8641,
+      "loss": 6.8266,
       "step": 347
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.8400000000000014e-06,
-      "loss": 6.9594,
+      "loss": 6.6888,
       "step": 348
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.860000000000001e-06,
-      "loss": 6.8262,
+      "loss": 6.7824,
       "step": 349
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.88e-06,
-      "loss": 6.6507,
+      "loss": 6.5344,
       "step": 350
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 6.9e-06,
-      "loss": 7.0301,
+      "loss": 7.1566,
       "step": 351
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 6.92e-06,
-      "loss": 7.0473,
+      "loss": 7.2001,
       "step": 352
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 6.9400000000000005e-06,
-      "loss": 7.0101,
+      "loss": 7.0835,
       "step": 353
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 6.96e-06,
-      "loss": 6.9832,
+      "loss": 7.023,
       "step": 354
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 6.98e-06,
-      "loss": 7.1293,
+      "loss": 7.0585,
       "step": 355
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7e-06,
-      "loss": 6.9812,
+      "loss": 7.1668,
       "step": 356
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.0200000000000006e-06,
-      "loss": 7.1903,
+      "loss": 7.0,
       "step": 357
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.04e-06,
-      "loss": 6.9624,
+      "loss": 7.0832,
       "step": 358
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.06e-06,
-      "loss": 7.0994,
+      "loss": 6.9888,
       "step": 359
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 7.08e-06,
-      "loss": 6.8535,
+      "loss": 6.9788,
       "step": 360
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 7.100000000000001e-06,
-      "loss": 6.9574,
+      "loss": 7.0177,
       "step": 361
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 7.1200000000000004e-06,
-      "loss": 7.0674,
+      "loss": 6.9336,
       "step": 362
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 7.14e-06,
-      "loss": 6.988,
+      "loss": 6.8755,
       "step": 363
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.16e-06,
-      "loss": 7.0264,
+      "loss": 6.938,
       "step": 364
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.180000000000001e-06,
-      "loss": 6.9866,
+      "loss": 7.0578,
       "step": 365
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.2000000000000005e-06,
-      "loss": 6.9413,
+      "loss": 7.0069,
       "step": 366
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.22e-06,
-      "loss": 6.9779,
+      "loss": 7.1085,
       "step": 367
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 7.24e-06,
-      "loss": 6.9478,
+      "loss": 7.0106,
       "step": 368
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 7.260000000000001e-06,
-      "loss": 6.9141,
+      "loss": 6.9103,
       "step": 369
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 7.280000000000001e-06,
-      "loss": 7.0886,
+      "loss": 6.9784,
       "step": 370
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 7.3e-06,
-      "loss": 6.9941,
+      "loss": 7.0409,
       "step": 371
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 7.32e-06,
-      "loss": 7.1728,
+      "loss": 7.069,
       "step": 372
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.340000000000001e-06,
-      "loss": 6.9864,
+      "loss": 6.9829,
       "step": 373
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.360000000000001e-06,
-      "loss": 7.0761,
+      "loss": 6.8754,
       "step": 374
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.3800000000000005e-06,
-      "loss": 6.8448,
+      "loss": 6.9719,
       "step": 375
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.4e-06,
-      "loss": 7.0047,
+      "loss": 6.9281,
       "step": 376
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 7.420000000000001e-06,
-      "loss": 6.8696,
+      "loss": 6.9857,
       "step": 377
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 7.440000000000001e-06,
-      "loss": 6.9608,
+      "loss": 6.9805,
       "step": 378
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 7.4600000000000006e-06,
-      "loss": 6.9635,
+      "loss": 6.9681,
       "step": 379
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.48e-06,
-      "loss": 7.0253,
+      "loss": 6.962,
       "step": 380
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 7.0607,
+      "loss": 7.0683,
       "step": 381
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.520000000000001e-06,
-      "loss": 6.955,
+      "loss": 7.0167,
       "step": 382
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.540000000000001e-06,
-      "loss": 7.0939,
+      "loss": 7.0039,
       "step": 383
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.5600000000000005e-06,
-      "loss": 7.07,
+      "loss": 6.9635,
       "step": 384
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.58e-06,
-      "loss": 6.9421,
+      "loss": 7.0326,
       "step": 385
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 7.600000000000001e-06,
-      "loss": 6.9391,
+      "loss": 6.9914,
       "step": 386
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 7.620000000000001e-06,
-      "loss": 6.9566,
+      "loss": 6.9522,
       "step": 387
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 7.640000000000001e-06,
-      "loss": 6.9496,
+      "loss": 7.0191,
       "step": 388
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.660000000000001e-06,
-      "loss": 6.9117,
+      "loss": 6.9369,
       "step": 389
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.680000000000001e-06,
-      "loss": 7.1914,
+      "loss": 6.8405,
       "step": 390
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.7e-06,
-      "loss": 7.0119,
+      "loss": 6.9025,
       "step": 391
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.72e-06,
-      "loss": 6.9804,
+      "loss": 6.942,
       "step": 392
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.74e-06,
-      "loss": 6.9426,
+      "loss": 6.8378,
       "step": 393
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.76e-06,
-      "loss": 7.1092,
+      "loss": 6.8714,
       "step": 394
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 7.78e-06,
-      "loss": 6.9243,
+      "loss": 6.8833,
       "step": 395
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 7.800000000000002e-06,
-      "loss": 6.6944,
+      "loss": 6.9241,
       "step": 396
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.820000000000001e-06,
-      "loss": 6.6262,
+      "loss": 6.8455,
       "step": 397
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.840000000000001e-06,
-      "loss": 6.7822,
+      "loss": 6.7152,
       "step": 398
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.860000000000001e-06,
-      "loss": 6.6832,
+      "loss": 6.7795,
       "step": 399
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.88e-06,
-      "loss": 6.4803,
+      "loss": 6.3986,
       "step": 400
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.9e-06,
-      "loss": 7.1814,
+      "loss": 7.1826,
       "step": 401
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.92e-06,
-      "loss": 7.1672,
+      "loss": 7.1695,
       "step": 402
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.94e-06,
-      "loss": 7.0461,
+      "loss": 7.2669,
       "step": 403
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 7.960000000000002e-06,
-      "loss": 7.0087,
+      "loss": 7.1548,
       "step": 404
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 7.980000000000002e-06,
-      "loss": 7.1767,
+      "loss": 6.9942,
       "step": 405
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 6.9908,
+      "loss": 7.041,
       "step": 406
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.020000000000001e-06,
-      "loss": 6.9835,
+      "loss": 7.0299,
       "step": 407
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.040000000000001e-06,
-      "loss": 6.928,
+      "loss": 6.9149,
       "step": 408
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.06e-06,
-      "loss": 7.153,
+      "loss": 6.8887,
       "step": 409
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.08e-06,
-      "loss": 7.0491,
+      "loss": 6.9186,
       "step": 410
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.1e-06,
-      "loss": 7.0002,
+      "loss": 7.0415,
       "step": 411
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.120000000000002e-06,
-      "loss": 6.9823,
+      "loss": 6.9518,
       "step": 412
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 8.14e-06,
-      "loss": 6.9551,
+      "loss": 6.9932,
       "step": 413
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 8.16e-06,
-      "loss": 6.8327,
+      "loss": 6.9789,
       "step": 414
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.18e-06,
-      "loss": 6.879,
+      "loss": 7.0404,
       "step": 415
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.2e-06,
-      "loss": 6.9875,
+      "loss": 7.0041,
       "step": 416
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.220000000000001e-06,
-      "loss": 6.9204,
+      "loss": 6.9786,
       "step": 417
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.24e-06,
-      "loss": 6.9946,
+      "loss": 6.8628,
       "step": 418
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.26e-06,
-      "loss": 6.9319,
+      "loss": 6.9054,
       "step": 419
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.28e-06,
-      "loss": 6.9402,
+      "loss": 6.9641,
       "step": 420
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.3e-06,
-      "loss": 6.9857,
+      "loss": 7.002,
       "step": 421
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 8.32e-06,
-      "loss": 7.0409,
+      "loss": 6.8865,
       "step": 422
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 8.34e-06,
-      "loss": 6.9763,
+      "loss": 6.93,
       "step": 423
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.36e-06,
-      "loss": 6.9611,
+      "loss": 7.0314,
       "step": 424
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.380000000000001e-06,
-      "loss": 6.996,
+      "loss": 6.8996,
       "step": 425
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 7.1048,
+      "loss": 6.9738,
       "step": 426
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.42e-06,
-      "loss": 6.8986,
+      "loss": 6.9319,
       "step": 427
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.44e-06,
-      "loss": 7.0939,
+      "loss": 6.8332,
       "step": 428
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.46e-06,
-      "loss": 6.9146,
+      "loss": 6.8213,
       "step": 429
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.48e-06,
-      "loss": 6.9294,
+      "loss": 6.8614,
       "step": 430
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.97,
       "learning_rate": 8.5e-06,
-      "loss": 7.0619,
+      "loss": 6.9083,
       "step": 431
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.97,
       "learning_rate": 8.52e-06,
-      "loss": 7.0372,
+      "loss": 7.0035,
       "step": 432
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.97,
       "learning_rate": 8.540000000000001e-06,
-      "loss": 6.7141,
+      "loss": 6.8831,
       "step": 433
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.97,
       "learning_rate": 8.560000000000001e-06,
-      "loss": 6.9286,
+      "loss": 6.8566,
       "step": 434
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 8.580000000000001e-06,
-      "loss": 6.8576,
+      "loss": 6.9702,
       "step": 435
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 8.6e-06,
-      "loss": 7.0541,
+      "loss": 6.8319,
       "step": 436
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 8.62e-06,
-      "loss": 6.9776,
+      "loss": 6.8398,
       "step": 437
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 8.64e-06,
-      "loss": 7.0326,
+      "loss": 6.9327,
       "step": 438
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 8.66e-06,
-      "loss": 6.7962,
+      "loss": 6.8757,
       "step": 439
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.99,
       "learning_rate": 8.68e-06,
-      "loss": 6.9775,
+      "loss": 6.8738,
       "step": 440
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.99,
       "learning_rate": 8.700000000000001e-06,
-      "loss": 7.0289,
+      "loss": 6.7065,
       "step": 441
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.99,
       "learning_rate": 8.720000000000001e-06,
-      "loss": 6.8959,
+      "loss": 6.9136,
       "step": 442
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.99,
       "learning_rate": 8.740000000000001e-06,
-      "loss": 6.8555,
+      "loss": 6.5767,
       "step": 443
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 8.76e-06,
-      "loss": 6.9988,
+      "loss": 6.8314,
       "step": 444
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 8.78e-06,
-      "loss": 6.7196,
+      "loss": 6.5081,
       "step": 445
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 8.8e-06,
-      "loss": 6.7692,
+      "loss": 6.3829,
       "step": 446
     },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.82e-06,
-      "loss": 6.7891,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.84e-06,
-      "loss": 6.7611,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.860000000000002e-06,
-      "loss": 6.5288,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.880000000000001e-06,
-      "loss": 6.5406,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.900000000000001e-06,
-      "loss": 6.9948,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.920000000000001e-06,
-      "loss": 7.0961,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.94e-06,
-      "loss": 7.1564,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.96e-06,
-      "loss": 6.9755,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.98e-06,
-      "loss": 7.0177,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9e-06,
-      "loss": 7.0646,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.020000000000002e-06,
-      "loss": 6.9024,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.040000000000002e-06,
-      "loss": 6.9524,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.060000000000001e-06,
-      "loss": 6.9504,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.080000000000001e-06,
-      "loss": 7.0475,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.100000000000001e-06,
-      "loss": 6.8543,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.12e-06,
-      "loss": 6.9814,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.14e-06,
-      "loss": 6.8515,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.16e-06,
-      "loss": 6.8429,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.180000000000002e-06,
-      "loss": 6.9463,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.200000000000002e-06,
-      "loss": 6.9336,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.220000000000002e-06,
-      "loss": 6.8806,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.240000000000001e-06,
-      "loss": 6.9954,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.260000000000001e-06,
-      "loss": 6.8983,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.280000000000001e-06,
-      "loss": 6.997,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.3e-06,
-      "loss": 6.8817,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.32e-06,
-      "loss": 6.9046,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.340000000000002e-06,
-      "loss": 6.8909,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.360000000000002e-06,
-      "loss": 6.8197,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.38e-06,
-      "loss": 6.9758,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.4e-06,
-      "loss": 6.9858,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.42e-06,
-      "loss": 7.0484,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.440000000000001e-06,
-      "loss": 6.8071,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.460000000000001e-06,
-      "loss": 6.8754,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.48e-06,
-      "loss": 6.8545,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.5e-06,
-      "loss": 6.8623,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.52e-06,
-      "loss": 6.8574,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.54e-06,
-      "loss": 6.858,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.56e-06,
-      "loss": 6.9532,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.58e-06,
-      "loss": 6.9293,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 6.8754,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.620000000000001e-06,
-      "loss": 6.9566,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.640000000000001e-06,
-      "loss": 6.8534,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.66e-06,
-      "loss": 7.0255,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.68e-06,
-      "loss": 7.0672,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.7e-06,
-      "loss": 6.8342,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.72e-06,
-      "loss": 6.9103,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.74e-06,
-      "loss": 6.8983,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.760000000000001e-06,
-      "loss": 6.8642,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 6.9534,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 6.7265,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.820000000000001e-06,
-      "loss": 6.4172,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.84e-06,
-      "loss": 7.0219,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.86e-06,
-      "loss": 6.8133,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.88e-06,
-      "loss": 6.4267,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 6.822406768798828,
-      "eval_runtime": 447.9918,
-      "eval_samples_per_second": 5.897,
-      "eval_steps_per_second": 0.739,
-      "eval_wer": 0.9493819076532334,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.9e-06,
-      "loss": 7.1911,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.920000000000002e-06,
-      "loss": 6.8386,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.940000000000001e-06,
-      "loss": 7.0032,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 7.0327,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.980000000000001e-06,
-      "loss": 7.0476,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1e-05,
-      "loss": 7.0127,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.974489795918369e-06,
-      "loss": 6.9811,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.948979591836737e-06,
-      "loss": 6.8669,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.923469387755103e-06,
-      "loss": 6.9249,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.89795918367347e-06,
-      "loss": 6.973,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.872448979591838e-06,
-      "loss": 6.8868,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.846938775510205e-06,
-      "loss": 6.927,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.821428571428573e-06,
-      "loss": 7.0246,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.795918367346939e-06,
-      "loss": 7.0053,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.770408163265307e-06,
-      "loss": 6.978,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.744897959183674e-06,
-      "loss": 6.7923,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.719387755102042e-06,
-      "loss": 6.9865,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.693877551020408e-06,
-      "loss": 6.8256,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.668367346938776e-06,
-      "loss": 7.0044,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.642857142857144e-06,
-      "loss": 6.9858,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.617346938775512e-06,
-      "loss": 6.7904,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.591836734693878e-06,
-      "loss": 6.7929,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.566326530612246e-06,
-      "loss": 6.719,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.540816326530612e-06,
-      "loss": 6.8147,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.51530612244898e-06,
-      "loss": 6.9299,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.489795918367348e-06,
-      "loss": 6.7332,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.464285714285714e-06,
-      "loss": 6.8183,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.438775510204082e-06,
-      "loss": 6.8812,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.41326530612245e-06,
-      "loss": 6.9139,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.387755102040818e-06,
-      "loss": 6.944,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.362244897959184e-06,
-      "loss": 6.8158,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.336734693877552e-06,
-      "loss": 6.739,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.31122448979592e-06,
-      "loss": 6.7184,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.285714285714288e-06,
-      "loss": 6.6897,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.260204081632654e-06,
-      "loss": 6.7133,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.234693877551022e-06,
-      "loss": 6.888,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.209183673469388e-06,
-      "loss": 6.803,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.183673469387756e-06,
-      "loss": 6.8851,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.158163265306124e-06,
-      "loss": 6.8424,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.13265306122449e-06,
-      "loss": 6.7896,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.107142857142858e-06,
-      "loss": 6.9052,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.081632653061225e-06,
-      "loss": 6.8106,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.056122448979593e-06,
-      "loss": 6.6204,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.03061224489796e-06,
-      "loss": 6.5529,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.005102040816327e-06,
-      "loss": 6.6243,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.979591836734695e-06,
-      "loss": 6.6069,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.954081632653061e-06,
-      "loss": 6.8388,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.92857142857143e-06,
-      "loss": 6.6985,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.903061224489795e-06,
-      "loss": 6.3144,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.877551020408163e-06,
-      "loss": 6.6693,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.852040816326531e-06,
-      "loss": 6.6779,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.826530612244899e-06,
-      "loss": 6.863,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.801020408163265e-06,
-      "loss": 6.6801,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.775510204081633e-06,
-      "loss": 6.6504,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.750000000000001e-06,
-      "loss": 6.6853,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.724489795918369e-06,
-      "loss": 6.6195,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.698979591836737e-06,
-      "loss": 6.5403,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.673469387755103e-06,
-      "loss": 6.6437,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.64795918367347e-06,
-      "loss": 6.6095,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.622448979591837e-06,
-      "loss": 6.4877,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.596938775510205e-06,
-      "loss": 6.6462,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.571428571428571e-06,
-      "loss": 6.4643,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.545918367346939e-06,
-      "loss": 6.5061,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.520408163265307e-06,
-      "loss": 6.5157,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.494897959183675e-06,
-      "loss": 6.683,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.469387755102042e-06,
-      "loss": 6.3558,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.443877551020409e-06,
-      "loss": 6.5497,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.418367346938776e-06,
-      "loss": 6.425,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.392857142857144e-06,
-      "loss": 6.4337,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.36734693877551e-06,
-      "loss": 6.5868,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.341836734693878e-06,
-      "loss": 6.3913,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.316326530612246e-06,
-      "loss": 6.6029,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.290816326530612e-06,
-      "loss": 6.3791,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.26530612244898e-06,
-      "loss": 6.2576,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.239795918367348e-06,
-      "loss": 6.4795,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.214285714285714e-06,
-      "loss": 6.3123,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.188775510204082e-06,
-      "loss": 6.425,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.16326530612245e-06,
-      "loss": 6.4618,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.137755102040818e-06,
-      "loss": 6.2996,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.112244897959184e-06,
-      "loss": 6.5554,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.086734693877552e-06,
-      "loss": 6.5343,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.06122448979592e-06,
-      "loss": 6.3481,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.035714285714286e-06,
-      "loss": 6.2527,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.010204081632654e-06,
-      "loss": 6.3562,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.98469387755102e-06,
-      "loss": 6.4252,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.959183673469388e-06,
-      "loss": 6.4332,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.933673469387756e-06,
-      "loss": 6.4452,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.908163265306124e-06,
-      "loss": 6.3556,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.88265306122449e-06,
-      "loss": 6.2002,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.857142857142858e-06,
-      "loss": 6.2415,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.831632653061226e-06,
-      "loss": 6.3251,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.806122448979593e-06,
-      "loss": 6.3577,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.78061224489796e-06,
-      "loss": 6.4383,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.755102040816327e-06,
-      "loss": 6.2113,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.729591836734695e-06,
-      "loss": 6.369,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.704081632653061e-06,
-      "loss": 6.174,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.67857142857143e-06,
-      "loss": 6.2996,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.653061224489796e-06,
-      "loss": 6.2867,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.627551020408163e-06,
-      "loss": 6.2394,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.602040816326531e-06,
-      "loss": 6.1727,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.576530612244899e-06,
-      "loss": 6.4233,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.551020408163265e-06,
-      "loss": 6.1755,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.525510204081633e-06,
-      "loss": 6.2927,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 6.2481,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.474489795918368e-06,
-      "loss": 6.285,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.448979591836736e-06,
-      "loss": 5.9824,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.423469387755102e-06,
-      "loss": 6.2035,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.39795918367347e-06,
-      "loss": 6.2667,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.372448979591838e-06,
-      "loss": 6.2396,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.346938775510205e-06,
-      "loss": 6.242,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.321428571428572e-06,
-      "loss": 6.1696,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.295918367346939e-06,
-      "loss": 6.2707,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.270408163265307e-06,
-      "loss": 6.1647,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.244897959183675e-06,
-      "loss": 6.1739,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.219387755102042e-06,
-      "loss": 6.3845,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.193877551020409e-06,
-      "loss": 6.0286,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.168367346938776e-06,
-      "loss": 6.0793,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.1428571428571436e-06,
-      "loss": 6.0592,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.1173469387755114e-06,
-      "loss": 6.1299,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.091836734693878e-06,
-      "loss": 6.1894,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.0663265306122455e-06,
-      "loss": 6.2057,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.0408163265306125e-06,
-      "loss": 6.079,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.01530612244898e-06,
-      "loss": 6.1331,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.989795918367348e-06,
-      "loss": 5.968,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.964285714285714e-06,
-      "loss": 6.2897,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.938775510204082e-06,
-      "loss": 6.1582,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.91326530612245e-06,
-      "loss": 6.2631,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.887755102040817e-06,
-      "loss": 6.2091,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.862244897959184e-06,
-      "loss": 6.1292,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.836734693877551e-06,
-      "loss": 6.3536,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.811224489795919e-06,
-      "loss": 6.0919,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.785714285714287e-06,
-      "loss": 6.0478,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.760204081632653e-06,
-      "loss": 6.1582,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.734693877551021e-06,
-      "loss": 6.024,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.709183673469388e-06,
-      "loss": 6.1876,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.683673469387756e-06,
-      "loss": 6.1385,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.658163265306124e-06,
-      "loss": 6.1735,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.63265306122449e-06,
-      "loss": 5.9006,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.607142857142858e-06,
-      "loss": 6.2276,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.581632653061225e-06,
-      "loss": 5.9284,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.556122448979593e-06,
-      "loss": 6.0407,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.530612244897959e-06,
-      "loss": 6.3591,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.505102040816327e-06,
-      "loss": 6.1123,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.4795918367346946e-06,
-      "loss": 5.9066,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.454081632653062e-06,
-      "loss": 5.9831,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.4285714285714295e-06,
-      "loss": 6.1945,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.403061224489796e-06,
-      "loss": 5.9348,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.3775510204081635e-06,
-      "loss": 5.8005,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.352040816326531e-06,
-      "loss": 6.0812,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.326530612244899e-06,
-      "loss": 6.1251,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.301020408163265e-06,
-      "loss": 6.1855,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.275510204081633e-06,
-      "loss": 6.1263,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.25e-06,
-      "loss": 6.0457,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.224489795918368e-06,
-      "loss": 6.0228,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.198979591836736e-06,
-      "loss": 6.0734,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.173469387755102e-06,
-      "loss": 6.0462,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.14795918367347e-06,
-      "loss": 5.8885,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.122448979591837e-06,
-      "loss": 6.2115,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.096938775510205e-06,
-      "loss": 5.9691,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.071428571428571e-06,
-      "loss": 5.9754,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.045918367346939e-06,
-      "loss": 5.8972,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.020408163265307e-06,
-      "loss": 5.9596,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.994897959183674e-06,
-      "loss": 6.1242,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.969387755102042e-06,
-      "loss": 5.8532,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.943877551020408e-06,
-      "loss": 5.9176,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.918367346938776e-06,
-      "loss": 6.0338,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.892857142857144e-06,
-      "loss": 5.9436,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.867346938775511e-06,
-      "loss": 5.9501,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.841836734693878e-06,
-      "loss": 6.0369,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.816326530612246e-06,
-      "loss": 5.9592,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.790816326530613e-06,
-      "loss": 5.8014,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.7653061224489805e-06,
-      "loss": 5.9577,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.739795918367348e-06,
-      "loss": 5.8901,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.7142857142857145e-06,
-      "loss": 5.8914,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.688775510204082e-06,
-      "loss": 5.896,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.663265306122449e-06,
-      "loss": 6.0052,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.637755102040817e-06,
-      "loss": 5.8798,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.6122448979591834e-06,
-      "loss": 5.7341,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.586734693877551e-06,
-      "loss": 5.8964,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.561224489795919e-06,
-      "loss": 5.8531,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.535714285714286e-06,
-      "loss": 6.0326,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.510204081632653e-06,
-      "loss": 5.8104,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.48469387755102e-06,
-      "loss": 5.8037,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.459183673469388e-06,
-      "loss": 5.9225,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.433673469387756e-06,
-      "loss": 5.9709,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.408163265306123e-06,
-      "loss": 5.7156,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.38265306122449e-06,
-      "loss": 5.8575,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.357142857142857e-06,
-      "loss": 6.0424,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.331632653061225e-06,
-      "loss": 5.8257,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.306122448979593e-06,
-      "loss": 5.992,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.280612244897959e-06,
-      "loss": 5.7803,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.255102040816327e-06,
-      "loss": 5.8612,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.229591836734695e-06,
-      "loss": 5.76,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.204081632653062e-06,
-      "loss": 6.0264,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1785714285714296e-06,
-      "loss": 5.8621,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.153061224489796e-06,
-      "loss": 5.9304,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.127551020408164e-06,
-      "loss": 5.8043,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1020408163265315e-06,
-      "loss": 5.6251,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0765306122448985e-06,
-      "loss": 5.9134,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0510204081632655e-06,
-      "loss": 5.7145,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 5.0255102040816325e-06,
-      "loss": 5.9505,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 5e-06,
-      "loss": 5.9508,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.974489795918368e-06,
-      "loss": 5.7676,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.948979591836735e-06,
-      "loss": 5.7054,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.923469387755102e-06,
-      "loss": 5.7923,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.897959183673469e-06,
-      "loss": 5.9246,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.872448979591837e-06,
-      "loss": 5.8495,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.846938775510204e-06,
-      "loss": 6.0147,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.821428571428572e-06,
-      "loss": 5.7805,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.795918367346939e-06,
-      "loss": 5.7268,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.770408163265306e-06,
-      "loss": 5.7298,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.744897959183674e-06,
-      "loss": 5.6547,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.719387755102041e-06,
-      "loss": 5.7235,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.693877551020409e-06,
-      "loss": 5.853,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.668367346938776e-06,
-      "loss": 5.8477,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.642857142857144e-06,
-      "loss": 5.8202,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.617346938775511e-06,
-      "loss": 6.043,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.591836734693878e-06,
-      "loss": 5.8297,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.566326530612245e-06,
-      "loss": 5.5851,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.540816326530613e-06,
-      "loss": 5.8014,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.51530612244898e-06,
-      "loss": 5.7181,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.489795918367348e-06,
-      "loss": 5.647,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.464285714285715e-06,
-      "loss": 5.6772,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.438775510204082e-06,
-      "loss": 5.8078,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.4132653061224495e-06,
-      "loss": 5.7443,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.3877551020408165e-06,
-      "loss": 5.9123,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.362244897959184e-06,
-      "loss": 5.82,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.336734693877551e-06,
-      "loss": 5.6459,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.3112244897959184e-06,
-      "loss": 5.6433,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.2857142857142855e-06,
-      "loss": 5.8695,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.260204081632653e-06,
-      "loss": 5.9662,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.234693877551021e-06,
-      "loss": 5.854,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.209183673469388e-06,
-      "loss": 5.7486,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.183673469387755e-06,
-      "loss": 5.8271,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.158163265306123e-06,
-      "loss": 5.6711,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.13265306122449e-06,
-      "loss": 5.8048,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.107142857142857e-06,
-      "loss": 5.8217,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.081632653061225e-06,
-      "loss": 5.8255,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.056122448979592e-06,
-      "loss": 5.7545,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.03061224489796e-06,
-      "loss": 5.8673,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.005102040816327e-06,
-      "loss": 5.9113,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.979591836734694e-06,
-      "loss": 5.884,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.954081632653062e-06,
-      "loss": 5.7103,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.928571428571429e-06,
-      "loss": 5.7773,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.903061224489797e-06,
-      "loss": 5.7214,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.877551020408164e-06,
-      "loss": 6.0602,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.852040816326531e-06,
-      "loss": 5.6094,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.826530612244898e-06,
-      "loss": 5.9302,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.8010204081632656e-06,
-      "loss": 5.9017,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7755102040816327e-06,
-      "loss": 5.8448,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7500000000000005e-06,
-      "loss": 5.798,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.724489795918368e-06,
-      "loss": 5.8921,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.698979591836735e-06,
-      "loss": 5.7987,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.6734693877551024e-06,
-      "loss": 5.6745,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.6479591836734694e-06,
-      "loss": 5.7771,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.6224489795918373e-06,
-      "loss": 5.8616,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5969387755102043e-06,
-      "loss": 5.6304,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5714285714285718e-06,
-      "loss": 5.7058,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.545918367346939e-06,
-      "loss": 5.6406,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5204081632653062e-06,
-      "loss": 5.6383,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.494897959183674e-06,
-      "loss": 5.7418,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.469387755102041e-06,
-      "loss": 5.6793,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.4438775510204086e-06,
-      "loss": 5.5261,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.4183673469387756e-06,
-      "loss": 5.5328,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.3928571428571435e-06,
-      "loss": 5.6617,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.3673469387755105e-06,
-      "loss": 5.6348,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.341836734693878e-06,
-      "loss": 5.6628,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.316326530612245e-06,
-      "loss": 5.6829,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2908163265306124e-06,
-      "loss": 5.7563,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2653061224489794e-06,
-      "loss": 5.5559,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2397959183673473e-06,
-      "loss": 5.7773,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.2142857142857147e-06,
-      "loss": 5.7777,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.1887755102040818e-06,
-      "loss": 5.7062,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.1632653061224496e-06,
-      "loss": 5.5077,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.1377551020408166e-06,
-      "loss": 5.5002,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.112244897959184e-06,
-      "loss": 5.6265,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.086734693877551e-06,
-      "loss": 5.4815,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0612244897959185e-06,
-      "loss": 5.7339,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0357142857142856e-06,
-      "loss": 5.6302,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0102040816326534e-06,
-      "loss": 5.5082,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.984693877551021e-06,
-      "loss": 5.6323,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.959183673469388e-06,
-      "loss": 5.5845,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.9336734693877553e-06,
-      "loss": 5.6786,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.908163265306123e-06,
-      "loss": 5.5859,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8826530612244902e-06,
-      "loss": 5.5684,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8571428571428573e-06,
-      "loss": 5.6191,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8316326530612247e-06,
-      "loss": 5.7101,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8061224489795917e-06,
-      "loss": 5.6566,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7806122448979596e-06,
-      "loss": 5.6676,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.7551020408163266e-06,
-      "loss": 5.6757,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.729591836734694e-06,
-      "loss": 5.582,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.7040816326530615e-06,
-      "loss": 5.4637,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6785714285714285e-06,
-      "loss": 5.5952,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6530612244897964e-06,
-      "loss": 5.6437,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6275510204081634e-06,
-      "loss": 5.8045,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.602040816326531e-06,
-      "loss": 5.6708,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.576530612244898e-06,
-      "loss": 5.8274,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5510204081632657e-06,
-      "loss": 5.674,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.5255102040816328e-06,
-      "loss": 6.0943,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.5e-06,
-      "loss": 5.6152,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.4744897959183676e-06,
-      "loss": 5.7035,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.4489795918367347e-06,
-      "loss": 5.6179,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.423469387755102e-06,
-      "loss": 5.7335,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3979591836734696e-06,
-      "loss": 5.6298,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.372448979591837e-06,
-      "loss": 5.4279,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3469387755102044e-06,
-      "loss": 5.6543,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.321428571428572e-06,
-      "loss": 5.6383,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.295918367346939e-06,
-      "loss": 5.5104,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.2704081632653064e-06,
-      "loss": 5.6737,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.244897959183674e-06,
-      "loss": 5.5356,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.219387755102041e-06,
-      "loss": 5.7576,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1938775510204083e-06,
-      "loss": 5.6626,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1683673469387757e-06,
-      "loss": 5.5138,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1428571428571427e-06,
-      "loss": 5.5147,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1173469387755106e-06,
-      "loss": 5.6588,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0918367346938776e-06,
-      "loss": 5.5035,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 2.066326530612245e-06,
-      "loss": 5.6381,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 2.0408163265306125e-06,
-      "loss": 5.5644,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 2.01530612244898e-06,
-      "loss": 5.5479,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.989795918367347e-06,
-      "loss": 5.6088,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.9642857142857144e-06,
-      "loss": 5.5956,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.938775510204082e-06,
-      "loss": 5.5904,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.913265306122449e-06,
-      "loss": 5.5847,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8877551020408163e-06,
-      "loss": 5.5374,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.862244897959184e-06,
-      "loss": 5.7972,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.8367346938775512e-06,
-      "loss": 5.7153,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.8112244897959187e-06,
-      "loss": 5.5736,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7857142857142859e-06,
-      "loss": 5.5445,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7602040816326531e-06,
-      "loss": 5.536,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7346938775510206e-06,
-      "loss": 5.6138,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7091836734693878e-06,
-      "loss": 5.5778,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6836734693877552e-06,
-      "loss": 5.5391,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6581632653061225e-06,
-      "loss": 5.6355,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6326530612244897e-06,
-      "loss": 5.5561,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.6071428571428574e-06,
-      "loss": 5.6282,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5816326530612248e-06,
-      "loss": 5.7254,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.556122448979592e-06,
-      "loss": 5.6199,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5306122448979593e-06,
-      "loss": 5.604,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5051020408163267e-06,
-      "loss": 5.4757,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.479591836734694e-06,
-      "loss": 5.515,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4540816326530614e-06,
-      "loss": 5.8493,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4285714285714286e-06,
-      "loss": 5.6032,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.4030612244897959e-06,
-      "loss": 5.8956,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3775510204081633e-06,
-      "loss": 5.6506,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3520408163265307e-06,
-      "loss": 5.7816,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3265306122448982e-06,
-      "loss": 5.8381,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3010204081632654e-06,
-      "loss": 5.6069,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2755102040816329e-06,
-      "loss": 6.087,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.25e-06,
-      "loss": 5.8074,
-      "step": 849
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2244897959183673e-06,
-      "loss": 5.5123,
-      "step": 850
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1989795918367348e-06,
-      "loss": 5.5092,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.1734693877551022e-06,
-      "loss": 5.5448,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.1479591836734695e-06,
-      "loss": 5.6363,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.122448979591837e-06,
-      "loss": 5.5432,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0969387755102041e-06,
-      "loss": 5.5898,
-      "step": 855
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0714285714285714e-06,
-      "loss": 5.6369,
-      "step": 856
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0459183673469388e-06,
-      "loss": 5.7121,
-      "step": 857
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0204081632653063e-06,
-      "loss": 5.58,
-      "step": 858
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.948979591836735e-07,
-      "loss": 5.4047,
-      "step": 859
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.69387755102041e-07,
-      "loss": 5.6375,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 9.438775510204082e-07,
-      "loss": 5.7298,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 9.183673469387756e-07,
-      "loss": 5.637,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.928571428571429e-07,
-      "loss": 5.583,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.673469387755103e-07,
-      "loss": 5.6558,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.418367346938776e-07,
-      "loss": 5.663,
-      "step": 865
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.163265306122449e-07,
-      "loss": 5.5068,
-      "step": 866
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.908163265306124e-07,
-      "loss": 5.5933,
-      "step": 867
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.653061224489796e-07,
-      "loss": 5.5118,
-      "step": 868
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.39795918367347e-07,
-      "loss": 5.5633,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 7.142857142857143e-07,
-      "loss": 5.5898,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.887755102040817e-07,
-      "loss": 5.5085,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.632653061224491e-07,
-      "loss": 5.583,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.377551020408164e-07,
-      "loss": 5.4993,
-      "step": 873
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.122448979591837e-07,
-      "loss": 5.6236,
-      "step": 874
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.867346938775511e-07,
-      "loss": 5.6637,
-      "step": 875
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.612244897959184e-07,
-      "loss": 5.4688,
-      "step": 876
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.357142857142857e-07,
-      "loss": 5.4333,
-      "step": 877
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.102040816326531e-07,
-      "loss": 5.5523,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.846938775510205e-07,
-      "loss": 5.6512,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.591836734693878e-07,
-      "loss": 5.848,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.3367346938775514e-07,
-      "loss": 5.4165,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.0816326530612243e-07,
-      "loss": 5.5008,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.826530612244898e-07,
-      "loss": 5.5307,
-      "step": 883
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.5714285714285716e-07,
-      "loss": 5.6546,
-      "step": 884
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.3163265306122455e-07,
-      "loss": 5.5449,
-      "step": 885
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.0612244897959183e-07,
-      "loss": 5.6176,
-      "step": 886
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.806122448979592e-07,
-      "loss": 5.3798,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.5510204081632656e-07,
-      "loss": 5.6492,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.295918367346939e-07,
-      "loss": 5.6348,
-      "step": 889
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.0408163265306121e-07,
-      "loss": 5.4948,
-      "step": 890
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.7857142857142858e-07,
-      "loss": 5.713,
-      "step": 891
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.5306122448979592e-07,
-      "loss": 5.4356,
-      "step": 892
-    },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 6.699045791754274,
-      "train_runtime": 5652.3105,
-      "train_samples_per_second": 5.049,
-      "train_steps_per_second": 0.158
+      "train_loss": 7.25599230030727,
+      "train_runtime": 4990.6204,
+      "train_samples_per_second": 5.718,
+      "train_steps_per_second": 0.089
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,