diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
-  "global_step": 446,
+  "epoch": 0.9995796553173603,
+  "global_step": 1189,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,2690 +10,7166 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.0656,
+      "loss": 9.1423,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.6633,
+      "loss": 8.7113,
       "step": 2
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 2e-08,
-      "loss": 8.9119,
+      "loss": 9.1739,
       "step": 3
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 2e-08,
-      "loss": 8.8743,
+      "loss": 9.1294,
       "step": 4
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4e-08,
-      "loss": 8.7322,
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 8.799,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 8.8432,
+      "learning_rate": 4e-08,
+      "loss": 8.8979,
       "step": 6
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 6.000000000000001e-08,
-      "loss": 8.7598,
+      "loss": 8.7477,
       "step": 7
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 8e-08,
-      "loss": 8.7179,
+      "loss": 8.6932,
       "step": 8
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 1.0000000000000001e-07,
-      "loss": 8.7009,
+      "loss": 8.9655,
       "step": 9
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 8.8273,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 8.8356,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 8.6765,
+      "epoch": 0.01,
+      "learning_rate": 1.4e-07,
+      "loss": 8.7868,
       "step": 11
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 8.7786,
+      "epoch": 0.01,
+      "learning_rate": 1.6e-07,
+      "loss": 8.9447,
       "step": 12
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.4e-07,
-      "loss": 8.6871,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 8.7593,
       "step": 13
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.6e-07,
-      "loss": 8.8664,
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 8.6256,
       "step": 14
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.8e-07,
-      "loss": 8.658,
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 8.9289,
       "step": 15
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 8.8457,
+      "epoch": 0.01,
+      "learning_rate": 2.2e-07,
+      "loss": 8.4827,
       "step": 16
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.2e-07,
-      "loss": 8.5633,
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 8.8611,
       "step": 17
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 8.6972,
+      "epoch": 0.02,
+      "learning_rate": 2.6e-07,
+      "loss": 9.1329,
       "step": 18
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.6e-07,
-      "loss": 8.7013,
+      "epoch": 0.02,
+      "learning_rate": 2.8e-07,
+      "loss": 8.6605,
       "step": 19
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.8e-07,
-      "loss": 8.6702,
+      "epoch": 0.02,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 8.7434,
       "step": 20
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 8.9022,
+      "epoch": 0.02,
+      "learning_rate": 3.2e-07,
+      "loss": 8.4078,
       "step": 21
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 3.2e-07,
-      "loss": 8.8576,
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 8.3026,
       "step": 22
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 8.7779,
+      "epoch": 0.02,
+      "learning_rate": 3.6e-07,
+      "loss": 8.763,
       "step": 23
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 3.6e-07,
-      "loss": 8.834,
+      "loss": 8.7404,
       "step": 24
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 3.8e-07,
-      "loss": 8.5806,
+      "loss": 8.4678,
       "step": 25
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 4.0000000000000003e-07,
-      "loss": 8.4857,
+      "loss": 8.5886,
       "step": 26
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 4.2000000000000006e-07,
-      "loss": 8.2719,
+      "loss": 8.6279,
       "step": 27
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 4.4e-07,
-      "loss": 8.3136,
+      "loss": 8.4111,
       "step": 28
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.02,
       "learning_rate": 4.6000000000000004e-07,
-      "loss": 8.5229,
+      "loss": 8.4223,
       "step": 29
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 4.800000000000001e-07,
-      "loss": 8.2262,
+      "loss": 8.483,
       "step": 30
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 8.3328,
+      "loss": 8.5132,
       "step": 31
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 5.2e-07,
-      "loss": 8.2745,
+      "loss": 8.0765,
       "step": 32
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 5.4e-07,
-      "loss": 8.1359,
+      "loss": 8.5574,
       "step": 33
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 5.6e-07,
-      "loss": 8.2812,
+      "loss": 8.1588,
       "step": 34
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 5.800000000000001e-07,
-      "loss": 8.2066,
+      "loss": 9.2147,
       "step": 35
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 6.000000000000001e-07,
-      "loss": 8.2059,
+      "loss": 8.375,
       "step": 36
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 6.200000000000001e-07,
-      "loss": 8.1422,
+      "loss": 8.3187,
       "step": 37
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 6.4e-07,
-      "loss": 8.1902,
+      "loss": 8.2902,
       "step": 38
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 6.6e-07,
-      "loss": 8.1267,
+      "loss": 8.4031,
       "step": 39
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 6.800000000000001e-07,
-      "loss": 8.2181,
+      "loss": 8.0942,
       "step": 40
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 7.000000000000001e-07,
-      "loss": 8.0245,
+      "loss": 8.2915,
       "step": 41
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.04,
       "learning_rate": 7.2e-07,
-      "loss": 8.1449,
+      "loss": 8.5093,
       "step": 42
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 7.4e-07,
-      "loss": 7.998,
+      "loss": 8.2665,
       "step": 43
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 7.6e-07,
-      "loss": 8.1898,
+      "loss": 8.1763,
       "step": 44
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 7.8e-07,
-      "loss": 8.2973,
+      "loss": 8.2778,
       "step": 45
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 8.000000000000001e-07,
-      "loss": 8.1341,
+      "loss": 8.3132,
       "step": 46
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 8.200000000000001e-07,
-      "loss": 8.1824,
+      "loss": 8.2355,
       "step": 47
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 8.400000000000001e-07,
-      "loss": 8.4063,
+      "loss": 8.3881,
       "step": 48
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 8.6e-07,
-      "loss": 8.26,
+      "loss": 8.4433,
       "step": 49
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 8.8e-07,
-      "loss": 8.3837,
+      "loss": 8.5924,
       "step": 50
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 9.000000000000001e-07,
-      "loss": 7.823,
+      "loss": 8.123,
       "step": 51
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.04,
       "learning_rate": 9.200000000000001e-07,
-      "loss": 7.7067,
+      "loss": 7.8029,
       "step": 52
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.04,
       "learning_rate": 9.400000000000001e-07,
-      "loss": 7.8324,
+      "loss": 7.9512,
       "step": 53
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.05,
       "learning_rate": 9.600000000000001e-07,
-      "loss": 7.7619,
+      "loss": 7.7151,
       "step": 54
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.05,
       "learning_rate": 9.800000000000001e-07,
-      "loss": 7.5628,
+      "loss": 7.9318,
       "step": 55
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 7.6529,
+      "loss": 7.6555,
       "step": 56
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.02e-06,
-      "loss": 7.6872,
+      "loss": 8.0422,
       "step": 57
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.04e-06,
-      "loss": 7.7812,
+      "loss": 7.7045,
       "step": 58
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.06e-06,
-      "loss": 7.6103,
+      "loss": 7.7043,
       "step": 59
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.08e-06,
-      "loss": 7.5827,
+      "loss": 7.6798,
       "step": 60
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.1e-06,
-      "loss": 7.5597,
+      "loss": 7.5884,
       "step": 61
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.12e-06,
-      "loss": 7.5961,
+      "loss": 7.5546,
       "step": 62
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.14e-06,
-      "loss": 7.521,
+      "loss": 7.7126,
       "step": 63
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.1600000000000001e-06,
-      "loss": 7.5574,
+      "loss": 7.6394,
       "step": 64
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.05,
       "learning_rate": 1.1800000000000001e-06,
-      "loss": 7.5383,
+      "loss": 7.5483,
       "step": 65
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.2000000000000002e-06,
-      "loss": 7.5139,
+      "loss": 7.7911,
       "step": 66
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.2200000000000002e-06,
-      "loss": 7.4725,
+      "loss": 7.6219,
       "step": 67
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.2400000000000002e-06,
-      "loss": 7.4597,
+      "loss": 7.468,
       "step": 68
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.26e-06,
-      "loss": 7.5113,
+      "loss": 7.5091,
       "step": 69
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.28e-06,
-      "loss": 7.4046,
+      "loss": 7.4986,
       "step": 70
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.3e-06,
-      "loss": 7.5338,
+      "loss": 7.7941,
       "step": 71
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.32e-06,
-      "loss": 7.513,
+      "loss": 7.7057,
       "step": 72
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.34e-06,
-      "loss": 7.4987,
+      "loss": 7.508,
       "step": 73
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.3600000000000001e-06,
-      "loss": 7.5031,
+      "loss": 7.714,
       "step": 74
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.3800000000000001e-06,
-      "loss": 7.4272,
+      "loss": 7.3634,
       "step": 75
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 7.4681,
+      "loss": 7.849,
       "step": 76
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.42e-06,
-      "loss": 7.4724,
+      "loss": 7.5857,
       "step": 77
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.07,
       "learning_rate": 1.44e-06,
-      "loss": 7.4617,
+      "loss": 7.2865,
       "step": 78
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.46e-06,
-      "loss": 7.3632,
+      "loss": 7.4498,
       "step": 79
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.48e-06,
-      "loss": 7.4406,
+      "loss": 7.4346,
       "step": 80
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.5e-06,
-      "loss": 7.3949,
+      "loss": 7.5561,
       "step": 81
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.52e-06,
-      "loss": 7.4337,
+      "loss": 7.567,
       "step": 82
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.54e-06,
-      "loss": 7.4132,
+      "loss": 7.2907,
       "step": 83
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.56e-06,
-      "loss": 7.3994,
+      "loss": 7.5245,
       "step": 84
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.5800000000000001e-06,
-      "loss": 7.355,
+      "loss": 7.45,
       "step": 85
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.6000000000000001e-06,
-      "loss": 7.4217,
+      "loss": 7.4697,
       "step": 86
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 1.6200000000000002e-06,
-      "loss": 7.3948,
+      "loss": 7.39,
       "step": 87
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 1.6400000000000002e-06,
-      "loss": 7.3221,
+      "loss": 7.4728,
       "step": 88
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 1.6600000000000002e-06,
-      "loss": 7.2875,
+      "loss": 7.5737,
       "step": 89
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.08,
       "learning_rate": 1.6800000000000002e-06,
-      "loss": 7.4339,
+      "loss": 7.41,
       "step": 90
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.08,
       "learning_rate": 1.7000000000000002e-06,
-      "loss": 7.3627,
+      "loss": 7.6328,
       "step": 91
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.72e-06,
-      "loss": 7.5915,
+      "loss": 7.5404,
       "step": 92
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.74e-06,
-      "loss": 7.4667,
+      "loss": 7.2639,
       "step": 93
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.76e-06,
-      "loss": 7.3207,
+      "loss": 7.5851,
       "step": 94
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.7800000000000001e-06,
-      "loss": 7.282,
+      "loss": 7.8058,
       "step": 95
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.8000000000000001e-06,
-      "loss": 7.5122,
+      "loss": 7.4325,
       "step": 96
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.8200000000000002e-06,
-      "loss": 7.4713,
+      "loss": 7.671,
       "step": 97
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.8400000000000002e-06,
-      "loss": 7.2456,
+      "loss": 7.5897,
       "step": 98
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.8600000000000002e-06,
-      "loss": 7.5184,
+      "loss": 7.4808,
       "step": 99
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.8800000000000002e-06,
-      "loss": 7.3295,
+      "loss": 7.5564,
       "step": 100
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.08,
       "learning_rate": 1.9000000000000002e-06,
-      "loss": 7.3192,
+      "loss": 7.2784,
       "step": 101
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 1.9200000000000003e-06,
-      "loss": 7.2951,
+      "loss": 7.2384,
       "step": 102
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 1.94e-06,
-      "loss": 7.1776,
+      "loss": 7.2829,
       "step": 103
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 1.9600000000000003e-06,
-      "loss": 7.2282,
+      "loss": 7.2878,
       "step": 104
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 1.98e-06,
-      "loss": 7.2341,
+      "loss": 7.2126,
       "step": 105
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 7.2799,
+      "loss": 7.2989,
       "step": 106
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.02e-06,
-      "loss": 7.223,
+      "loss": 7.2767,
       "step": 107
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.04e-06,
-      "loss": 7.2384,
+      "loss": 7.3321,
       "step": 108
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.06e-06,
-      "loss": 7.126,
+      "loss": 7.4267,
       "step": 109
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.08e-06,
-      "loss": 7.2243,
+      "loss": 7.2614,
       "step": 110
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.1000000000000002e-06,
-      "loss": 7.1182,
+      "loss": 7.3644,
       "step": 111
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.12e-06,
-      "loss": 7.1275,
+      "loss": 7.2905,
       "step": 112
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.1400000000000003e-06,
-      "loss": 7.2087,
+      "loss": 7.1891,
       "step": 113
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.16e-06,
-      "loss": 7.2448,
+      "loss": 7.2206,
       "step": 114
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.1800000000000003e-06,
-      "loss": 7.174,
+      "loss": 7.3103,
       "step": 115
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.2e-06,
-      "loss": 7.2965,
+      "loss": 7.4248,
       "step": 116
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.2200000000000003e-06,
-      "loss": 7.2543,
+      "loss": 7.256,
       "step": 117
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.24e-06,
-      "loss": 7.2287,
+      "loss": 7.1282,
       "step": 118
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.2600000000000004e-06,
-      "loss": 7.139,
+      "loss": 7.0698,
       "step": 119
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.28e-06,
-      "loss": 7.2388,
+      "loss": 7.2876,
       "step": 120
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.3000000000000004e-06,
-      "loss": 7.1634,
+      "loss": 7.3798,
       "step": 121
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.3200000000000002e-06,
-      "loss": 7.1059,
+      "loss": 7.2382,
       "step": 122
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.1,
       "learning_rate": 2.3400000000000005e-06,
-      "loss": 7.1523,
+      "loss": 7.2945,
       "step": 123
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.1,
       "learning_rate": 2.3600000000000003e-06,
-      "loss": 7.1194,
+      "loss": 7.2533,
       "step": 124
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 2.38e-06,
-      "loss": 7.2067,
+      "loss": 7.2129,
       "step": 125
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 2.4000000000000003e-06,
-      "loss": 7.2303,
+      "loss": 7.1037,
       "step": 126
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 2.42e-06,
-      "loss": 7.2051,
+      "loss": 7.4396,
       "step": 127
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.4400000000000004e-06,
-      "loss": 7.1689,
+      "loss": 7.1695,
       "step": 128
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.46e-06,
-      "loss": 7.2276,
+      "loss": 7.211,
       "step": 129
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.4800000000000004e-06,
-      "loss": 7.2736,
+      "loss": 7.1477,
       "step": 130
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.5e-06,
-      "loss": 7.0741,
+      "loss": 7.0502,
       "step": 131
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.52e-06,
-      "loss": 7.2548,
+      "loss": 7.2168,
       "step": 132
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.5400000000000002e-06,
-      "loss": 7.1224,
+      "loss": 7.363,
       "step": 133
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.56e-06,
-      "loss": 7.1548,
+      "loss": 7.1792,
       "step": 134
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.5800000000000003e-06,
-      "loss": 7.2653,
+      "loss": 7.227,
       "step": 135
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.6e-06,
-      "loss": 7.0389,
+      "loss": 7.1339,
       "step": 136
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.6200000000000003e-06,
-      "loss": 7.2487,
+      "loss": 7.1581,
       "step": 137
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.64e-06,
-      "loss": 7.173,
+      "loss": 7.2339,
       "step": 138
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.6600000000000004e-06,
-      "loss": 7.1693,
+      "loss": 7.1666,
       "step": 139
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.68e-06,
-      "loss": 7.2814,
+      "loss": 7.2801,
       "step": 140
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.7000000000000004e-06,
-      "loss": 7.2482,
+      "loss": 7.2069,
       "step": 141
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.7200000000000002e-06,
-      "loss": 7.2497,
+      "loss": 7.2248,
       "step": 142
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.7400000000000004e-06,
-      "loss": 7.0239,
+      "loss": 7.1036,
       "step": 143
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.7600000000000003e-06,
-      "loss": 7.1039,
+      "loss": 7.2377,
       "step": 144
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.7800000000000005e-06,
-      "loss": 7.0526,
+      "loss": 7.1325,
       "step": 145
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.8000000000000003e-06,
-      "loss": 7.0991,
+      "loss": 7.3692,
       "step": 146
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.82e-06,
-      "loss": 7.1917,
+      "loss": 7.12,
       "step": 147
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.84e-06,
-      "loss": 7.2096,
+      "loss": 6.9214,
       "step": 148
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.13,
       "learning_rate": 2.86e-06,
-      "loss": 6.9993,
+      "loss": 7.1352,
       "step": 149
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.88e-06,
-      "loss": 6.9164,
+      "loss": 7.1289,
       "step": 150
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.9e-06,
-      "loss": 7.1531,
+      "loss": 7.3238,
       "step": 151
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.92e-06,
-      "loss": 7.1666,
+      "loss": 7.369,
       "step": 152
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.9400000000000002e-06,
-      "loss": 7.1133,
+      "loss": 7.2302,
       "step": 153
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 2.96e-06,
-      "loss": 7.0933,
+      "loss": 7.2603,
       "step": 154
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 2.9800000000000003e-06,
-      "loss": 7.0639,
+      "loss": 7.1517,
       "step": 155
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3e-06,
-      "loss": 7.1154,
+      "loss": 7.1303,
       "step": 156
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3.0200000000000003e-06,
-      "loss": 7.1863,
+      "loss": 7.1392,
       "step": 157
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3.04e-06,
-      "loss": 7.1349,
+      "loss": 7.2775,
       "step": 158
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.13,
       "learning_rate": 3.0600000000000003e-06,
-      "loss": 7.1162,
+      "loss": 7.0583,
       "step": 159
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.13,
       "learning_rate": 3.08e-06,
-      "loss": 7.1441,
+      "loss": 7.0259,
       "step": 160
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.14,
       "learning_rate": 3.1000000000000004e-06,
-      "loss": 7.0902,
+      "loss": 7.1242,
       "step": 161
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.14,
       "learning_rate": 3.12e-06,
-      "loss": 7.1615,
+      "loss": 7.2233,
       "step": 162
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.1400000000000004e-06,
-      "loss": 7.1423,
+      "loss": 7.2768,
       "step": 163
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.1600000000000002e-06,
-      "loss": 7.1207,
+      "loss": 7.132,
       "step": 164
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.1800000000000005e-06,
-      "loss": 7.0631,
+      "loss": 7.111,
       "step": 165
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.2000000000000003e-06,
-      "loss": 7.1458,
+      "loss": 7.0004,
       "step": 166
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.2200000000000005e-06,
-      "loss": 7.0482,
+      "loss": 7.1518,
       "step": 167
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.2400000000000003e-06,
-      "loss": 7.0045,
+      "loss": 7.1699,
       "step": 168
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.2600000000000006e-06,
-      "loss": 7.0607,
+      "loss": 7.1734,
       "step": 169
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.2800000000000004e-06,
-      "loss": 7.1924,
+      "loss": 7.0623,
       "step": 170
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.3000000000000006e-06,
-      "loss": 7.1058,
+      "loss": 7.1003,
       "step": 171
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.14,
       "learning_rate": 3.3200000000000004e-06,
-      "loss": 7.1291,
+      "loss": 7.0884,
       "step": 172
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.3400000000000006e-06,
-      "loss": 7.0849,
+      "loss": 6.9313,
       "step": 173
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.3600000000000004e-06,
-      "loss": 7.1391,
+      "loss": 7.1607,
       "step": 174
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.3800000000000007e-06,
-      "loss": 7.1101,
+      "loss": 7.1983,
       "step": 175
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.4000000000000005e-06,
-      "loss": 7.0369,
+      "loss": 7.3195,
       "step": 176
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.4200000000000007e-06,
-      "loss": 7.0363,
+      "loss": 7.2551,
       "step": 177
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.44e-06,
-      "loss": 7.1012,
+      "loss": 7.1921,
       "step": 178
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.46e-06,
-      "loss": 7.0995,
+      "loss": 7.1054,
       "step": 179
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.48e-06,
-      "loss": 7.1561,
+      "loss": 7.1595,
       "step": 180
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.5e-06,
-      "loss": 7.1402,
+      "loss": 7.2107,
       "step": 181
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.52e-06,
-      "loss": 7.0451,
+      "loss": 7.2974,
       "step": 182
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.54e-06,
-      "loss": 7.148,
+      "loss": 7.0981,
       "step": 183
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.5600000000000002e-06,
-      "loss": 7.1851,
+      "loss": 7.1042,
       "step": 184
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.16,
       "learning_rate": 3.58e-06,
-      "loss": 7.1082,
+      "loss": 6.9914,
       "step": 185
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.6000000000000003e-06,
-      "loss": 6.9746,
+      "loss": 6.9904,
       "step": 186
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.62e-06,
-      "loss": 7.0704,
+      "loss": 7.2776,
       "step": 187
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.6400000000000003e-06,
-      "loss": 7.0528,
+      "loss": 7.0688,
       "step": 188
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.66e-06,
-      "loss": 7.1322,
+      "loss": 7.0942,
       "step": 189
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.6800000000000003e-06,
-      "loss": 7.1972,
+      "loss": 7.0486,
       "step": 190
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.7e-06,
-      "loss": 7.2024,
+      "loss": 7.293,
       "step": 191
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.7200000000000004e-06,
-      "loss": 7.0408,
+      "loss": 7.306,
       "step": 192
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.74e-06,
-      "loss": 7.1215,
+      "loss": 7.1269,
       "step": 193
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.7600000000000004e-06,
-      "loss": 7.0417,
+      "loss": 7.0907,
       "step": 194
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.16,
       "learning_rate": 3.7800000000000002e-06,
-      "loss": 7.01,
+      "loss": 6.9632,
       "step": 195
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.16,
       "learning_rate": 3.8000000000000005e-06,
-      "loss": 6.9221,
+      "loss": 6.9237,
       "step": 196
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.17,
       "learning_rate": 3.820000000000001e-06,
-      "loss": 6.8631,
+      "loss": 7.2833,
       "step": 197
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.17,
       "learning_rate": 3.8400000000000005e-06,
-      "loss": 7.045,
+      "loss": 6.9452,
       "step": 198
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.86e-06,
-      "loss": 6.8861,
+      "loss": 6.8769,
       "step": 199
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.88e-06,
-      "loss": 6.7337,
+      "loss": 6.5161,
       "step": 200
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.900000000000001e-06,
-      "loss": 7.2072,
+      "loss": 7.1482,
       "step": 201
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.920000000000001e-06,
-      "loss": 7.1981,
+      "loss": 7.0983,
       "step": 202
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 3.94e-06,
-      "loss": 7.1899,
+      "loss": 7.0931,
       "step": 203
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 3.96e-06,
-      "loss": 7.1453,
+      "loss": 7.022,
       "step": 204
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 3.980000000000001e-06,
-      "loss": 7.1685,
+      "loss": 7.0267,
       "step": 205
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 7.1609,
+      "loss": 7.1611,
       "step": 206
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 4.0200000000000005e-06,
-      "loss": 7.0495,
+      "loss": 7.1138,
       "step": 207
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.17,
       "learning_rate": 4.04e-06,
-      "loss": 7.187,
+      "loss": 7.3329,
       "step": 208
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 4.060000000000001e-06,
-      "loss": 7.1006,
+      "loss": 7.1811,
       "step": 209
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 4.08e-06,
-      "loss": 7.1509,
+      "loss": 7.2872,
       "step": 210
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 4.1e-06,
-      "loss": 7.0645,
+      "loss": 7.0531,
       "step": 211
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.12e-06,
-      "loss": 7.1027,
+      "loss": 6.9959,
       "step": 212
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.14e-06,
-      "loss": 6.9497,
+      "loss": 7.0875,
       "step": 213
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.16e-06,
-      "loss": 7.0604,
+      "loss": 7.0946,
       "step": 214
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.18e-06,
-      "loss": 7.032,
+      "loss": 7.0841,
       "step": 215
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.2000000000000004e-06,
-      "loss": 7.0713,
+      "loss": 7.0711,
       "step": 216
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.22e-06,
-      "loss": 7.059,
+      "loss": 7.1116,
       "step": 217
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.24e-06,
-      "loss": 7.0995,
+      "loss": 7.1287,
       "step": 218
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.26e-06,
-      "loss": 7.0596,
+      "loss": 7.0188,
       "step": 219
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.2800000000000005e-06,
-      "loss": 7.0901,
+      "loss": 7.2101,
       "step": 220
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.3e-06,
-      "loss": 7.1409,
+      "loss": 7.008,
       "step": 221
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.32e-06,
-      "loss": 7.1144,
+      "loss": 7.3396,
       "step": 222
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.34e-06,
-      "loss": 7.0433,
+      "loss": 7.2366,
       "step": 223
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.360000000000001e-06,
-      "loss": 6.9968,
+      "loss": 7.2528,
       "step": 224
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.38e-06,
-      "loss": 7.1748,
+      "loss": 7.0548,
       "step": 225
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.4e-06,
-      "loss": 7.0851,
+      "loss": 6.9545,
       "step": 226
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.42e-06,
-      "loss": 7.178,
+      "loss": 7.1837,
       "step": 227
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.440000000000001e-06,
-      "loss": 7.1072,
+      "loss": 7.172,
       "step": 228
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.4600000000000005e-06,
-      "loss": 7.0404,
+      "loss": 7.0596,
       "step": 229
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.19,
       "learning_rate": 4.48e-06,
-      "loss": 7.0697,
+      "loss": 7.1248,
       "step": 230
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.19,
       "learning_rate": 4.5e-06,
-      "loss": 7.1125,
+      "loss": 7.1866,
       "step": 231
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 4.520000000000001e-06,
-      "loss": 6.9526,
+      "loss": 7.1138,
       "step": 232
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 4.540000000000001e-06,
-      "loss": 7.0077,
+      "loss": 7.076,
       "step": 233
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 4.56e-06,
-      "loss": 7.0224,
+      "loss": 6.9825,
       "step": 234
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.58e-06,
-      "loss": 7.1392,
+      "loss": 7.1614,
       "step": 235
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.600000000000001e-06,
-      "loss": 7.0185,
+      "loss": 7.1638,
       "step": 236
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.620000000000001e-06,
-      "loss": 7.1857,
+      "loss": 7.1362,
       "step": 237
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.6400000000000005e-06,
-      "loss": 7.0496,
+      "loss": 7.0607,
       "step": 238
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.66e-06,
-      "loss": 6.9863,
+      "loss": 7.1012,
       "step": 239
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.680000000000001e-06,
-      "loss": 7.2381,
+      "loss": 6.9509,
       "step": 240
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.7e-06,
-      "loss": 7.0243,
+      "loss": 7.1166,
       "step": 241
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.7200000000000005e-06,
-      "loss": 7.0274,
+      "loss": 7.095,
       "step": 242
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.74e-06,
-      "loss": 7.1441,
+      "loss": 7.1828,
       "step": 243
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.76e-06,
-      "loss": 6.9591,
+      "loss": 7.2223,
       "step": 244
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.78e-06,
-      "loss": 6.9841,
+      "loss": 6.9206,
       "step": 245
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.800000000000001e-06,
-      "loss": 6.9644,
+      "loss": 6.9058,
       "step": 246
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.8200000000000004e-06,
-      "loss": 6.7741,
+      "loss": 7.4591,
       "step": 247
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.84e-06,
-      "loss": 7.0463,
+      "loss": 6.8475,
       "step": 248
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.86e-06,
-      "loss": 6.9203,
+      "loss": 6.8085,
       "step": 249
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.880000000000001e-06,
-      "loss": 6.7171,
+      "loss": 6.6114,
       "step": 250
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.9000000000000005e-06,
-      "loss": 7.203,
+      "loss": 7.2398,
       "step": 251
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 4.92e-06,
-      "loss": 7.1808,
+      "loss": 6.9727,
       "step": 252
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 4.94e-06,
-      "loss": 7.1442,
+      "loss": 7.1992,
       "step": 253
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 4.960000000000001e-06,
-      "loss": 7.1738,
+      "loss": 6.9822,
       "step": 254
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 4.980000000000001e-06,
-      "loss": 7.2137,
+      "loss": 7.0725,
       "step": 255
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.22,
       "learning_rate": 5e-06,
-      "loss": 7.1239,
+      "loss": 7.1447,
       "step": 256
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.02e-06,
-      "loss": 7.0826,
+      "loss": 7.07,
       "step": 257
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.04e-06,
-      "loss": 7.1231,
+      "loss": 7.1652,
       "step": 258
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.060000000000001e-06,
-      "loss": 7.0327,
+      "loss": 7.0423,
       "step": 259
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.0800000000000005e-06,
-      "loss": 7.062,
+      "loss": 6.9902,
       "step": 260
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.1e-06,
-      "loss": 7.0684,
+      "loss": 7.0468,
       "step": 261
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.12e-06,
-      "loss": 7.0394,
+      "loss": 7.0185,
       "step": 262
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.140000000000001e-06,
-      "loss": 7.1074,
+      "loss": 6.9692,
       "step": 263
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.1600000000000006e-06,
-      "loss": 7.1088,
+      "loss": 7.099,
       "step": 264
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.18e-06,
-      "loss": 7.1337,
+      "loss": 7.051,
       "step": 265
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.22,
       "learning_rate": 5.2e-06,
-      "loss": 6.9456,
+      "loss": 7.0721,
       "step": 266
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.22,
       "learning_rate": 5.220000000000001e-06,
-      "loss": 7.0948,
+      "loss": 6.994,
       "step": 267
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.23,
       "learning_rate": 5.240000000000001e-06,
-      "loss": 6.9807,
+      "loss": 6.9929,
       "step": 268
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.23,
       "learning_rate": 5.2600000000000005e-06,
-      "loss": 7.0879,
+      "loss": 7.0564,
       "step": 269
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.28e-06,
-      "loss": 7.1338,
+      "loss": 7.0759,
       "step": 270
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.300000000000001e-06,
-      "loss": 7.0202,
+      "loss": 7.1145,
       "step": 271
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.320000000000001e-06,
-      "loss": 7.0207,
+      "loss": 7.0028,
       "step": 272
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.3400000000000005e-06,
-      "loss": 6.9342,
+      "loss": 7.1359,
       "step": 273
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.36e-06,
-      "loss": 6.9222,
+      "loss": 7.1386,
       "step": 274
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.380000000000001e-06,
-      "loss": 7.0924,
+      "loss": 7.1558,
       "step": 275
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.400000000000001e-06,
-      "loss": 7.0018,
+      "loss": 6.9148,
       "step": 276
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.420000000000001e-06,
-      "loss": 6.9829,
+      "loss": 7.0105,
       "step": 277
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.4400000000000004e-06,
-      "loss": 7.1071,
+      "loss": 7.0457,
       "step": 278
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.23,
       "learning_rate": 5.460000000000001e-06,
-      "loss": 7.1579,
+      "loss": 7.1764,
       "step": 279
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.480000000000001e-06,
-      "loss": 7.1568,
+      "loss": 6.9304,
       "step": 280
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 7.0662,
+      "loss": 6.9352,
       "step": 281
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.5200000000000005e-06,
-      "loss": 7.0609,
+      "loss": 7.0308,
       "step": 282
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.540000000000001e-06,
-      "loss": 6.9644,
+      "loss": 7.2257,
       "step": 283
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.560000000000001e-06,
-      "loss": 7.0189,
+      "loss": 7.1661,
       "step": 284
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.580000000000001e-06,
-      "loss": 7.059,
+      "loss": 7.0588,
       "step": 285
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.600000000000001e-06,
-      "loss": 7.1024,
+      "loss": 7.0741,
       "step": 286
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.620000000000001e-06,
-      "loss": 7.0941,
+      "loss": 6.9665,
       "step": 287
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.64e-06,
-      "loss": 7.0827,
+      "loss": 7.0131,
       "step": 288
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.66e-06,
-      "loss": 7.0403,
+      "loss": 7.1136,
       "step": 289
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.68e-06,
-      "loss": 7.0994,
+      "loss": 6.9755,
       "step": 290
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.7e-06,
-      "loss": 7.1633,
+      "loss": 7.2509,
       "step": 291
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.25,
       "learning_rate": 5.72e-06,
-      "loss": 7.0665,
+      "loss": 7.0928,
       "step": 292
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.74e-06,
-      "loss": 6.9603,
+      "loss": 6.7829,
       "step": 293
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.76e-06,
-      "loss": 6.9473,
+      "loss": 6.898,
       "step": 294
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.78e-06,
-      "loss": 6.9004,
+      "loss": 7.03,
       "step": 295
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.8e-06,
-      "loss": 6.866,
+      "loss": 7.1157,
       "step": 296
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.82e-06,
-      "loss": 6.9382,
+      "loss": 6.8165,
       "step": 297
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.84e-06,
-      "loss": 7.0233,
+      "loss": 6.9187,
       "step": 298
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.86e-06,
-      "loss": 6.6025,
+      "loss": 7.0252,
       "step": 299
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.8800000000000005e-06,
-      "loss": 6.6428,
+      "loss": 6.6754,
       "step": 300
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.9e-06,
-      "loss": 7.2423,
+      "loss": 6.9993,
       "step": 301
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.25,
       "learning_rate": 5.92e-06,
-      "loss": 7.1925,
+      "loss": 7.1661,
       "step": 302
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.25,
       "learning_rate": 5.94e-06,
-      "loss": 7.2078,
+      "loss": 7.2031,
       "step": 303
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.26,
       "learning_rate": 5.9600000000000005e-06,
-      "loss": 7.0933,
+      "loss": 6.9499,
       "step": 304
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.26,
       "learning_rate": 5.98e-06,
-      "loss": 7.255,
+      "loss": 7.0675,
       "step": 305
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6e-06,
-      "loss": 7.0256,
+      "loss": 7.0218,
       "step": 306
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6.02e-06,
-      "loss": 6.9909,
+      "loss": 7.0893,
       "step": 307
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6.040000000000001e-06,
-      "loss": 7.1496,
+      "loss": 6.9799,
       "step": 308
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6.0600000000000004e-06,
-      "loss": 6.9642,
+      "loss": 6.9179,
       "step": 309
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.08e-06,
-      "loss": 7.0763,
+      "loss": 7.1749,
       "step": 310
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.1e-06,
-      "loss": 7.0671,
+      "loss": 6.8198,
       "step": 311
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.120000000000001e-06,
-      "loss": 7.0533,
+      "loss": 6.9023,
       "step": 312
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.1400000000000005e-06,
-      "loss": 6.982,
+      "loss": 6.9893,
       "step": 313
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.16e-06,
-      "loss": 6.9866,
+      "loss": 7.0538,
       "step": 314
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.26,
       "learning_rate": 6.18e-06,
-      "loss": 7.1002,
+      "loss": 6.98,
       "step": 315
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 6.200000000000001e-06,
-      "loss": 7.0481,
+      "loss": 7.2463,
       "step": 316
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 6.220000000000001e-06,
-      "loss": 7.014,
+      "loss": 7.0738,
       "step": 317
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 6.24e-06,
-      "loss": 6.9361,
+      "loss": 7.1088,
       "step": 318
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.26e-06,
-      "loss": 7.0155,
+      "loss": 6.9552,
       "step": 319
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.280000000000001e-06,
-      "loss": 7.0522,
+      "loss": 6.9688,
       "step": 320
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.300000000000001e-06,
-      "loss": 6.9896,
+      "loss": 6.9983,
       "step": 321
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.3200000000000005e-06,
-      "loss": 7.0202,
+      "loss": 6.9825,
       "step": 322
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.34e-06,
-      "loss": 7.078,
+      "loss": 6.9598,
       "step": 323
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.360000000000001e-06,
-      "loss": 7.0054,
+      "loss": 7.0496,
       "step": 324
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.380000000000001e-06,
-      "loss": 7.0511,
+      "loss": 6.8867,
       "step": 325
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.4000000000000006e-06,
-      "loss": 7.0643,
+      "loss": 7.1072,
       "step": 326
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.42e-06,
-      "loss": 7.047,
+      "loss": 7.043,
       "step": 327
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.440000000000001e-06,
-      "loss": 6.9796,
+      "loss": 7.0664,
       "step": 328
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.460000000000001e-06,
-      "loss": 7.0924,
+      "loss": 6.9956,
       "step": 329
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.480000000000001e-06,
-      "loss": 7.0856,
+      "loss": 7.1473,
       "step": 330
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 7.0372,
+      "loss": 7.0503,
       "step": 331
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.520000000000001e-06,
-      "loss": 6.9479,
+      "loss": 7.1276,
       "step": 332
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.540000000000001e-06,
-      "loss": 7.0917,
+      "loss": 7.1452,
       "step": 333
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.560000000000001e-06,
-      "loss": 7.0307,
+      "loss": 6.9883,
       "step": 334
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.5800000000000005e-06,
-      "loss": 7.124,
+      "loss": 7.065,
       "step": 335
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.600000000000001e-06,
-      "loss": 7.0226,
+      "loss": 7.1492,
       "step": 336
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 6.620000000000001e-06,
-      "loss": 6.9063,
+      "loss": 6.887,
       "step": 337
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 6.640000000000001e-06,
-      "loss": 7.0728,
+      "loss": 7.1826,
       "step": 338
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 6.660000000000001e-06,
-      "loss": 7.0572,
+      "loss": 6.9978,
       "step": 339
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.29,
       "learning_rate": 6.680000000000001e-06,
-      "loss": 7.0749,
+      "loss": 7.0577,
       "step": 340
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.29,
       "learning_rate": 6.700000000000001e-06,
-      "loss": 6.969,
+      "loss": 6.9807,
       "step": 341
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.720000000000001e-06,
-      "loss": 7.0759,
+      "loss": 7.1097,
       "step": 342
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.740000000000001e-06,
-      "loss": 6.9305,
+      "loss": 6.8924,
       "step": 343
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.760000000000001e-06,
-      "loss": 6.9073,
+      "loss": 7.0879,
       "step": 344
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.780000000000001e-06,
-      "loss": 6.881,
+      "loss": 6.8813,
       "step": 345
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.800000000000001e-06,
-      "loss": 6.9599,
+      "loss": 6.9154,
       "step": 346
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.820000000000001e-06,
-      "loss": 6.8266,
+      "loss": 7.035,
       "step": 347
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.8400000000000014e-06,
-      "loss": 6.6888,
+      "loss": 7.0761,
       "step": 348
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.860000000000001e-06,
-      "loss": 6.7824,
+      "loss": 6.7615,
       "step": 349
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.88e-06,
-      "loss": 6.5344,
+      "loss": 6.8788,
       "step": 350
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 6.9e-06,
-      "loss": 7.1566,
+      "loss": 7.2978,
       "step": 351
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 6.92e-06,
-      "loss": 7.2001,
+      "loss": 7.1531,
       "step": 352
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 6.9400000000000005e-06,
-      "loss": 7.0835,
+      "loss": 7.0533,
       "step": 353
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 6.96e-06,
-      "loss": 7.023,
+      "loss": 7.1514,
       "step": 354
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 6.98e-06,
-      "loss": 7.0585,
+      "loss": 7.0157,
       "step": 355
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7e-06,
-      "loss": 7.1668,
+      "loss": 7.1494,
       "step": 356
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.0200000000000006e-06,
-      "loss": 7.0,
+      "loss": 7.1684,
       "step": 357
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.04e-06,
-      "loss": 7.0832,
+      "loss": 6.9543,
       "step": 358
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.06e-06,
-      "loss": 6.9888,
+      "loss": 7.0365,
       "step": 359
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 7.08e-06,
-      "loss": 6.9788,
+      "loss": 7.1179,
       "step": 360
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 7.100000000000001e-06,
-      "loss": 7.0177,
+      "loss": 6.9828,
       "step": 361
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 7.1200000000000004e-06,
-      "loss": 6.9336,
+      "loss": 6.9125,
       "step": 362
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.31,
       "learning_rate": 7.14e-06,
-      "loss": 6.8755,
+      "loss": 7.1676,
       "step": 363
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.16e-06,
-      "loss": 6.938,
+      "loss": 7.0857,
       "step": 364
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.180000000000001e-06,
-      "loss": 7.0578,
+      "loss": 6.9913,
       "step": 365
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.2000000000000005e-06,
-      "loss": 7.0069,
+      "loss": 7.0921,
       "step": 366
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.22e-06,
-      "loss": 7.1085,
+      "loss": 7.0941,
       "step": 367
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.24e-06,
-      "loss": 7.0106,
+      "loss": 7.2049,
       "step": 368
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.260000000000001e-06,
-      "loss": 6.9103,
+      "loss": 7.0583,
       "step": 369
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.280000000000001e-06,
-      "loss": 6.9784,
+      "loss": 7.124,
       "step": 370
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.3e-06,
-      "loss": 7.0409,
+      "loss": 6.9423,
       "step": 371
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.32e-06,
-      "loss": 7.069,
+      "loss": 6.9826,
       "step": 372
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.31,
       "learning_rate": 7.340000000000001e-06,
-      "loss": 6.9829,
+      "loss": 6.9073,
       "step": 373
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.31,
       "learning_rate": 7.360000000000001e-06,
-      "loss": 6.8754,
+      "loss": 6.9879,
       "step": 374
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.32,
       "learning_rate": 7.3800000000000005e-06,
-      "loss": 6.9719,
+      "loss": 6.9442,
       "step": 375
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.32,
       "learning_rate": 7.4e-06,
-      "loss": 6.9281,
+      "loss": 7.1176,
       "step": 376
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.420000000000001e-06,
-      "loss": 6.9857,
+      "loss": 6.9775,
       "step": 377
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.440000000000001e-06,
-      "loss": 6.9805,
+      "loss": 7.0327,
       "step": 378
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.4600000000000006e-06,
-      "loss": 6.9681,
+      "loss": 7.0447,
       "step": 379
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.48e-06,
-      "loss": 6.962,
+      "loss": 6.9938,
       "step": 380
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 7.0683,
+      "loss": 7.0039,
       "step": 381
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.520000000000001e-06,
-      "loss": 7.0167,
+      "loss": 7.0229,
       "step": 382
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.540000000000001e-06,
-      "loss": 7.0039,
+      "loss": 6.8511,
       "step": 383
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.5600000000000005e-06,
-      "loss": 6.9635,
+      "loss": 7.2123,
       "step": 384
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.58e-06,
-      "loss": 7.0326,
+      "loss": 6.9928,
       "step": 385
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.32,
       "learning_rate": 7.600000000000001e-06,
-      "loss": 6.9914,
+      "loss": 6.914,
       "step": 386
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.620000000000001e-06,
-      "loss": 6.9522,
+      "loss": 6.9887,
       "step": 387
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.640000000000001e-06,
-      "loss": 7.0191,
+      "loss": 7.0655,
       "step": 388
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.660000000000001e-06,
-      "loss": 6.9369,
+      "loss": 7.1643,
       "step": 389
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.680000000000001e-06,
-      "loss": 6.8405,
+      "loss": 7.3189,
       "step": 390
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.7e-06,
-      "loss": 6.9025,
+      "loss": 7.1428,
       "step": 391
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.72e-06,
-      "loss": 6.942,
+      "loss": 6.9643,
       "step": 392
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.74e-06,
-      "loss": 6.8378,
+      "loss": 6.8432,
       "step": 393
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.76e-06,
-      "loss": 6.8714,
+      "loss": 7.026,
       "step": 394
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.78e-06,
-      "loss": 6.8833,
+      "loss": 6.8437,
       "step": 395
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.800000000000002e-06,
-      "loss": 6.9241,
+      "loss": 6.9199,
       "step": 396
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.820000000000001e-06,
-      "loss": 6.8455,
+      "loss": 6.715,
       "step": 397
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.840000000000001e-06,
-      "loss": 6.7152,
+      "loss": 6.7204,
       "step": 398
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.34,
       "learning_rate": 7.860000000000001e-06,
-      "loss": 6.7795,
+      "loss": 6.6869,
       "step": 399
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.88e-06,
-      "loss": 6.3986,
+      "loss": 6.4494,
       "step": 400
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.9e-06,
-      "loss": 7.1826,
+      "loss": 7.0713,
       "step": 401
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.92e-06,
-      "loss": 7.1695,
+      "loss": 7.04,
       "step": 402
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.94e-06,
-      "loss": 7.2669,
+      "loss": 7.0701,
       "step": 403
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 7.960000000000002e-06,
-      "loss": 7.1548,
+      "loss": 7.1369,
       "step": 404
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 7.980000000000002e-06,
-      "loss": 6.9942,
+      "loss": 6.908,
       "step": 405
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 7.041,
+      "loss": 7.0375,
       "step": 406
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8.020000000000001e-06,
-      "loss": 7.0299,
+      "loss": 7.0368,
       "step": 407
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8.040000000000001e-06,
-      "loss": 6.9149,
+      "loss": 7.0894,
       "step": 408
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.34,
       "learning_rate": 8.06e-06,
-      "loss": 6.8887,
+      "loss": 6.9395,
       "step": 409
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.34,
       "learning_rate": 8.08e-06,
-      "loss": 6.9186,
+      "loss": 7.14,
       "step": 410
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.35,
       "learning_rate": 8.1e-06,
-      "loss": 7.0415,
+      "loss": 7.0961,
       "step": 411
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.35,
       "learning_rate": 8.120000000000002e-06,
-      "loss": 6.9518,
+      "loss": 7.0724,
       "step": 412
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.14e-06,
-      "loss": 6.9932,
+      "loss": 7.0039,
       "step": 413
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.16e-06,
-      "loss": 6.9789,
+      "loss": 6.9363,
       "step": 414
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.18e-06,
-      "loss": 7.0404,
+      "loss": 6.9689,
       "step": 415
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.2e-06,
-      "loss": 7.0041,
+      "loss": 7.2013,
       "step": 416
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.220000000000001e-06,
-      "loss": 6.9786,
+      "loss": 6.9674,
       "step": 417
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.24e-06,
-      "loss": 6.8628,
+      "loss": 6.8554,
       "step": 418
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.26e-06,
-      "loss": 6.9054,
+      "loss": 6.9712,
       "step": 419
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.28e-06,
-      "loss": 6.9641,
+      "loss": 7.144,
       "step": 420
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.3e-06,
-      "loss": 7.002,
+      "loss": 7.0049,
       "step": 421
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.35,
       "learning_rate": 8.32e-06,
-      "loss": 6.8865,
+      "loss": 6.8933,
       "step": 422
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 8.34e-06,
-      "loss": 6.93,
+      "loss": 7.0381,
       "step": 423
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 8.36e-06,
-      "loss": 7.0314,
+      "loss": 6.9631,
       "step": 424
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 8.380000000000001e-06,
-      "loss": 6.8996,
+      "loss": 7.1917,
       "step": 425
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 6.9738,
+      "loss": 7.0247,
       "step": 426
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.42e-06,
-      "loss": 6.9319,
+      "loss": 6.9609,
       "step": 427
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.44e-06,
-      "loss": 6.8332,
+      "loss": 7.0366,
       "step": 428
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.46e-06,
-      "loss": 6.8213,
+      "loss": 7.0985,
       "step": 429
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.48e-06,
-      "loss": 6.8614,
+      "loss": 7.0134,
       "step": 430
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.5e-06,
-      "loss": 6.9083,
+      "loss": 7.0934,
       "step": 431
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.52e-06,
-      "loss": 7.0035,
+      "loss": 6.8146,
       "step": 432
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.540000000000001e-06,
-      "loss": 6.8831,
+      "loss": 7.082,
       "step": 433
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.560000000000001e-06,
-      "loss": 6.8566,
+      "loss": 7.0223,
       "step": 434
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.580000000000001e-06,
-      "loss": 6.9702,
+      "loss": 7.0763,
       "step": 435
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.6e-06,
-      "loss": 6.8319,
+      "loss": 6.8739,
       "step": 436
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.62e-06,
-      "loss": 6.8398,
+      "loss": 6.9112,
       "step": 437
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.64e-06,
-      "loss": 6.9327,
+      "loss": 6.8633,
       "step": 438
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.66e-06,
-      "loss": 6.8757,
+      "loss": 7.1312,
       "step": 439
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.68e-06,
-      "loss": 6.8738,
+      "loss": 6.9781,
       "step": 440
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.700000000000001e-06,
-      "loss": 6.7065,
+      "loss": 7.0472,
       "step": 441
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.720000000000001e-06,
-      "loss": 6.9136,
+      "loss": 6.9132,
       "step": 442
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.740000000000001e-06,
-      "loss": 6.5767,
+      "loss": 6.9745,
       "step": 443
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 8.76e-06,
-      "loss": 6.8314,
+      "loss": 6.8829,
       "step": 444
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 8.78e-06,
-      "loss": 6.5081,
+      "loss": 6.683,
       "step": 445
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 8.8e-06,
-      "loss": 6.3829,
+      "loss": 6.9166,
       "step": 446
     },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.82e-06,
+      "loss": 6.7985,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.84e-06,
+      "loss": 6.9687,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 6.6958,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 6.494,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 6.9844,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 6.9767,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.94e-06,
+      "loss": 7.0195,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.96e-06,
+      "loss": 7.0104,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.98e-06,
+      "loss": 6.9913,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9e-06,
+      "loss": 6.9891,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 7.0139,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 7.1273,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 7.0984,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 7.0295,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 6.9996,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.12e-06,
+      "loss": 6.9963,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.14e-06,
+      "loss": 7.013,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.16e-06,
+      "loss": 7.0542,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 6.9944,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 6.9589,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 6.845,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 6.9199,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 6.7606,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 7.1722,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.3e-06,
+      "loss": 6.9564,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.32e-06,
+      "loss": 7.0807,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 7.0357,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 7.1361,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.38e-06,
+      "loss": 6.9145,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.4e-06,
+      "loss": 6.8052,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.42e-06,
+      "loss": 6.7791,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 6.9294,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 6.9155,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.48e-06,
+      "loss": 7.0952,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.5e-06,
+      "loss": 6.8783,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.52e-06,
+      "loss": 6.9119,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.54e-06,
+      "loss": 6.9301,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.56e-06,
+      "loss": 6.9932,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.58e-06,
+      "loss": 7.028,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 6.817,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 6.8959,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 7.0075,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.66e-06,
+      "loss": 7.0015,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.68e-06,
+      "loss": 6.8851,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.7e-06,
+      "loss": 7.0308,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.72e-06,
+      "loss": 7.0487,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.74e-06,
+      "loss": 7.0695,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 7.0122,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 6.9017,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 6.7882,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 6.7723,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.84e-06,
+      "loss": 6.7281,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.86e-06,
+      "loss": 6.6603,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.88e-06,
+      "loss": 6.5785,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 6.863341808319092,
+      "eval_runtime": 760.7122,
+      "eval_samples_per_second": 3.473,
+      "eval_steps_per_second": 0.291,
+      "eval_wer": 1.2730321249280712,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.9e-06,
+      "loss": 7.0143,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 6.9906,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 6.9772,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 6.9237,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 6.9579,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 7.1024,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.985486211901307e-06,
+      "loss": 6.9507,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.970972423802613e-06,
+      "loss": 6.9772,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.95645863570392e-06,
+      "loss": 6.8646,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.941944847605226e-06,
+      "loss": 6.9444,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.927431059506532e-06,
+      "loss": 6.7739,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.912917271407838e-06,
+      "loss": 6.985,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.898403483309144e-06,
+      "loss": 6.9392,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.883889695210452e-06,
+      "loss": 6.8675,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.869375907111757e-06,
+      "loss": 6.9742,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.854862119013063e-06,
+      "loss": 6.8329,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.84034833091437e-06,
+      "loss": 6.9738,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.825834542815675e-06,
+      "loss": 6.8553,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.811320754716981e-06,
+      "loss": 7.1065,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.79680696661829e-06,
+      "loss": 6.8504,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.782293178519594e-06,
+      "loss": 6.9724,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.7677793904209e-06,
+      "loss": 6.9496,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.753265602322208e-06,
+      "loss": 6.9618,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.738751814223512e-06,
+      "loss": 6.8863,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.724238026124818e-06,
+      "loss": 6.8763,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.709724238026126e-06,
+      "loss": 6.8101,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.69521044992743e-06,
+      "loss": 6.949,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.680696661828739e-06,
+      "loss": 7.0043,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.666182873730045e-06,
+      "loss": 6.9053,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.651669085631351e-06,
+      "loss": 6.9482,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.637155297532657e-06,
+      "loss": 6.8926,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.622641509433963e-06,
+      "loss": 6.9342,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.60812772133527e-06,
+      "loss": 6.8225,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.593613933236576e-06,
+      "loss": 6.8438,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.579100145137882e-06,
+      "loss": 7.0229,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.564586357039188e-06,
+      "loss": 6.824,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.550072568940494e-06,
+      "loss": 6.8623,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.5355587808418e-06,
+      "loss": 6.7549,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.521044992743107e-06,
+      "loss": 6.9337,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.506531204644413e-06,
+      "loss": 7.0024,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.49201741654572e-06,
+      "loss": 7.0414,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.477503628447025e-06,
+      "loss": 6.8472,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.462989840348332e-06,
+      "loss": 6.6774,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.448476052249638e-06,
+      "loss": 7.047,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 6.8967,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.41944847605225e-06,
+      "loss": 6.7912,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.404934687953556e-06,
+      "loss": 6.5542,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.390420899854863e-06,
+      "loss": 6.7136,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.375907111756169e-06,
+      "loss": 6.6257,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.361393323657475e-06,
+      "loss": 6.5128,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.346879535558781e-06,
+      "loss": 7.1436,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.332365747460089e-06,
+      "loss": 7.0321,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.317851959361393e-06,
+      "loss": 7.0403,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.3033381712627e-06,
+      "loss": 6.9365,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.288824383164008e-06,
+      "loss": 7.0607,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.274310595065312e-06,
+      "loss": 6.9458,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.259796806966618e-06,
+      "loss": 6.969,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 6.8947,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 7.0284,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.216255442670537e-06,
+      "loss": 6.8622,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.201741654571845e-06,
+      "loss": 7.1464,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.187227866473151e-06,
+      "loss": 6.7642,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.172714078374455e-06,
+      "loss": 7.0256,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.158200290275763e-06,
+      "loss": 6.6748,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.14368650217707e-06,
+      "loss": 6.8414,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.129172714078376e-06,
+      "loss": 6.9019,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.114658925979682e-06,
+      "loss": 6.8464,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.100145137880988e-06,
+      "loss": 6.8911,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.085631349782294e-06,
+      "loss": 6.8572,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.0711175616836e-06,
+      "loss": 6.8903,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.056603773584907e-06,
+      "loss": 6.9655,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.042089985486213e-06,
+      "loss": 6.9244,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.027576197387519e-06,
+      "loss": 6.8847,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.013062409288825e-06,
+      "loss": 6.9668,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.998548621190131e-06,
+      "loss": 7.1118,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.984034833091438e-06,
+      "loss": 7.0144,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.969521044992744e-06,
+      "loss": 6.9883,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.95500725689405e-06,
+      "loss": 6.9234,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.940493468795356e-06,
+      "loss": 6.842,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.925979680696662e-06,
+      "loss": 7.0538,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.911465892597968e-06,
+      "loss": 7.0362,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.896952104499275e-06,
+      "loss": 6.834,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.882438316400581e-06,
+      "loss": 6.8916,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.867924528301887e-06,
+      "loss": 6.827,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.853410740203193e-06,
+      "loss": 7.0698,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.8388969521045e-06,
+      "loss": 6.9355,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.824383164005806e-06,
+      "loss": 7.0138,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.809869375907114e-06,
+      "loss": 6.8661,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.795355587808418e-06,
+      "loss": 6.7788,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.780841799709724e-06,
+      "loss": 6.8941,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.766328011611032e-06,
+      "loss": 6.9536,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.751814223512337e-06,
+      "loss": 6.824,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.737300435413644e-06,
+      "loss": 6.98,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.72278664731495e-06,
+      "loss": 6.7571,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 6.694,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.693759071117563e-06,
+      "loss": 6.7125,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.67924528301887e-06,
+      "loss": 6.6153,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.664731494920174e-06,
+      "loss": 6.6813,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.650217706821482e-06,
+      "loss": 6.5217,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.635703918722788e-06,
+      "loss": 6.3799,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.621190130624092e-06,
+      "loss": 6.9017,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.6066763425254e-06,
+      "loss": 7.0753,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.592162554426706e-06,
+      "loss": 7.1271,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.577648766328013e-06,
+      "loss": 6.9628,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.563134978229319e-06,
+      "loss": 7.0422,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.548621190130625e-06,
+      "loss": 7.1291,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.534107402031931e-06,
+      "loss": 6.8313,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.519593613933237e-06,
+      "loss": 6.9726,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.505079825834544e-06,
+      "loss": 6.9317,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 6.963,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.476052249637156e-06,
+      "loss": 7.0122,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 6.8501,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.447024673439768e-06,
+      "loss": 6.8686,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.432510885341074e-06,
+      "loss": 6.8321,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.41799709724238e-06,
+      "loss": 6.9304,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.403483309143687e-06,
+      "loss": 6.7895,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.388969521044995e-06,
+      "loss": 6.9016,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.3744557329463e-06,
+      "loss": 6.8863,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.359941944847605e-06,
+      "loss": 6.9302,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.345428156748913e-06,
+      "loss": 7.0617,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.330914368650218e-06,
+      "loss": 6.7005,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.316400580551524e-06,
+      "loss": 6.8083,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.301886792452832e-06,
+      "loss": 6.8251,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.287373004354136e-06,
+      "loss": 6.8973,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.272859216255443e-06,
+      "loss": 6.9695,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.25834542815675e-06,
+      "loss": 7.0215,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.243831640058055e-06,
+      "loss": 6.965,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.229317851959361e-06,
+      "loss": 6.7352,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.214804063860669e-06,
+      "loss": 6.811,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.200290275761974e-06,
+      "loss": 6.8311,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.185776487663281e-06,
+      "loss": 6.9017,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.171262699564588e-06,
+      "loss": 6.9718,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.156748911465894e-06,
+      "loss": 6.8392,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.1422351233672e-06,
+      "loss": 6.985,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.127721335268506e-06,
+      "loss": 6.9158,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.113207547169812e-06,
+      "loss": 6.9385,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.098693759071119e-06,
+      "loss": 6.8239,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.084179970972425e-06,
+      "loss": 6.8259,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.069666182873731e-06,
+      "loss": 6.8866,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.055152394775037e-06,
+      "loss": 7.0857,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.040638606676343e-06,
+      "loss": 6.7744,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.02612481857765e-06,
+      "loss": 6.7829,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.011611030478956e-06,
+      "loss": 7.0232,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.997097242380262e-06,
+      "loss": 6.8237,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.982583454281568e-06,
+      "loss": 6.7989,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.968069666182874e-06,
+      "loss": 6.7534,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.95355587808418e-06,
+      "loss": 6.3743,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.939042089985487e-06,
+      "loss": 6.8223,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 6.8269,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.910014513788099e-06,
+      "loss": 6.328,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.895500725689405e-06,
+      "loss": 7.1547,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.880986937590711e-06,
+      "loss": 6.9365,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.866473149492018e-06,
+      "loss": 7.0605,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.851959361393324e-06,
+      "loss": 7.0724,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.837445573294632e-06,
+      "loss": 7.0274,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.822931785195936e-06,
+      "loss": 6.9956,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.808417997097242e-06,
+      "loss": 6.8714,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.79390420899855e-06,
+      "loss": 6.9926,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.779390420899855e-06,
+      "loss": 6.8888,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.764876632801161e-06,
+      "loss": 6.914,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.750362844702469e-06,
+      "loss": 6.8434,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.735849056603775e-06,
+      "loss": 6.9537,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.72133526850508e-06,
+      "loss": 6.714,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.706821480406387e-06,
+      "loss": 6.9964,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 6.9463,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.677793904208998e-06,
+      "loss": 6.8083,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.663280116110306e-06,
+      "loss": 6.9482,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.648766328011612e-06,
+      "loss": 6.8766,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.634252539912917e-06,
+      "loss": 7.1899,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.619738751814224e-06,
+      "loss": 7.0247,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.605224963715531e-06,
+      "loss": 6.8884,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.590711175616836e-06,
+      "loss": 6.8426,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.576197387518143e-06,
+      "loss": 6.8704,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.561683599419449e-06,
+      "loss": 6.836,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.5471698113207555e-06,
+      "loss": 6.998,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.532656023222062e-06,
+      "loss": 6.8403,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.518142235123368e-06,
+      "loss": 7.0563,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.503628447024675e-06,
+      "loss": 7.0132,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.48911465892598e-06,
+      "loss": 6.9056,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.4746008708272864e-06,
+      "loss": 7.0372,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.4600870827285935e-06,
+      "loss": 6.7229,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.445573294629899e-06,
+      "loss": 6.7849,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.431059506531205e-06,
+      "loss": 6.854,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.416545718432512e-06,
+      "loss": 6.8085,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.402031930333817e-06,
+      "loss": 6.8613,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.387518142235124e-06,
+      "loss": 6.8526,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.373004354136431e-06,
+      "loss": 6.9369,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 6.8889,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.343976777939042e-06,
+      "loss": 7.065,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.329462989840349e-06,
+      "loss": 6.8501,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.314949201741655e-06,
+      "loss": 7.0383,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.300435413642961e-06,
+      "loss": 6.7872,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.285921625544268e-06,
+      "loss": 6.7519,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.271407837445574e-06,
+      "loss": 6.9189,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.25689404934688e-06,
+      "loss": 6.8036,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.242380261248186e-06,
+      "loss": 6.5608,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.227866473149493e-06,
+      "loss": 6.9249,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.213352685050799e-06,
+      "loss": 6.8296,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.198838896952105e-06,
+      "loss": 6.4702,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.184325108853412e-06,
+      "loss": 6.7627,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 7.0823,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1552975326560235e-06,
+      "loss": 6.8861,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1407837445573305e-06,
+      "loss": 7.0545,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.126269956458637e-06,
+      "loss": 6.8919,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.111756168359942e-06,
+      "loss": 6.9492,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.097242380261249e-06,
+      "loss": 6.9989,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.082728592162555e-06,
+      "loss": 6.9493,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.068214804063861e-06,
+      "loss": 6.849,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.053701015965168e-06,
+      "loss": 6.9369,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.039187227866474e-06,
+      "loss": 7.0079,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.024673439767779e-06,
+      "loss": 6.8203,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.010159651669086e-06,
+      "loss": 7.0967,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9956458635703924e-06,
+      "loss": 6.9258,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.981132075471699e-06,
+      "loss": 6.9394,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.966618287373005e-06,
+      "loss": 6.8369,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.952104499274311e-06,
+      "loss": 6.846,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937590711175617e-06,
+      "loss": 6.8751,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 6.976,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.90856313497823e-06,
+      "loss": 6.9466,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894049346879537e-06,
+      "loss": 6.878,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879535558780842e-06,
+      "loss": 6.7352,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865021770682149e-06,
+      "loss": 6.8065,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850507982583455e-06,
+      "loss": 6.8175,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8359941944847605e-06,
+      "loss": 6.7807,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8214804063860676e-06,
+      "loss": 6.8933,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.806966618287374e-06,
+      "loss": 6.8557,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.792452830188679e-06,
+      "loss": 6.8194,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.777939042089986e-06,
+      "loss": 6.9714,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.763425253991292e-06,
+      "loss": 6.8809,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.748911465892598e-06,
+      "loss": 6.9403,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.734397677793905e-06,
+      "loss": 6.8424,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719883889695211e-06,
+      "loss": 6.9194,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705370101596517e-06,
+      "loss": 6.9287,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690856313497823e-06,
+      "loss": 6.8134,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6763425253991295e-06,
+      "loss": 6.9138,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6618287373004365e-06,
+      "loss": 6.8621,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647314949201742e-06,
+      "loss": 7.0345,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632801161103049e-06,
+      "loss": 7.0137,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618287373004355e-06,
+      "loss": 7.0006,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 6.861,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5892597968069675e-06,
+      "loss": 7.02,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574746008708274e-06,
+      "loss": 6.9999,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560232220609579e-06,
+      "loss": 6.9767,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.545718432510886e-06,
+      "loss": 6.7665,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.531204644412192e-06,
+      "loss": 6.7314,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5166908563134976e-06,
+      "loss": 6.6778,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502177068214805e-06,
+      "loss": 6.7894,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487663280116111e-06,
+      "loss": 7.0063,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473149492017418e-06,
+      "loss": 6.5721,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458635703918723e-06,
+      "loss": 6.2697,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444121915820029e-06,
+      "loss": 6.9765,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429608127721336e-06,
+      "loss": 7.1245,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415094339622642e-06,
+      "loss": 6.9324,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400580551523948e-06,
+      "loss": 6.9144,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386066763425255e-06,
+      "loss": 7.0733,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.37155297532656e-06,
+      "loss": 6.9949,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.357039187227867e-06,
+      "loss": 6.9225,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.3425253991291735e-06,
+      "loss": 7.0314,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.328011611030479e-06,
+      "loss": 6.7945,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.313497822931786e-06,
+      "loss": 6.964,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.298984034833092e-06,
+      "loss": 6.9162,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2844702467343975e-06,
+      "loss": 6.7735,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2699564586357045e-06,
+      "loss": 6.85,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.255442670537011e-06,
+      "loss": 6.7586,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.240928882438318e-06,
+      "loss": 7.1247,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 6.9207,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211901306240929e-06,
+      "loss": 6.9186,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.197387518142236e-06,
+      "loss": 6.7365,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.182873730043542e-06,
+      "loss": 6.8695,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.168359941944848e-06,
+      "loss": 6.9282,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 6.8974,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.13933236574746e-06,
+      "loss": 7.0249,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.124818577648766e-06,
+      "loss": 6.8865,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.1103047895500734e-06,
+      "loss": 6.7236,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.095791001451379e-06,
+      "loss": 6.9998,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.081277213352685e-06,
+      "loss": 6.8417,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.066763425253992e-06,
+      "loss": 6.8121,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.052249637155298e-06,
+      "loss": 6.8639,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.037735849056604e-06,
+      "loss": 6.927,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.023222060957911e-06,
+      "loss": 7.0968,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.008708272859218e-06,
+      "loss": 6.9577,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.994194484760523e-06,
+      "loss": 6.8211,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.979680696661829e-06,
+      "loss": 6.6943,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.965166908563136e-06,
+      "loss": 6.9421,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9506531204644415e-06,
+      "loss": 7.0147,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.936139332365748e-06,
+      "loss": 6.7209,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.921625544267055e-06,
+      "loss": 7.0326,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.90711175616836e-06,
+      "loss": 6.827,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.892597968069666e-06,
+      "loss": 6.8603,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.878084179970973e-06,
+      "loss": 6.7833,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.863570391872279e-06,
+      "loss": 6.9714,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.849056603773585e-06,
+      "loss": 6.8076,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.834542815674892e-06,
+      "loss": 6.7926,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.820029027576198e-06,
+      "loss": 6.7087,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8055152394775034e-06,
+      "loss": 6.6682,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7910014513788105e-06,
+      "loss": 6.6057,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.776487663280117e-06,
+      "loss": 6.6032,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.761973875181423e-06,
+      "loss": 6.586,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.747460087082729e-06,
+      "loss": 6.5468,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.732946298984035e-06,
+      "loss": 6.3228,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7184325108853414e-06,
+      "loss": 6.9353,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.703918722786648e-06,
+      "loss": 6.8317,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.689404934687955e-06,
+      "loss": 6.8783,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.67489114658926e-06,
+      "loss": 6.9472,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 6.9673,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.645863570391873e-06,
+      "loss": 6.8245,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6313497822931794e-06,
+      "loss": 6.8775,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.616835994194485e-06,
+      "loss": 6.939,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.602322206095792e-06,
+      "loss": 6.9086,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587808417997098e-06,
+      "loss": 6.9131,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.573294629898403e-06,
+      "loss": 6.8137,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.55878084179971e-06,
+      "loss": 6.8864,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5442670537010166e-06,
+      "loss": 6.7745,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.529753265602322e-06,
+      "loss": 7.0997,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.515239477503629e-06,
+      "loss": 7.0315,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.500725689404935e-06,
+      "loss": 6.7453,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.486211901306241e-06,
+      "loss": 6.8937,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4716981132075475e-06,
+      "loss": 6.6849,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.457184325108854e-06,
+      "loss": 6.8958,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.44267053701016e-06,
+      "loss": 6.9158,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.428156748911466e-06,
+      "loss": 6.8899,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.413642960812773e-06,
+      "loss": 6.8385,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.399129172714079e-06,
+      "loss": 6.9148,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 6.996,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.370101596516692e-06,
+      "loss": 6.8569,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.355587808417998e-06,
+      "loss": 6.8315,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.341074020319303e-06,
+      "loss": 6.9643,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.32656023222061e-06,
+      "loss": 6.9543,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.3120464441219165e-06,
+      "loss": 6.8956,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.297532656023222e-06,
+      "loss": 7.0792,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 6.9263,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.268505079825835e-06,
+      "loss": 6.7032,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.25399129172714e-06,
+      "loss": 6.7837,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.239477503628447e-06,
+      "loss": 6.9698,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.224963715529754e-06,
+      "loss": 6.9227,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.210449927431061e-06,
+      "loss": 7.0405,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.195936139332366e-06,
+      "loss": 6.9694,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.181422351233672e-06,
+      "loss": 6.6384,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.166908563134979e-06,
+      "loss": 7.0486,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1523947750362846e-06,
+      "loss": 6.8472,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.137880986937592e-06,
+      "loss": 6.9925,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.123367198838898e-06,
+      "loss": 6.7869,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108853410740203e-06,
+      "loss": 7.0557,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.09433962264151e-06,
+      "loss": 6.645,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.079825834542816e-06,
+      "loss": 6.8268,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.065312046444122e-06,
+      "loss": 6.753,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.050798258345429e-06,
+      "loss": 6.7694,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.036284470246735e-06,
+      "loss": 6.4818,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.02177068214804e-06,
+      "loss": 6.8192,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.007256894049347e-06,
+      "loss": 6.6748,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9927431059506535e-06,
+      "loss": 7.1003,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.97822931785196e-06,
+      "loss": 6.9059,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.963715529753266e-06,
+      "loss": 6.9902,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.949201741654572e-06,
+      "loss": 7.0545,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.934687953555878e-06,
+      "loss": 6.9411,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.920174165457185e-06,
+      "loss": 6.9627,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.905660377358491e-06,
+      "loss": 6.7708,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.891146589259797e-06,
+      "loss": 6.9756,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.876632801161104e-06,
+      "loss": 6.7789,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.862119013062409e-06,
+      "loss": 6.8748,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.847605224963715e-06,
+      "loss": 6.9197,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8330914368650224e-06,
+      "loss": 6.9325,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.818577648766329e-06,
+      "loss": 6.739,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.804063860667635e-06,
+      "loss": 6.9291,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.789550072568941e-06,
+      "loss": 6.9372,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.775036284470247e-06,
+      "loss": 6.8033,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.760522496371553e-06,
+      "loss": 6.766,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.74600870827286e-06,
+      "loss": 6.7955,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.731494920174166e-06,
+      "loss": 6.7609,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 6.7356,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.702467343976778e-06,
+      "loss": 6.8248,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.687953555878084e-06,
+      "loss": 6.6899,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6734397677793905e-06,
+      "loss": 7.0141,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.658925979680697e-06,
+      "loss": 6.9566,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.644412191582004e-06,
+      "loss": 6.9616,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.629898403483309e-06,
+      "loss": 6.7778,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 6.7939,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.600870827285922e-06,
+      "loss": 6.9759,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.586357039187228e-06,
+      "loss": 6.8179,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.571843251088535e-06,
+      "loss": 6.9923,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.557329462989841e-06,
+      "loss": 6.8301,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.542815674891147e-06,
+      "loss": 6.9272,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 6.9453,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5137880986937595e-06,
+      "loss": 6.9743,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.499274310595066e-06,
+      "loss": 6.7763,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.484760522496372e-06,
+      "loss": 6.9135,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.470246734397678e-06,
+      "loss": 6.8415,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.455732946298984e-06,
+      "loss": 6.8536,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.4412191582002904e-06,
+      "loss": 6.9403,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.426705370101597e-06,
+      "loss": 6.7382,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.412191582002903e-06,
+      "loss": 6.9703,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.397677793904209e-06,
+      "loss": 6.8839,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.383164005805516e-06,
+      "loss": 6.7099,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.368650217706822e-06,
+      "loss": 6.7211,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 6.7592,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.339622641509435e-06,
+      "loss": 6.7586,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.325108853410741e-06,
+      "loss": 6.5655,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.310595065312046e-06,
+      "loss": 6.3672,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.296081277213353e-06,
+      "loss": 6.5872,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.281567489114659e-06,
+      "loss": 6.209,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2670537010159656e-06,
+      "loss": 7.0228,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252539912917272e-06,
+      "loss": 6.9513,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.238026124818578e-06,
+      "loss": 6.9614,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.223512336719884e-06,
+      "loss": 6.7462,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.20899854862119e-06,
+      "loss": 6.932,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.194484760522497e-06,
+      "loss": 6.9618,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.179970972423803e-06,
+      "loss": 6.8221,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.165457184325109e-06,
+      "loss": 7.196,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.150943396226416e-06,
+      "loss": 6.9068,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.136429608127721e-06,
+      "loss": 6.92,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1219158200290275e-06,
+      "loss": 6.8494,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1074020319303345e-06,
+      "loss": 6.9218,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.092888243831641e-06,
+      "loss": 6.8718,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.078374455732947e-06,
+      "loss": 6.7455,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.063860667634253e-06,
+      "loss": 6.8827,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.049346879535559e-06,
+      "loss": 6.9519,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0348330914368655e-06,
+      "loss": 6.8735,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.020319303338172e-06,
+      "loss": 6.8876,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.005805515239478e-06,
+      "loss": 6.9762,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.991291727140784e-06,
+      "loss": 6.9191,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.97677793904209e-06,
+      "loss": 6.8247,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 7.0048,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.947750362844703e-06,
+      "loss": 6.9299,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.933236574746009e-06,
+      "loss": 6.9704,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.918722786647316e-06,
+      "loss": 6.768,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.904208998548621e-06,
+      "loss": 6.8617,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.889695210449927e-06,
+      "loss": 6.9783,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.875181422351234e-06,
+      "loss": 6.9152,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.86066763425254e-06,
+      "loss": 6.7962,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 6.8732,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.831640058055153e-06,
+      "loss": 6.9002,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.817126269956458e-06,
+      "loss": 6.9278,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.8026124818577654e-06,
+      "loss": 7.0025,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7880986937590716e-06,
+      "loss": 6.886,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7735849056603777e-06,
+      "loss": 6.8703,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.759071117561684e-06,
+      "loss": 6.898,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.74455732946299e-06,
+      "loss": 6.8803,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7300435413642967e-06,
+      "loss": 7.0353,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7155297532656025e-06,
+      "loss": 6.874,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7010159651669087e-06,
+      "loss": 6.7774,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6865021770682153e-06,
+      "loss": 6.7775,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.671988388969521e-06,
+      "loss": 6.9426,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6574746008708277e-06,
+      "loss": 6.6207,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.642960812772134e-06,
+      "loss": 6.5535,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.62844702467344e-06,
+      "loss": 6.9448,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6139332365747467e-06,
+      "loss": 6.6878,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5994194484760525e-06,
+      "loss": 6.574,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5849056603773586e-06,
+      "loss": 6.4714,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5703918722786653e-06,
+      "loss": 6.4663,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.555878084179971e-06,
+      "loss": 6.4119,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5413642960812776e-06,
+      "loss": 6.9545,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.526850507982584e-06,
+      "loss": 6.9492,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5123367198838896e-06,
+      "loss": 6.8786,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4978229317851962e-06,
+      "loss": 6.7108,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4833091436865024e-06,
+      "loss": 6.8923,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4687953555878086e-06,
+      "loss": 7.0149,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.454281567489115e-06,
+      "loss": 6.8889,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.439767779390421e-06,
+      "loss": 7.0583,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4252539912917276e-06,
+      "loss": 6.919,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4107402031930338e-06,
+      "loss": 6.7794,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 6.8407,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.381712626995646e-06,
+      "loss": 6.7961,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3671988388969524e-06,
+      "loss": 6.7505,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3526850507982585e-06,
+      "loss": 6.8841,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3381712626995647e-06,
+      "loss": 6.9314,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.323657474600871e-06,
+      "loss": 6.8862,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3091436865021775e-06,
+      "loss": 7.0835,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2946298984034837e-06,
+      "loss": 6.9177,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2801161103047895e-06,
+      "loss": 6.6316,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.265602322206096e-06,
+      "loss": 6.9209,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2510885341074023e-06,
+      "loss": 6.7641,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.236574746008709e-06,
+      "loss": 6.7326,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2220609579100147e-06,
+      "loss": 6.808,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207547169811321e-06,
+      "loss": 6.8766,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1930333817126275e-06,
+      "loss": 6.9098,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1785195936139337e-06,
+      "loss": 6.9424,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1640058055152394e-06,
+      "loss": 6.8173,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.149492017416546e-06,
+      "loss": 6.8762,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1349782293178522e-06,
+      "loss": 6.79,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.120464441219159e-06,
+      "loss": 6.8736,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1059506531204646e-06,
+      "loss": 7.1317,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.091436865021771e-06,
+      "loss": 6.9109,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 6.9198,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.062409288824383e-06,
+      "loss": 7.0885,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0478955007256894e-06,
+      "loss": 6.7748,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.033381712626996e-06,
+      "loss": 6.8582,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 6.9278,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.004354136429609e-06,
+      "loss": 6.8761,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9898403483309146e-06,
+      "loss": 6.8446,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9753265602322208e-06,
+      "loss": 6.7847,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9608127721335274e-06,
+      "loss": 6.8721,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.946298984034833e-06,
+      "loss": 6.8442,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9317851959361393e-06,
+      "loss": 6.7476,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.917271407837446e-06,
+      "loss": 6.9989,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9027576197387517e-06,
+      "loss": 6.818,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8882438316400583e-06,
+      "loss": 7.1432,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8737300435413645e-06,
+      "loss": 6.8235,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8592162554426707e-06,
+      "loss": 6.653,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8447024673439773e-06,
+      "loss": 6.6371,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 6.495,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 6.767850399017334,
+      "eval_runtime": 769.9568,
+      "eval_samples_per_second": 3.431,
+      "eval_steps_per_second": 0.287,
+      "eval_wer": 1.4752663849038634,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8156748911465897e-06,
+      "loss": 6.8706,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.801161103047896e-06,
+      "loss": 6.9277,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7866473149492017e-06,
+      "loss": 6.8924,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7721335268505083e-06,
+      "loss": 6.9364,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7576197387518145e-06,
+      "loss": 6.8358,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7431059506531207e-06,
+      "loss": 6.9334,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.728592162554427e-06,
+      "loss": 6.829,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.714078374455733e-06,
+      "loss": 6.6969,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6995645863570397e-06,
+      "loss": 6.9654,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.685050798258346e-06,
+      "loss": 6.7021,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6705370101596516e-06,
+      "loss": 6.8105,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6560232220609582e-06,
+      "loss": 6.8654,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6415094339622644e-06,
+      "loss": 6.7323,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.62699564586357e-06,
+      "loss": 6.7648,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.612481857764877e-06,
+      "loss": 6.7337,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.597968069666183e-06,
+      "loss": 6.9257,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5834542815674896e-06,
+      "loss": 6.7598,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.568940493468796e-06,
+      "loss": 6.8418,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5544267053701016e-06,
+      "loss": 6.9608,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.539912917271408e-06,
+      "loss": 6.7484,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5253991291727144e-06,
+      "loss": 6.9334,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.51088534107402e-06,
+      "loss": 6.9865,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4963715529753268e-06,
+      "loss": 7.0334,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.481857764876633e-06,
+      "loss": 6.8062,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.467343976777939e-06,
+      "loss": 6.7737,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 6.7982,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.438316400580552e-06,
+      "loss": 6.7906,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4238026124818577e-06,
+      "loss": 6.8239,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4092888243831643e-06,
+      "loss": 6.8813,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3947750362844705e-06,
+      "loss": 6.8053,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3802612481857767e-06,
+      "loss": 6.9033,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.365747460087083e-06,
+      "loss": 6.722,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.351233671988389e-06,
+      "loss": 6.8812,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3367198838896953e-06,
+      "loss": 6.8061,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.322206095791002e-06,
+      "loss": 6.8649,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 6.9663,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.293178519593614e-06,
+      "loss": 6.6947,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2786647314949205e-06,
+      "loss": 6.9788,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2641509433962266e-06,
+      "loss": 6.8507,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.249637155297533e-06,
+      "loss": 6.7087,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.235123367198839e-06,
+      "loss": 6.7418,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2206095791001452e-06,
+      "loss": 6.7797,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2060957910014514e-06,
+      "loss": 6.7308,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.191582002902758e-06,
+      "loss": 6.7953,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.177068214804064e-06,
+      "loss": 6.7766,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1625544267053704e-06,
+      "loss": 6.7918,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1480406386066766e-06,
+      "loss": 6.5738,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1335268505079828e-06,
+      "loss": 6.8215,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.119013062409289e-06,
+      "loss": 6.8907,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.104499274310595e-06,
+      "loss": 6.4309,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0899854862119014e-06,
+      "loss": 6.9973,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.075471698113208e-06,
+      "loss": 6.8308,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0609579100145137e-06,
+      "loss": 6.8409,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0464441219158204e-06,
+      "loss": 6.9626,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0319303338171265e-06,
+      "loss": 6.942,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0174165457184327e-06,
+      "loss": 6.9222,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.002902757619739e-06,
+      "loss": 6.9238,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.988388969521045e-06,
+      "loss": 6.7578,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9738751814223513e-06,
+      "loss": 6.9271,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.959361393323658e-06,
+      "loss": 6.9286,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9448476052249637e-06,
+      "loss": 6.9462,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.93033381712627e-06,
+      "loss": 6.8382,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9158200290275765e-06,
+      "loss": 6.9456,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9013062409288827e-06,
+      "loss": 6.8105,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8867924528301889e-06,
+      "loss": 6.8244,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.872278664731495e-06,
+      "loss": 6.8595,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8577648766328013e-06,
+      "loss": 6.7694,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8432510885341077e-06,
+      "loss": 6.8864,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8287373004354138e-06,
+      "loss": 6.9903,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.81422351233672e-06,
+      "loss": 6.9048,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7997097242380262e-06,
+      "loss": 6.7377,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7851959361393326e-06,
+      "loss": 6.8922,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7706821480406388e-06,
+      "loss": 6.6749,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7561683599419448e-06,
+      "loss": 6.8178,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7416545718432512e-06,
+      "loss": 6.5481,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7271407837445576e-06,
+      "loss": 6.9033,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7126269956458638e-06,
+      "loss": 7.0286,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6981132075471698e-06,
+      "loss": 6.935,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6835994194484762e-06,
+      "loss": 6.7556,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6690856313497824e-06,
+      "loss": 6.8288,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6545718432510888e-06,
+      "loss": 6.8435,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6400580551523947e-06,
+      "loss": 7.1217,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6255442670537012e-06,
+      "loss": 6.6545,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6110304789550073e-06,
+      "loss": 6.8334,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5965166908563137e-06,
+      "loss": 6.8292,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5820029027576197e-06,
+      "loss": 6.876,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5674891146589261e-06,
+      "loss": 6.8266,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5529753265602323e-06,
+      "loss": 6.8529,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 6.8762,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5239477503628447e-06,
+      "loss": 6.7678,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 6.8384,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4949201741654573e-06,
+      "loss": 6.6931,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4804063860667637e-06,
+      "loss": 6.7146,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4658925979680697e-06,
+      "loss": 6.9039,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4513788098693759e-06,
+      "loss": 6.6635,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4368650217706823e-06,
+      "loss": 6.8665,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4223512336719887e-06,
+      "loss": 6.6563,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4078374455732949e-06,
+      "loss": 6.8987,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3933236574746008e-06,
+      "loss": 6.8854,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3788098693759072e-06,
+      "loss": 6.1351,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3642960812772134e-06,
+      "loss": 6.8762,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3497822931785198e-06,
+      "loss": 6.9102,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3352685050798258e-06,
+      "loss": 6.9588,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3207547169811322e-06,
+      "loss": 6.7828,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3062409288824384e-06,
+      "loss": 6.7559,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2917271407837448e-06,
+      "loss": 6.7355,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2772133526850508e-06,
+      "loss": 7.061,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2626995645863572e-06,
+      "loss": 7.031,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2481857764876634e-06,
+      "loss": 6.8649,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2336719883889696e-06,
+      "loss": 6.7787,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.219158200290276e-06,
+      "loss": 6.794,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2046444121915822e-06,
+      "loss": 6.9153,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1901306240928884e-06,
+      "loss": 6.692,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1756168359941945e-06,
+      "loss": 6.7696,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.161103047895501e-06,
+      "loss": 6.9986,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.146589259796807e-06,
+      "loss": 6.9248,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1320754716981133e-06,
+      "loss": 6.9554,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1175616835994195e-06,
+      "loss": 6.8741,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1030478955007257e-06,
+      "loss": 6.6611,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.088534107402032e-06,
+      "loss": 6.9804,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0740203193033383e-06,
+      "loss": 6.8872,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0595065312046445e-06,
+      "loss": 7.0589,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0449927431059507e-06,
+      "loss": 6.8507,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0304789550072569e-06,
+      "loss": 6.9875,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0159651669085633e-06,
+      "loss": 7.0288,
+      "step": 1125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0014513788098695e-06,
+      "loss": 6.798,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.869375907111757e-07,
+      "loss": 6.9097,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.724238026124818e-07,
+      "loss": 6.7965,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.579100145137882e-07,
+      "loss": 6.8586,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.433962264150944e-07,
+      "loss": 6.8131,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.288824383164006e-07,
+      "loss": 6.7052,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.143686502177069e-07,
+      "loss": 6.6866,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.998548621190131e-07,
+      "loss": 6.982,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.853410740203194e-07,
+      "loss": 6.8548,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.708272859216256e-07,
+      "loss": 6.9743,
+      "step": 1135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.563134978229319e-07,
+      "loss": 6.8077,
+      "step": 1136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.417997097242381e-07,
+      "loss": 6.8983,
+      "step": 1137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.272859216255444e-07,
+      "loss": 6.8462,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.127721335268506e-07,
+      "loss": 6.7949,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.982583454281569e-07,
+      "loss": 6.7141,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.837445573294631e-07,
+      "loss": 6.9175,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 6.8364,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 6.814,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.402031930333818e-07,
+      "loss": 6.8656,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.256894049346879e-07,
+      "loss": 6.6052,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111756168359943e-07,
+      "loss": 6.9527,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.966618287373004e-07,
+      "loss": 6.6223,
+      "step": 1147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.821480406386067e-07,
+      "loss": 6.5225,
+      "step": 1148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.676342525399129e-07,
+      "loss": 6.7791,
+      "step": 1149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.531204644412192e-07,
+      "loss": 6.5205,
+      "step": 1150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.386066763425254e-07,
+      "loss": 6.9171,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.240928882438317e-07,
+      "loss": 6.7792,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.09579100145138e-07,
+      "loss": 7.0013,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.950653120464442e-07,
+      "loss": 6.8375,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.805515239477505e-07,
+      "loss": 6.868,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.660377358490567e-07,
+      "loss": 7.0463,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.515239477503629e-07,
+      "loss": 6.825,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.370101596516691e-07,
+      "loss": 6.7783,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.224963715529753e-07,
+      "loss": 6.8332,
+      "step": 1159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.079825834542816e-07,
+      "loss": 6.872,
+      "step": 1160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.934687953555878e-07,
+      "loss": 7.0363,
+      "step": 1161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.789550072568941e-07,
+      "loss": 6.8874,
+      "step": 1162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644412191582003e-07,
+      "loss": 6.994,
+      "step": 1163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4992743105950656e-07,
+      "loss": 6.8419,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.354136429608128e-07,
+      "loss": 6.8114,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2089985486211904e-07,
+      "loss": 7.0351,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.063860667634253e-07,
+      "loss": 6.6599,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9187227866473153e-07,
+      "loss": 6.8608,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 6.7875,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6284470246734397e-07,
+      "loss": 6.7666,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.483309143686502e-07,
+      "loss": 6.8301,
+      "step": 1171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3381712626995645e-07,
+      "loss": 6.7421,
+      "step": 1172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.193033381712627e-07,
+      "loss": 6.9527,
+      "step": 1173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.04789550072569e-07,
+      "loss": 6.9352,
+      "step": 1174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9027576197387524e-07,
+      "loss": 6.6471,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.757619738751814e-07,
+      "loss": 6.6596,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6124818577648767e-07,
+      "loss": 6.8253,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.467343976777939e-07,
+      "loss": 6.9835,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3222060957910016e-07,
+      "loss": 6.597,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.177068214804064e-07,
+      "loss": 7.0269,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0319303338171264e-07,
+      "loss": 6.7609,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.886792452830189e-07,
+      "loss": 6.9373,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.741654571843251e-07,
+      "loss": 6.8113,
+      "step": 1183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5965166908563135e-07,
+      "loss": 6.7609,
+      "step": 1184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4513788098693762e-07,
+      "loss": 6.7291,
+      "step": 1185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3062409288824383e-07,
+      "loss": 6.6586,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1611030478955008e-07,
+      "loss": 6.6503,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0159651669085632e-07,
+      "loss": 6.545,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.708272859216255e-08,
+      "loss": 6.4964,
+      "step": 1189
+    },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 7.25599230030727,
-      "train_runtime": 4990.6204,
-      "train_samples_per_second": 5.718,
-      "train_steps_per_second": 0.089
+      "train_loss": 7.0317410712887805,
+      "train_runtime": 6576.7974,
+      "train_samples_per_second": 4.339,
+      "train_steps_per_second": 0.181
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,