diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,7 +2,7 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,5375 +10,2690 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 4.8105,
+      "loss": 4.7746,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.0644,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 4.9289,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 4.9238,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.8883,
       "step": 3
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.9152,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 4.8789,
       "step": 4
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.8e-07,
-      "loss": 4.7529,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.8296,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.8704,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.8764,
       "step": 6
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.8358,
+      "epoch": 0.02,
+      "learning_rate": 3.6e-07,
+      "loss": 4.8601,
       "step": 7
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.6e-07,
-      "loss": 4.7615,
+      "epoch": 0.02,
+      "learning_rate": 4.2e-07,
+      "loss": 4.9133,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.2e-07,
-      "loss": 4.9183,
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.916,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 4.8171,
+      "epoch": 0.02,
+      "learning_rate": 5.4e-07,
+      "loss": 4.7759,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 5.4e-07,
-      "loss": 4.882,
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.8958,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.8536,
+      "epoch": 0.03,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 4.8645,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.599999999999999e-07,
-      "loss": 4.8384,
+      "epoch": 0.03,
+      "learning_rate": 7.2e-07,
+      "loss": 4.8147,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.2e-07,
-      "loss": 4.7957,
+      "epoch": 0.03,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 4.8259,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.799999999999999e-07,
-      "loss": 4.7335,
+      "epoch": 0.03,
+      "learning_rate": 8.4e-07,
+      "loss": 4.7034,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.4e-07,
-      "loss": 4.8636,
+      "epoch": 0.04,
+      "learning_rate": 9e-07,
+      "loss": 4.8451,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 9e-07,
-      "loss": 4.8799,
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 4.7678,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 4.9855,
+      "epoch": 0.04,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 4.8715,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.0200000000000002e-06,
-      "loss": 4.6993,
+      "epoch": 0.04,
+      "learning_rate": 1.08e-06,
+      "loss": 4.6656,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.08e-06,
-      "loss": 4.8669,
+      "epoch": 0.04,
+      "learning_rate": 1.14e-06,
+      "loss": 4.8439,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.14e-06,
-      "loss": 4.796,
+      "epoch": 0.05,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.8241,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.7392,
+      "epoch": 0.05,
+      "learning_rate": 1.26e-06,
+      "loss": 4.7418,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.26e-06,
-      "loss": 4.7951,
+      "epoch": 0.05,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 4.8523,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.3199999999999999e-06,
-      "loss": 4.912,
+      "epoch": 0.05,
+      "learning_rate": 1.38e-06,
+      "loss": 4.8442,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.38e-06,
-      "loss": 4.7894,
+      "epoch": 0.06,
+      "learning_rate": 1.44e-06,
+      "loss": 4.6269,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.44e-06,
-      "loss": 4.9259,
+      "epoch": 0.06,
+      "learning_rate": 1.5e-06,
+      "loss": 4.8907,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.5e-06,
-      "loss": 4.6791,
+      "epoch": 0.06,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 4.7369,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.5599999999999999e-06,
-      "loss": 4.6056,
+      "epoch": 0.06,
+      "learning_rate": 1.62e-06,
+      "loss": 4.5892,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.62e-06,
-      "loss": 4.6111,
+      "epoch": 0.07,
+      "learning_rate": 1.68e-06,
+      "loss": 4.564,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.68e-06,
-      "loss": 4.7901,
+      "epoch": 0.07,
+      "learning_rate": 1.74e-06,
+      "loss": 4.7634,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.74e-06,
-      "loss": 4.7425,
+      "epoch": 0.07,
+      "learning_rate": 1.8e-06,
+      "loss": 4.735,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.8e-06,
-      "loss": 4.533,
+      "epoch": 0.07,
+      "learning_rate": 1.86e-06,
+      "loss": 4.5768,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.86e-06,
-      "loss": 4.637,
+      "epoch": 0.07,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 4.6719,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.4931,
+      "epoch": 0.08,
+      "learning_rate": 1.98e-06,
+      "loss": 4.5934,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.98e-06,
-      "loss": 4.7015,
+      "epoch": 0.08,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 4.6764,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.0400000000000004e-06,
-      "loss": 4.6452,
+      "epoch": 0.08,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.6146,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.6921,
+      "epoch": 0.08,
+      "learning_rate": 2.16e-06,
+      "loss": 4.6754,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.16e-06,
-      "loss": 4.6732,
+      "epoch": 0.09,
+      "learning_rate": 2.22e-06,
+      "loss": 4.6119,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.22e-06,
-      "loss": 4.8648,
+      "epoch": 0.09,
+      "learning_rate": 2.28e-06,
+      "loss": 4.7453,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.28e-06,
-      "loss": 4.6368,
+      "epoch": 0.09,
+      "learning_rate": 2.34e-06,
+      "loss": 4.5306,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.34e-06,
-      "loss": 4.4882,
+      "epoch": 0.09,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.595,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.6836,
+      "epoch": 0.09,
+      "learning_rate": 2.46e-06,
+      "loss": 4.7049,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.46e-06,
-      "loss": 4.4747,
+      "epoch": 0.1,
+      "learning_rate": 2.52e-06,
+      "loss": 4.4633,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.52e-06,
-      "loss": 4.6423,
+      "epoch": 0.1,
+      "learning_rate": 2.58e-06,
+      "loss": 4.6159,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.58e-06,
-      "loss": 4.7406,
+      "epoch": 0.1,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 4.7318,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.6399999999999997e-06,
-      "loss": 4.871,
+      "epoch": 0.1,
+      "learning_rate": 2.7e-06,
+      "loss": 4.8312,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.7e-06,
-      "loss": 4.756,
+      "epoch": 0.11,
+      "learning_rate": 2.76e-06,
+      "loss": 4.7198,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.76e-06,
-      "loss": 5.13,
+      "epoch": 0.11,
+      "learning_rate": 2.82e-06,
+      "loss": 4.9727,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.76e-06,
-      "loss": 5.2669,
+      "epoch": 0.11,
+      "learning_rate": 2.88e-06,
+      "loss": 5.057,
       "step": 49
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.82e-06,
-      "loss": 4.9543,
+      "epoch": 0.11,
+      "learning_rate": 2.88e-06,
+      "loss": 4.837,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.88e-06,
-      "loss": 4.3671,
+      "epoch": 0.11,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.4908,
       "step": 51
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.4949,
+      "epoch": 0.12,
+      "learning_rate": 3e-06,
+      "loss": 4.3346,
       "step": 52
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3e-06,
-      "loss": 4.3571,
+      "epoch": 0.12,
+      "learning_rate": 3.06e-06,
+      "loss": 4.5022,
       "step": 53
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.06e-06,
-      "loss": 4.4288,
+      "epoch": 0.12,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 4.4011,
       "step": 54
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.1199999999999998e-06,
-      "loss": 4.4257,
+      "epoch": 0.12,
+      "learning_rate": 3.18e-06,
+      "loss": 4.2133,
       "step": 55
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.18e-06,
-      "loss": 4.4883,
+      "epoch": 0.13,
+      "learning_rate": 3.24e-06,
+      "loss": 4.4502,
       "step": 56
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.24e-06,
-      "loss": 4.45,
+      "epoch": 0.13,
+      "learning_rate": 3.3e-06,
+      "loss": 4.4597,
       "step": 57
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.3e-06,
-      "loss": 4.6948,
+      "epoch": 0.13,
+      "learning_rate": 3.36e-06,
+      "loss": 4.5472,
       "step": 58
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.36e-06,
-      "loss": 4.429,
+      "epoch": 0.13,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 4.4784,
       "step": 59
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.4200000000000003e-06,
-      "loss": 4.286,
+      "epoch": 0.13,
+      "learning_rate": 3.48e-06,
+      "loss": 4.3192,
       "step": 60
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.48e-06,
-      "loss": 4.5093,
+      "epoch": 0.14,
+      "learning_rate": 3.54e-06,
+      "loss": 4.4132,
       "step": 61
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.54e-06,
-      "loss": 4.4479,
+      "epoch": 0.14,
+      "learning_rate": 3.6e-06,
+      "loss": 4.3723,
       "step": 62
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.6e-06,
-      "loss": 4.3196,
+      "epoch": 0.14,
+      "learning_rate": 3.66e-06,
+      "loss": 4.3143,
       "step": 63
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.66e-06,
-      "loss": 4.4664,
+      "epoch": 0.14,
+      "learning_rate": 3.72e-06,
+      "loss": 4.4111,
       "step": 64
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.72e-06,
-      "loss": 4.2118,
+      "epoch": 0.15,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.381,
       "step": 65
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.5129,
+      "epoch": 0.15,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.2726,
       "step": 66
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.2765,
+      "epoch": 0.15,
+      "learning_rate": 3.9e-06,
+      "loss": 4.4058,
       "step": 67
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.9e-06,
-      "loss": 4.3555,
+      "epoch": 0.15,
+      "learning_rate": 3.96e-06,
+      "loss": 4.269,
       "step": 68
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.96e-06,
-      "loss": 4.2651,
+      "epoch": 0.15,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.3351,
       "step": 69
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.3872,
+      "epoch": 0.16,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 4.2911,
       "step": 70
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.080000000000001e-06,
-      "loss": 4.5608,
+      "epoch": 0.16,
+      "learning_rate": 4.14e-06,
+      "loss": 4.434,
       "step": 71
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.14e-06,
-      "loss": 4.4479,
+      "epoch": 0.16,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.4303,
       "step": 72
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.3873,
+      "epoch": 0.16,
+      "learning_rate": 4.26e-06,
+      "loss": 4.2569,
       "step": 73
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.26e-06,
-      "loss": 4.4395,
+      "epoch": 0.17,
+      "learning_rate": 4.32e-06,
+      "loss": 4.3597,
       "step": 74
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.32e-06,
-      "loss": 4.2003,
+      "epoch": 0.17,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 4.3462,
       "step": 75
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.3799999999999996e-06,
-      "loss": 4.5215,
+      "epoch": 0.17,
+      "learning_rate": 4.44e-06,
+      "loss": 4.3001,
       "step": 76
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.44e-06,
-      "loss": 4.3927,
+      "epoch": 0.17,
+      "learning_rate": 4.5e-06,
+      "loss": 4.3445,
       "step": 77
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5e-06,
-      "loss": 4.3106,
+      "epoch": 0.17,
+      "learning_rate": 4.56e-06,
+      "loss": 4.2932,
       "step": 78
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.56e-06,
-      "loss": 4.1722,
+      "epoch": 0.18,
+      "learning_rate": 4.62e-06,
+      "loss": 4.267,
       "step": 79
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.62e-06,
-      "loss": 4.3752,
+      "epoch": 0.18,
+      "learning_rate": 4.68e-06,
+      "loss": 4.4182,
       "step": 80
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.68e-06,
-      "loss": 4.4628,
+      "epoch": 0.18,
+      "learning_rate": 4.74e-06,
+      "loss": 4.3713,
       "step": 81
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.74e-06,
-      "loss": 4.2682,
+      "epoch": 0.18,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.3061,
       "step": 82
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.4709,
+      "epoch": 0.19,
+      "learning_rate": 4.86e-06,
+      "loss": 4.4143,
       "step": 83
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.86e-06,
-      "loss": 4.4093,
+      "epoch": 0.19,
+      "learning_rate": 4.92e-06,
+      "loss": 4.2805,
       "step": 84
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.92e-06,
-      "loss": 4.3576,
+      "epoch": 0.19,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.3571,
       "step": 85
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.3302,
+      "epoch": 0.19,
+      "learning_rate": 5.04e-06,
+      "loss": 4.3305,
       "step": 86
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.04e-06,
-      "loss": 4.3482,
+      "epoch": 0.2,
+      "learning_rate": 5.1e-06,
+      "loss": 4.4209,
       "step": 87
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.1e-06,
-      "loss": 4.3374,
+      "epoch": 0.2,
+      "learning_rate": 5.16e-06,
+      "loss": 4.2915,
       "step": 88
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.16e-06,
-      "loss": 4.3922,
+      "epoch": 0.2,
+      "learning_rate": 5.22e-06,
+      "loss": 4.4348,
       "step": 89
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.22e-06,
-      "loss": 4.2921,
+      "epoch": 0.2,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 4.4178,
       "step": 90
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.279999999999999e-06,
-      "loss": 4.5432,
+      "epoch": 0.2,
+      "learning_rate": 5.34e-06,
+      "loss": 4.3064,
       "step": 91
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.34e-06,
-      "loss": 4.4631,
+      "epoch": 0.21,
+      "learning_rate": 5.4e-06,
+      "loss": 4.5617,
       "step": 92
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.4e-06,
-      "loss": 4.177,
+      "epoch": 0.21,
+      "learning_rate": 5.46e-06,
+      "loss": 4.4422,
       "step": 93
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.46e-06,
-      "loss": 4.3825,
+      "epoch": 0.21,
+      "learning_rate": 5.52e-06,
+      "loss": 4.4235,
       "step": 94
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.52e-06,
-      "loss": 4.6525,
+      "epoch": 0.21,
+      "learning_rate": 5.58e-06,
+      "loss": 4.3727,
       "step": 95
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.58e-06,
-      "loss": 4.5859,
+      "epoch": 0.22,
+      "learning_rate": 5.64e-06,
+      "loss": 4.5854,
       "step": 96
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.64e-06,
-      "loss": 4.3767,
+      "epoch": 0.22,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 4.5272,
       "step": 97
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 4.677,
+      "epoch": 0.22,
+      "learning_rate": 5.76e-06,
+      "loss": 4.3841,
       "step": 98
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.76e-06,
-      "loss": 4.4746,
+      "epoch": 0.22,
+      "learning_rate": 5.82e-06,
+      "loss": 4.711,
       "step": 99
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.82e-06,
-      "loss": 4.4723,
+      "epoch": 0.22,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.4718,
       "step": 100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.2406,
+      "epoch": 0.23,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 4.3046,
       "step": 101
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.940000000000001e-06,
-      "loss": 4.2459,
+      "epoch": 0.23,
+      "learning_rate": 6e-06,
+      "loss": 4.2435,
       "step": 102
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6e-06,
-      "loss": 4.3816,
+      "epoch": 0.23,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.113,
       "step": 103
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 4.2937,
+      "epoch": 0.23,
+      "learning_rate": 6.12e-06,
+      "loss": 4.1906,
       "step": 104
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.12e-06,
-      "loss": 4.117,
+      "epoch": 0.24,
+      "learning_rate": 6.18e-06,
+      "loss": 4.1371,
       "step": 105
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.18e-06,
-      "loss": 4.3714,
+      "epoch": 0.24,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 4.1835,
       "step": 106
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.2399999999999995e-06,
-      "loss": 4.3265,
+      "epoch": 0.24,
+      "learning_rate": 6.3e-06,
+      "loss": 4.28,
       "step": 107
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.3e-06,
-      "loss": 4.4825,
+      "epoch": 0.24,
+      "learning_rate": 6.36e-06,
+      "loss": 4.1821,
       "step": 108
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.36e-06,
-      "loss": 4.0523,
+      "epoch": 0.24,
+      "learning_rate": 6.42e-06,
+      "loss": 4.1884,
       "step": 109
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.42e-06,
-      "loss": 4.2058,
+      "epoch": 0.25,
+      "learning_rate": 6.48e-06,
+      "loss": 4.2333,
       "step": 110
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.48e-06,
-      "loss": 4.4007,
+      "epoch": 0.25,
+      "learning_rate": 6.54e-06,
+      "loss": 4.2675,
       "step": 111
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.54e-06,
-      "loss": 4.2399,
+      "epoch": 0.25,
+      "learning_rate": 6.6e-06,
+      "loss": 4.2355,
       "step": 112
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.6e-06,
-      "loss": 4.2235,
+      "epoch": 0.25,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.2424,
       "step": 113
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 4.1655,
+      "epoch": 0.26,
+      "learning_rate": 6.72e-06,
+      "loss": 4.2007,
       "step": 114
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.72e-06,
-      "loss": 4.2307,
+      "epoch": 0.26,
+      "learning_rate": 6.78e-06,
+      "loss": 4.2562,
       "step": 115
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.78e-06,
-      "loss": 4.0141,
+      "epoch": 0.26,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 4.3384,
       "step": 116
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.840000000000001e-06,
-      "loss": 4.3334,
+      "epoch": 0.26,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 4.217,
       "step": 117
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.900000000000001e-06,
-      "loss": 4.1807,
+      "epoch": 0.26,
+      "learning_rate": 6.96e-06,
+      "loss": 4.37,
       "step": 118
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.96e-06,
-      "loss": 4.3577,
+      "epoch": 0.27,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.2666,
       "step": 119
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 4.3768,
+      "epoch": 0.27,
+      "learning_rate": 7.08e-06,
+      "loss": 4.292,
       "step": 120
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.08e-06,
-      "loss": 4.3085,
+      "epoch": 0.27,
+      "learning_rate": 7.14e-06,
+      "loss": 4.1314,
       "step": 121
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.14e-06,
-      "loss": 4.2463,
+      "epoch": 0.27,
+      "learning_rate": 7.2e-06,
+      "loss": 4.2254,
       "step": 122
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.2e-06,
-      "loss": 4.1331,
+      "epoch": 0.28,
+      "learning_rate": 7.26e-06,
+      "loss": 4.236,
       "step": 123
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.26e-06,
-      "loss": 4.3671,
+      "epoch": 0.28,
+      "learning_rate": 7.32e-06,
+      "loss": 4.2277,
       "step": 124
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.32e-06,
-      "loss": 4.3023,
+      "epoch": 0.28,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.3383,
       "step": 125
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 4.3423,
+      "epoch": 0.28,
+      "learning_rate": 7.44e-06,
+      "loss": 4.3372,
       "step": 126
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.44e-06,
-      "loss": 4.4363,
+      "epoch": 0.28,
+      "learning_rate": 7.5e-06,
+      "loss": 4.2553,
       "step": 127
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.5e-06,
-      "loss": 4.2239,
+      "epoch": 0.29,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.2319,
       "step": 128
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 4.1795,
+      "epoch": 0.29,
+      "learning_rate": 7.62e-06,
+      "loss": 4.1354,
       "step": 129
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.62e-06,
-      "loss": 4.3581,
+      "epoch": 0.29,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.2859,
       "step": 130
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.680000000000001e-06,
-      "loss": 4.262,
+      "epoch": 0.29,
+      "learning_rate": 7.74e-06,
+      "loss": 4.1919,
       "step": 131
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.74e-06,
-      "loss": 4.3525,
+      "epoch": 0.3,
+      "learning_rate": 7.8e-06,
+      "loss": 4.2501,
       "step": 132
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.8e-06,
-      "loss": 4.3125,
+      "epoch": 0.3,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.1723,
       "step": 133
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 4.3189,
+      "epoch": 0.3,
+      "learning_rate": 7.92e-06,
+      "loss": 4.2081,
       "step": 134
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.92e-06,
-      "loss": 4.2083,
+      "epoch": 0.3,
+      "learning_rate": 7.98e-06,
+      "loss": 4.3765,
       "step": 135
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.98e-06,
-      "loss": 4.2304,
+      "epoch": 0.3,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 4.2239,
       "step": 136
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 4.5801,
+      "epoch": 0.31,
+      "learning_rate": 8.1e-06,
+      "loss": 4.3212,
       "step": 137
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 8.1e-06,
-      "loss": 4.3233,
+      "epoch": 0.31,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 4.3565,
       "step": 138
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.160000000000001e-06,
-      "loss": 4.36,
+      "epoch": 0.31,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 4.3689,
       "step": 139
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.220000000000001e-06,
-      "loss": 4.1527,
+      "epoch": 0.31,
+      "learning_rate": 8.28e-06,
+      "loss": 4.3316,
       "step": 140
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.28e-06,
-      "loss": 4.4406,
+      "epoch": 0.32,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 4.4513,
       "step": 141
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.340000000000001e-06,
-      "loss": 4.5201,
+      "epoch": 0.32,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.3836,
       "step": 142
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.4419,
+      "epoch": 0.32,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 4.3245,
       "step": 143
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.459999999999999e-06,
-      "loss": 4.351,
+      "epoch": 0.32,
+      "learning_rate": 8.52e-06,
+      "loss": 4.3805,
       "step": 144
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.52e-06,
-      "loss": 4.4423,
+      "epoch": 0.33,
+      "learning_rate": 8.58e-06,
+      "loss": 4.2924,
       "step": 145
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.58e-06,
-      "loss": 4.4382,
+      "epoch": 0.33,
+      "learning_rate": 8.64e-06,
+      "loss": 4.3797,
       "step": 146
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.64e-06,
-      "loss": 4.5181,
+      "epoch": 0.33,
+      "learning_rate": 8.7e-06,
+      "loss": 4.6306,
       "step": 147
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 8.7e-06,
-      "loss": 4.2803,
+      "epoch": 0.33,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 4.77,
       "step": 148
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 8.759999999999999e-06,
-      "loss": 4.6252,
+      "epoch": 0.33,
+      "learning_rate": 8.82e-06,
+      "loss": 4.5055,
       "step": 149
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 8.82e-06,
-      "loss": 4.2255,
+      "epoch": 0.34,
+      "learning_rate": 8.88e-06,
+      "loss": 4.6005,
       "step": 150
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 8.88e-06,
-      "loss": 4.3498,
+      "epoch": 0.34,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 4.1378,
       "step": 151
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 8.939999999999999e-06,
-      "loss": 4.2342,
+      "epoch": 0.34,
+      "learning_rate": 9e-06,
+      "loss": 4.2224,
       "step": 152
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9e-06,
-      "loss": 4.2839,
+      "epoch": 0.34,
+      "learning_rate": 9.06e-06,
+      "loss": 4.2385,
       "step": 153
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.06e-06,
-      "loss": 4.1556,
+      "epoch": 0.35,
+      "learning_rate": 9.12e-06,
+      "loss": 4.1296,
       "step": 154
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.12e-06,
-      "loss": 4.0843,
+      "epoch": 0.35,
+      "learning_rate": 9.18e-06,
+      "loss": 4.2382,
       "step": 155
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.18e-06,
-      "loss": 4.2043,
+      "epoch": 0.35,
+      "learning_rate": 9.24e-06,
+      "loss": 4.0975,
       "step": 156
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.24e-06,
-      "loss": 4.3446,
+      "epoch": 0.35,
+      "learning_rate": 9.3e-06,
+      "loss": 4.1335,
       "step": 157
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.3e-06,
-      "loss": 4.4503,
+      "epoch": 0.35,
+      "learning_rate": 9.36e-06,
+      "loss": 4.2322,
       "step": 158
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.36e-06,
-      "loss": 4.4753,
+      "epoch": 0.36,
+      "learning_rate": 9.42e-06,
+      "loss": 4.2269,
       "step": 159
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.42e-06,
-      "loss": 4.1959,
+      "epoch": 0.36,
+      "learning_rate": 9.48e-06,
+      "loss": 4.2434,
       "step": 160
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.48e-06,
-      "loss": 4.2175,
+      "epoch": 0.36,
+      "learning_rate": 9.54e-06,
+      "loss": 4.2971,
       "step": 161
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.54e-06,
-      "loss": 4.2721,
+      "epoch": 0.36,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.2479,
       "step": 162
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2518,
+      "epoch": 0.37,
+      "learning_rate": 9.66e-06,
+      "loss": 4.2019,
       "step": 163
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.66e-06,
-      "loss": 4.3758,
+      "epoch": 0.37,
+      "learning_rate": 9.72e-06,
+      "loss": 4.2686,
       "step": 164
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.72e-06,
-      "loss": 4.2946,
+      "epoch": 0.37,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.1306,
       "step": 165
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 4.3018,
+      "epoch": 0.37,
+      "learning_rate": 9.84e-06,
+      "loss": 4.1961,
       "step": 166
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 9.84e-06,
-      "loss": 4.2654,
+      "epoch": 0.37,
+      "learning_rate": 9.9e-06,
+      "loss": 4.1072,
       "step": 167
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 9.9e-06,
-      "loss": 4.0493,
+      "epoch": 0.38,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.1856,
       "step": 168
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 4.0436,
+      "epoch": 0.38,
+      "learning_rate": 1.002e-05,
+      "loss": 4.2457,
       "step": 169
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.002e-05,
-      "loss": 4.2211,
+      "epoch": 0.38,
+      "learning_rate": 1.008e-05,
+      "loss": 4.2701,
       "step": 170
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.008e-05,
-      "loss": 4.4244,
+      "epoch": 0.38,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 4.2464,
       "step": 171
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.0140000000000001e-05,
-      "loss": 4.3368,
+      "epoch": 0.39,
+      "learning_rate": 1.02e-05,
+      "loss": 4.3068,
       "step": 172
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.02e-05,
-      "loss": 4.272,
+      "epoch": 0.39,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 4.1162,
       "step": 173
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.0260000000000002e-05,
-      "loss": 4.1146,
+      "epoch": 0.39,
+      "learning_rate": 1.032e-05,
+      "loss": 4.2632,
       "step": 174
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.032e-05,
-      "loss": 4.244,
+      "epoch": 0.39,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 4.3134,
       "step": 175
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.0379999999999999e-05,
-      "loss": 4.0634,
+      "epoch": 0.39,
+      "learning_rate": 1.044e-05,
+      "loss": 4.1883,
       "step": 176
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.044e-05,
-      "loss": 4.0988,
+      "epoch": 0.4,
+      "learning_rate": 1.05e-05,
+      "loss": 4.1563,
       "step": 177
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.05e-05,
-      "loss": 4.1709,
+      "epoch": 0.4,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 4.2408,
       "step": 178
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.0559999999999999e-05,
-      "loss": 4.2762,
+      "epoch": 0.4,
+      "learning_rate": 1.062e-05,
+      "loss": 4.3123,
       "step": 179
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.062e-05,
-      "loss": 4.2628,
+      "epoch": 0.4,
+      "learning_rate": 1.068e-05,
+      "loss": 4.301,
       "step": 180
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.068e-05,
-      "loss": 4.304,
+      "epoch": 0.41,
+      "learning_rate": 1.074e-05,
+      "loss": 4.2258,
       "step": 181
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.074e-05,
-      "loss": 4.1144,
+      "epoch": 0.41,
+      "learning_rate": 1.08e-05,
+      "loss": 4.2896,
       "step": 182
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.08e-05,
-      "loss": 4.3378,
+      "epoch": 0.41,
+      "learning_rate": 1.086e-05,
+      "loss": 4.2775,
       "step": 183
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.086e-05,
-      "loss": 4.2838,
+      "epoch": 0.41,
+      "learning_rate": 1.092e-05,
+      "loss": 4.2849,
       "step": 184
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.092e-05,
-      "loss": 4.3166,
+      "epoch": 0.41,
+      "learning_rate": 1.098e-05,
+      "loss": 4.2654,
       "step": 185
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.098e-05,
-      "loss": 4.1774,
+      "epoch": 0.42,
+      "learning_rate": 1.104e-05,
+      "loss": 4.128,
       "step": 186
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.104e-05,
-      "loss": 4.3944,
+      "epoch": 0.42,
+      "learning_rate": 1.11e-05,
+      "loss": 4.1517,
       "step": 187
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.11e-05,
-      "loss": 4.1116,
+      "epoch": 0.42,
+      "learning_rate": 1.116e-05,
+      "loss": 4.1204,
       "step": 188
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.116e-05,
-      "loss": 4.2198,
+      "epoch": 0.42,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 4.2358,
       "step": 189
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.1220000000000001e-05,
-      "loss": 4.3477,
+      "epoch": 0.43,
+      "learning_rate": 1.128e-05,
+      "loss": 4.3392,
       "step": 190
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.128e-05,
-      "loss": 4.3497,
+      "epoch": 0.43,
+      "learning_rate": 1.134e-05,
+      "loss": 4.254,
       "step": 191
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.134e-05,
-      "loss": 4.2842,
+      "epoch": 0.43,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.2131,
       "step": 192
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.5254,
+      "epoch": 0.43,
+      "learning_rate": 1.146e-05,
+      "loss": 4.2087,
       "step": 193
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.146e-05,
-      "loss": 4.2371,
+      "epoch": 0.43,
+      "learning_rate": 1.152e-05,
+      "loss": 4.3236,
       "step": 194
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.152e-05,
-      "loss": 4.2678,
+      "epoch": 0.44,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 4.4341,
       "step": 195
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.1580000000000001e-05,
-      "loss": 4.5069,
+      "epoch": 0.44,
+      "learning_rate": 1.164e-05,
+      "loss": 4.3805,
       "step": 196
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.164e-05,
-      "loss": 4.4958,
+      "epoch": 0.44,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 4.3194,
       "step": 197
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.1700000000000001e-05,
-      "loss": 4.4464,
+      "epoch": 0.44,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 4.61,
       "step": 198
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.1760000000000001e-05,
-      "loss": 4.6181,
+      "epoch": 0.45,
+      "learning_rate": 1.182e-05,
+      "loss": 4.5467,
       "step": 199
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.182e-05,
-      "loss": 4.6803,
+      "epoch": 0.45,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 4.4941,
       "step": 200
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.1880000000000001e-05,
-      "loss": 4.0903,
+      "epoch": 0.45,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 4.2598,
       "step": 201
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.1940000000000001e-05,
-      "loss": 4.2021,
+      "epoch": 0.45,
+      "learning_rate": 1.2e-05,
+      "loss": 4.1943,
       "step": 202
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.2e-05,
-      "loss": 4.0922,
+      "epoch": 0.46,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 4.2136,
       "step": 203
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.2060000000000001e-05,
-      "loss": 4.0967,
+      "epoch": 0.46,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 4.209,
       "step": 204
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.2120000000000001e-05,
-      "loss": 4.0051,
+      "epoch": 0.46,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 4.2688,
       "step": 205
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.2180000000000002e-05,
-      "loss": 4.07,
+      "epoch": 0.46,
+      "learning_rate": 1.224e-05,
+      "loss": 4.2527,
       "step": 206
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.224e-05,
-      "loss": 4.1579,
+      "epoch": 0.46,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 4.1229,
       "step": 207
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.2299999999999999e-05,
-      "loss": 3.9931,
+      "epoch": 0.47,
+      "learning_rate": 1.236e-05,
+      "loss": 4.2149,
       "step": 208
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.236e-05,
-      "loss": 4.2358,
+      "epoch": 0.47,
+      "learning_rate": 1.242e-05,
+      "loss": 4.2757,
       "step": 209
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.242e-05,
-      "loss": 4.1737,
+      "epoch": 0.47,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 4.2353,
       "step": 210
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.2479999999999999e-05,
-      "loss": 4.1183,
+      "epoch": 0.47,
+      "learning_rate": 1.254e-05,
+      "loss": 4.1108,
       "step": 211
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.254e-05,
-      "loss": 4.2407,
+      "epoch": 0.48,
+      "learning_rate": 1.26e-05,
+      "loss": 4.16,
       "step": 212
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.26e-05,
-      "loss": 4.1897,
+      "epoch": 0.48,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 4.0885,
       "step": 213
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.2659999999999999e-05,
-      "loss": 4.2462,
+      "epoch": 0.48,
+      "learning_rate": 1.272e-05,
+      "loss": 4.1628,
       "step": 214
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.272e-05,
-      "loss": 4.1168,
+      "epoch": 0.48,
+      "learning_rate": 1.278e-05,
+      "loss": 4.209,
       "step": 215
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.278e-05,
-      "loss": 4.2695,
+      "epoch": 0.48,
+      "learning_rate": 1.284e-05,
+      "loss": 4.1769,
       "step": 216
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.284e-05,
-      "loss": 4.2268,
+      "epoch": 0.49,
+      "learning_rate": 1.29e-05,
+      "loss": 4.2236,
       "step": 217
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.29e-05,
-      "loss": 4.1414,
+      "epoch": 0.49,
+      "learning_rate": 1.296e-05,
+      "loss": 4.175,
       "step": 218
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.296e-05,
-      "loss": 4.2381,
+      "epoch": 0.49,
+      "learning_rate": 1.302e-05,
+      "loss": 4.1064,
       "step": 219
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.302e-05,
-      "loss": 4.1982,
+      "epoch": 0.49,
+      "learning_rate": 1.308e-05,
+      "loss": 4.2232,
       "step": 220
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.308e-05,
-      "loss": 4.1713,
+      "epoch": 0.5,
+      "learning_rate": 1.314e-05,
+      "loss": 4.1993,
       "step": 221
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.314e-05,
-      "loss": 4.2545,
+      "epoch": 0.5,
+      "learning_rate": 1.32e-05,
+      "loss": 4.1825,
       "step": 222
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.32e-05,
-      "loss": 4.2899,
+      "epoch": 0.5,
+      "learning_rate": 1.326e-05,
+      "loss": 4.2186,
       "step": 223
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.326e-05,
-      "loss": 4.2193,
+      "epoch": 0.5,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 4.1414,
       "step": 224
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.3320000000000001e-05,
-      "loss": 4.3013,
+      "epoch": 0.5,
+      "learning_rate": 1.338e-05,
+      "loss": 4.299,
       "step": 225
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.338e-05,
-      "loss": 4.4583,
+      "epoch": 0.51,
+      "learning_rate": 1.344e-05,
+      "loss": 4.1934,
       "step": 226
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.344e-05,
-      "loss": 4.212,
+      "epoch": 0.51,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 4.3162,
       "step": 227
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 4.1256,
+      "epoch": 0.51,
+      "learning_rate": 1.356e-05,
+      "loss": 4.262,
       "step": 228
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.356e-05,
-      "loss": 4.1234,
+      "epoch": 0.51,
+      "learning_rate": 1.362e-05,
+      "loss": 4.1437,
       "step": 229
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.362e-05,
-      "loss": 4.204,
+      "epoch": 0.52,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 4.2367,
       "step": 230
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.3680000000000001e-05,
-      "loss": 4.1709,
+      "epoch": 0.52,
+      "learning_rate": 1.374e-05,
+      "loss": 4.2988,
       "step": 231
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.374e-05,
-      "loss": 4.3005,
+      "epoch": 0.52,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.0968,
       "step": 232
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.1864,
+      "epoch": 0.52,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 4.1227,
       "step": 233
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.3860000000000001e-05,
-      "loss": 4.2975,
+      "epoch": 0.52,
+      "learning_rate": 1.392e-05,
+      "loss": 4.2133,
       "step": 234
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.392e-05,
-      "loss": 4.3514,
+      "epoch": 0.53,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 4.2843,
       "step": 235
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.3980000000000002e-05,
-      "loss": 4.2039,
+      "epoch": 0.53,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 4.2125,
       "step": 236
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.4040000000000001e-05,
-      "loss": 4.2195,
+      "epoch": 0.53,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 4.2384,
       "step": 237
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.4099999999999999e-05,
-      "loss": 4.4241,
+      "epoch": 0.53,
+      "learning_rate": 1.416e-05,
+      "loss": 4.1947,
       "step": 238
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.416e-05,
-      "loss": 4.2328,
+      "epoch": 0.54,
+      "learning_rate": 1.422e-05,
+      "loss": 4.1355,
       "step": 239
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.422e-05,
-      "loss": 4.3521,
+      "epoch": 0.54,
+      "learning_rate": 1.428e-05,
+      "loss": 4.2412,
       "step": 240
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.428e-05,
-      "loss": 4.3209,
+      "epoch": 0.54,
+      "learning_rate": 1.434e-05,
+      "loss": 4.2237,
       "step": 241
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.434e-05,
-      "loss": 4.4751,
+      "epoch": 0.54,
+      "learning_rate": 1.44e-05,
+      "loss": 4.2662,
       "step": 242
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.44e-05,
-      "loss": 4.327,
+      "epoch": 0.54,
+      "learning_rate": 1.446e-05,
+      "loss": 4.4641,
       "step": 243
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.446e-05,
-      "loss": 4.2864,
+      "epoch": 0.55,
+      "learning_rate": 1.452e-05,
+      "loss": 4.2923,
       "step": 244
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.452e-05,
-      "loss": 4.4874,
+      "epoch": 0.55,
+      "learning_rate": 1.458e-05,
+      "loss": 4.2855,
       "step": 245
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.458e-05,
-      "loss": 4.3515,
+      "epoch": 0.55,
+      "learning_rate": 1.464e-05,
+      "loss": 4.1766,
       "step": 246
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.464e-05,
-      "loss": 4.7349,
+      "epoch": 0.55,
+      "learning_rate": 1.47e-05,
+      "loss": 4.1232,
       "step": 247
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.47e-05,
-      "loss": 4.7552,
+      "epoch": 0.56,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 4.4921,
       "step": 248
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.4760000000000001e-05,
-      "loss": 4.5734,
-      "step": 249
-    },
-    {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 1.482e-05,
-      "loss": 4.2063,
-      "step": 250
+      "loss": 4.5888,
+      "step": 249
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 1.488e-05,
-      "loss": 4.3978,
-      "step": 251
+      "loss": 4.582,
+      "step": 250
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 1.4940000000000001e-05,
-      "loss": 4.0939,
-      "step": 252
+      "loss": 4.1951,
+      "step": 251
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 1.5e-05,
-      "loss": 4.1101,
-      "step": 253
+      "loss": 4.2269,
+      "step": 252
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 1.506e-05,
-      "loss": 4.2572,
-      "step": 254
+      "loss": 4.2679,
+      "step": 253
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 1.5120000000000001e-05,
-      "loss": 4.125,
-      "step": 255
+      "loss": 4.1101,
+      "step": 254
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 1.518e-05,
-      "loss": 4.1,
-      "step": 256
+      "loss": 4.1986,
+      "step": 255
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 1.524e-05,
-      "loss": 4.2977,
-      "step": 257
+      "loss": 4.179,
+      "step": 256
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 1.53e-05,
-      "loss": 4.1329,
-      "step": 258
+      "loss": 4.1623,
+      "step": 257
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 1.5360000000000002e-05,
-      "loss": 4.1073,
-      "step": 259
+      "loss": 4.1636,
+      "step": 258
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 1.542e-05,
-      "loss": 4.2496,
-      "step": 260
+      "loss": 4.1744,
+      "step": 259
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 1.548e-05,
-      "loss": 4.2685,
-      "step": 261
+      "loss": 4.1098,
+      "step": 260
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 1.554e-05,
-      "loss": 4.0245,
-      "step": 262
+      "loss": 4.2371,
+      "step": 261
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 1.56e-05,
-      "loss": 4.1365,
-      "step": 263
+      "loss": 4.1562,
+      "step": 262
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 1.5660000000000003e-05,
-      "loss": 4.1303,
-      "step": 264
+      "loss": 4.294,
+      "step": 263
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 1.5720000000000002e-05,
-      "loss": 4.2422,
-      "step": 265
+      "loss": 4.1616,
+      "step": 264
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 1.578e-05,
-      "loss": 4.307,
-      "step": 266
+      "loss": 4.2108,
+      "step": 265
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 1.584e-05,
-      "loss": 4.1291,
-      "step": 267
+      "loss": 4.1689,
+      "step": 266
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 1.59e-05,
-      "loss": 4.4503,
-      "step": 268
+      "loss": 4.2019,
+      "step": 267
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 1.596e-05,
-      "loss": 4.1821,
-      "step": 269
+      "loss": 4.1509,
+      "step": 268
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 1.6020000000000002e-05,
-      "loss": 4.3484,
-      "step": 270
+      "loss": 4.2425,
+      "step": 269
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 1.6080000000000002e-05,
-      "loss": 3.9466,
-      "step": 271
+      "loss": 4.1185,
+      "step": 270
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 1.614e-05,
-      "loss": 4.0963,
-      "step": 272
+      "loss": 4.0561,
+      "step": 271
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 1.62e-05,
-      "loss": 4.1321,
-      "step": 273
+      "loss": 4.2334,
+      "step": 272
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 1.626e-05,
-      "loss": 4.0624,
-      "step": 274
+      "loss": 4.03,
+      "step": 273
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 1.6320000000000003e-05,
-      "loss": 4.2895,
-      "step": 275
+      "loss": 4.1374,
+      "step": 274
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 1.6380000000000002e-05,
-      "loss": 4.225,
-      "step": 276
+      "loss": 4.2188,
+      "step": 275
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 1.6440000000000002e-05,
-      "loss": 4.1543,
-      "step": 277
+      "loss": 4.1667,
+      "step": 276
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 1.65e-05,
-      "loss": 4.1254,
-      "step": 278
+      "loss": 4.1154,
+      "step": 277
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 1.656e-05,
-      "loss": 4.1338,
-      "step": 279
+      "loss": 4.1921,
+      "step": 278
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 1.6620000000000004e-05,
-      "loss": 4.2958,
-      "step": 280
+      "loss": 4.1726,
+      "step": 279
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 1.6680000000000003e-05,
-      "loss": 4.0743,
-      "step": 281
+      "loss": 4.3145,
+      "step": 280
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 1.6740000000000002e-05,
-      "loss": 4.217,
-      "step": 282
+      "loss": 4.2087,
+      "step": 281
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 1.6800000000000002e-05,
-      "loss": 3.9742,
-      "step": 283
+      "loss": 4.0653,
+      "step": 282
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 1.686e-05,
-      "loss": 4.1994,
-      "step": 284
+      "loss": 4.0902,
+      "step": 283
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 1.6919999999999997e-05,
-      "loss": 4.2729,
-      "step": 285
+      "loss": 4.1814,
+      "step": 284
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 1.698e-05,
-      "loss": 4.1934,
-      "step": 286
+      "loss": 4.2828,
+      "step": 285
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 1.704e-05,
-      "loss": 4.3364,
-      "step": 287
+      "loss": 4.1693,
+      "step": 286
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 1.71e-05,
-      "loss": 4.2937,
-      "step": 288
+      "loss": 4.2818,
+      "step": 287
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 1.716e-05,
-      "loss": 4.3273,
-      "step": 289
+      "loss": 4.2331,
+      "step": 288
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 1.7219999999999998e-05,
-      "loss": 4.302,
-      "step": 290
+      "loss": 4.1357,
+      "step": 289
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 1.728e-05,
-      "loss": 4.3626,
-      "step": 291
+      "loss": 4.1985,
+      "step": 290
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 1.734e-05,
-      "loss": 4.2106,
-      "step": 292
+      "loss": 4.4431,
+      "step": 291
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 1.74e-05,
-      "loss": 4.306,
-      "step": 293
+      "loss": 4.3144,
+      "step": 292
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 1.746e-05,
-      "loss": 4.2628,
-      "step": 294
+      "loss": 4.3839,
+      "step": 293
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 1.7519999999999998e-05,
-      "loss": 4.1744,
-      "step": 295
+      "loss": 4.3518,
+      "step": 294
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 1.758e-05,
-      "loss": 4.4221,
-      "step": 296
+      "loss": 4.3501,
+      "step": 295
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 1.764e-05,
-      "loss": 4.5329,
-      "step": 297
+      "loss": 4.3526,
+      "step": 296
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 1.77e-05,
-      "loss": 4.3351,
-      "step": 298
+      "loss": 4.2403,
+      "step": 297
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 1.776e-05,
-      "loss": 4.738,
-      "step": 299
+      "loss": 4.5569,
+      "step": 298
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 1.782e-05,
-      "loss": 4.7495,
-      "step": 300
+      "loss": 4.3044,
+      "step": 299
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 1.7879999999999998e-05,
-      "loss": 4.0522,
-      "step": 301
+      "loss": 4.7419,
+      "step": 300
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 1.794e-05,
-      "loss": 4.1765,
-      "step": 302
+      "loss": 4.2255,
+      "step": 301
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 1.8e-05,
-      "loss": 4.0936,
-      "step": 303
+      "loss": 4.1911,
+      "step": 302
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 1.806e-05,
-      "loss": 4.2213,
-      "step": 304
+      "loss": 4.1913,
+      "step": 303
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 1.812e-05,
-      "loss": 4.0573,
-      "step": 305
+      "loss": 4.2035,
+      "step": 304
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 1.818e-05,
-      "loss": 4.122,
-      "step": 306
+      "loss": 4.1659,
+      "step": 305
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 1.824e-05,
-      "loss": 4.1466,
-      "step": 307
+      "loss": 4.1522,
+      "step": 306
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 1.83e-05,
-      "loss": 4.2568,
-      "step": 308
+      "loss": 4.1422,
+      "step": 307
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 1.836e-05,
-      "loss": 4.1553,
-      "step": 309
+      "loss": 4.148,
+      "step": 308
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 1.842e-05,
-      "loss": 4.2805,
-      "step": 310
+      "loss": 4.1458,
+      "step": 309
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 1.848e-05,
-      "loss": 4.3665,
-      "step": 311
+      "loss": 4.2421,
+      "step": 310
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 1.854e-05,
-      "loss": 4.2954,
-      "step": 312
+      "loss": 4.0367,
+      "step": 311
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 1.86e-05,
-      "loss": 4.1888,
-      "step": 313
+      "loss": 4.2137,
+      "step": 312
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 1.866e-05,
-      "loss": 4.2317,
-      "step": 314
+      "loss": 4.2195,
+      "step": 313
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 1.872e-05,
-      "loss": 4.0812,
-      "step": 315
+      "loss": 4.1313,
+      "step": 314
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 1.878e-05,
-      "loss": 4.2498,
-      "step": 316
+      "loss": 4.165,
+      "step": 315
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 1.884e-05,
-      "loss": 4.0964,
-      "step": 317
+      "loss": 4.1387,
+      "step": 316
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 1.8900000000000002e-05,
-      "loss": 4.1132,
-      "step": 318
+      "loss": 4.1316,
+      "step": 317
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 1.896e-05,
-      "loss": 4.2153,
-      "step": 319
+      "loss": 4.164,
+      "step": 318
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 1.902e-05,
-      "loss": 4.1501,
-      "step": 320
+      "loss": 4.1583,
+      "step": 319
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 1.908e-05,
-      "loss": 4.2813,
-      "step": 321
+      "loss": 4.2331,
+      "step": 320
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 1.914e-05,
-      "loss": 4.1588,
-      "step": 322
+      "loss": 4.1241,
+      "step": 321
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.0803,
-      "step": 323
+      "loss": 3.9994,
+      "step": 322
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 1.9260000000000002e-05,
-      "loss": 4.2773,
-      "step": 324
+      "loss": 4.1963,
+      "step": 323
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 1.932e-05,
-      "loss": 4.2866,
-      "step": 325
+      "loss": 4.05,
+      "step": 324
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 1.938e-05,
-      "loss": 4.2622,
-      "step": 326
+      "loss": 4.2003,
+      "step": 325
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 1.944e-05,
-      "loss": 4.1441,
-      "step": 327
+      "loss": 4.1707,
+      "step": 326
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 1.95e-05,
-      "loss": 4.1331,
-      "step": 328
+      "loss": 4.1238,
+      "step": 327
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 1.9560000000000002e-05,
-      "loss": 4.3182,
-      "step": 329
+      "loss": 4.0829,
+      "step": 328
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 1.9620000000000002e-05,
-      "loss": 4.3532,
-      "step": 330
+      "loss": 4.1401,
+      "step": 329
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 1.968e-05,
-      "loss": 4.1673,
-      "step": 331
+      "loss": 4.1984,
+      "step": 330
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 1.974e-05,
-      "loss": 4.2077,
-      "step": 332
+      "loss": 4.2043,
+      "step": 331
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 1.98e-05,
-      "loss": 4.2068,
-      "step": 333
+      "loss": 4.0911,
+      "step": 332
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 1.9860000000000003e-05,
-      "loss": 4.2471,
-      "step": 334
+      "loss": 4.2525,
+      "step": 333
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 1.9920000000000002e-05,
-      "loss": 4.3877,
-      "step": 335
+      "loss": 4.1515,
+      "step": 334
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 1.9980000000000002e-05,
-      "loss": 4.1347,
-      "step": 336
+      "loss": 4.3075,
+      "step": 335
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 2.004e-05,
-      "loss": 4.0476,
-      "step": 337
+      "loss": 4.1172,
+      "step": 336
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 2.01e-05,
-      "loss": 4.3133,
-      "step": 338
+      "loss": 4.1193,
+      "step": 337
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 2.016e-05,
-      "loss": 4.1127,
-      "step": 339
+      "loss": 4.2335,
+      "step": 338
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 2.0220000000000003e-05,
-      "loss": 4.2478,
-      "step": 340
+      "loss": 4.2131,
+      "step": 339
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 2.0280000000000002e-05,
-      "loss": 4.1574,
-      "step": 341
+      "loss": 4.264,
+      "step": 340
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 2.0340000000000002e-05,
-      "loss": 4.2871,
-      "step": 342
+      "loss": 4.1833,
+      "step": 341
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 2.04e-05,
-      "loss": 4.2293,
-      "step": 343
+      "loss": 4.3401,
+      "step": 342
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 2.046e-05,
-      "loss": 4.3118,
-      "step": 344
+      "loss": 4.2661,
+      "step": 343
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 2.0520000000000003e-05,
-      "loss": 4.3705,
-      "step": 345
+      "loss": 4.3547,
+      "step": 344
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 2.0580000000000003e-05,
-      "loss": 4.5409,
-      "step": 346
+      "loss": 4.096,
+      "step": 345
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 2.064e-05,
-      "loss": 4.3818,
-      "step": 347
+      "loss": 4.339,
+      "step": 346
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 2.07e-05,
-      "loss": 4.4166,
-      "step": 348
+      "loss": 4.3258,
+      "step": 347
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 2.0759999999999998e-05,
-      "loss": 4.5959,
-      "step": 349
+      "loss": 4.2641,
+      "step": 348
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 2.082e-05,
-      "loss": 4.4276,
-      "step": 350
+      "loss": 4.4812,
+      "step": 349
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 2.088e-05,
-      "loss": 4.1841,
-      "step": 351
+      "loss": 4.4627,
+      "step": 350
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 2.094e-05,
-      "loss": 4.2181,
-      "step": 352
+      "loss": 4.2444,
+      "step": 351
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 2.1e-05,
-      "loss": 4.1374,
-      "step": 353
+      "loss": 4.1637,
+      "step": 352
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 2.1059999999999998e-05,
-      "loss": 4.1394,
-      "step": 354
+      "loss": 4.1012,
+      "step": 353
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 2.1119999999999998e-05,
-      "loss": 4.2856,
-      "step": 355
+      "loss": 3.9923,
+      "step": 354
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 2.118e-05,
-      "loss": 4.0304,
-      "step": 356
+      "loss": 4.175,
+      "step": 355
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 2.124e-05,
-      "loss": 4.0897,
-      "step": 357
+      "loss": 4.2816,
+      "step": 356
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 2.13e-05,
-      "loss": 4.1678,
-      "step": 358
+      "loss": 4.1508,
+      "step": 357
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 2.136e-05,
-      "loss": 4.1266,
-      "step": 359
+      "loss": 4.2011,
+      "step": 358
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 2.1419999999999998e-05,
-      "loss": 4.2845,
-      "step": 360
+      "loss": 4.0786,
+      "step": 359
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 2.148e-05,
-      "loss": 4.1321,
-      "step": 361
+      "loss": 4.1572,
+      "step": 360
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 2.154e-05,
-      "loss": 4.1353,
-      "step": 362
+      "loss": 4.0844,
+      "step": 361
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 2.16e-05,
-      "loss": 4.0673,
-      "step": 363
+      "loss": 4.1476,
+      "step": 362
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 2.166e-05,
-      "loss": 4.2948,
-      "step": 364
+      "loss": 4.0828,
+      "step": 363
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 2.172e-05,
-      "loss": 4.0433,
-      "step": 365
+      "loss": 4.1726,
+      "step": 364
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 2.178e-05,
-      "loss": 4.1878,
-      "step": 366
+      "loss": 4.2892,
+      "step": 365
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 2.184e-05,
-      "loss": 4.0997,
-      "step": 367
+      "loss": 4.0957,
+      "step": 366
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 2.19e-05,
-      "loss": 4.2037,
-      "step": 368
+      "loss": 4.1822,
+      "step": 367
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 2.196e-05,
-      "loss": 4.2507,
-      "step": 369
+      "loss": 4.1647,
+      "step": 368
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 2.202e-05,
-      "loss": 4.2778,
-      "step": 370
+      "loss": 4.0817,
+      "step": 369
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 2.208e-05,
-      "loss": 4.0861,
-      "step": 371
+      "loss": 4.0894,
+      "step": 370
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 2.214e-05,
-      "loss": 4.3873,
-      "step": 372
+      "loss": 4.079,
+      "step": 371
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 2.22e-05,
-      "loss": 4.1633,
-      "step": 373
+      "loss": 4.1652,
+      "step": 372
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 2.226e-05,
-      "loss": 4.2557,
-      "step": 374
+      "loss": 4.1596,
+      "step": 373
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 2.232e-05,
-      "loss": 4.105,
-      "step": 375
+      "loss": 3.9787,
+      "step": 374
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 2.238e-05,
-      "loss": 4.1985,
-      "step": 376
+      "loss": 4.1849,
+      "step": 375
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 2.2440000000000002e-05,
-      "loss": 3.9895,
-      "step": 377
+      "loss": 4.0419,
+      "step": 376
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 2.25e-05,
-      "loss": 4.3159,
-      "step": 378
+      "loss": 4.145,
+      "step": 377
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 2.256e-05,
-      "loss": 4.2099,
-      "step": 379
+      "loss": 4.1548,
+      "step": 378
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 2.262e-05,
-      "loss": 4.3025,
-      "step": 380
+      "loss": 4.0737,
+      "step": 379
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 2.268e-05,
-      "loss": 4.2785,
-      "step": 381
+      "loss": 4.096,
+      "step": 380
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 2.274e-05,
-      "loss": 4.1978,
-      "step": 382
+      "loss": 4.1361,
+      "step": 381
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.2591,
-      "step": 383
+      "loss": 4.2105,
+      "step": 382
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 2.286e-05,
-      "loss": 4.2084,
-      "step": 384
+      "loss": 4.1431,
+      "step": 383
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 2.292e-05,
-      "loss": 4.1318,
-      "step": 385
+      "loss": 4.0593,
+      "step": 384
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 2.298e-05,
-      "loss": 4.1975,
-      "step": 386
+      "loss": 4.16,
+      "step": 385
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 2.304e-05,
-      "loss": 4.1145,
-      "step": 387
+      "loss": 4.0336,
+      "step": 386
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 2.3100000000000002e-05,
-      "loss": 4.007,
-      "step": 388
+      "loss": 4.2153,
+      "step": 387
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 2.3160000000000002e-05,
-      "loss": 4.153,
-      "step": 389
+      "loss": 4.2426,
+      "step": 388
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 2.322e-05,
-      "loss": 4.2659,
-      "step": 390
+      "loss": 4.1324,
+      "step": 389
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 2.328e-05,
-      "loss": 4.3676,
-      "step": 391
+      "loss": 4.126,
+      "step": 390
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 2.334e-05,
-      "loss": 4.0016,
-      "step": 392
+      "loss": 4.2552,
+      "step": 391
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.1861,
-      "step": 393
+      "loss": 4.0773,
+      "step": 392
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 2.3460000000000002e-05,
-      "loss": 4.325,
-      "step": 394
+      "loss": 4.0848,
+      "step": 393
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 2.3520000000000002e-05,
-      "loss": 4.5118,
-      "step": 395
+      "loss": 4.1795,
+      "step": 394
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 2.358e-05,
-      "loss": 4.2486,
-      "step": 396
+      "loss": 4.3042,
+      "step": 395
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 2.364e-05,
-      "loss": 4.2231,
-      "step": 397
+      "loss": 4.2524,
+      "step": 396
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 2.37e-05,
-      "loss": 4.6555,
-      "step": 398
+      "loss": 4.4209,
+      "step": 397
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 2.3760000000000003e-05,
-      "loss": 4.428,
-      "step": 399
+      "loss": 4.3868,
+      "step": 398
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 2.3820000000000002e-05,
-      "loss": 4.5045,
-      "step": 400
+      "loss": 4.5116,
+      "step": 399
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 2.3880000000000002e-05,
-      "loss": 4.2833,
-      "step": 401
+      "loss": 4.4561,
+      "step": 400
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 2.394e-05,
-      "loss": 4.1244,
-      "step": 402
+      "loss": 4.1715,
+      "step": 401
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 2.4e-05,
-      "loss": 4.235,
-      "step": 403
+      "loss": 4.018,
+      "step": 402
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 2.4060000000000003e-05,
-      "loss": 4.17,
-      "step": 404
+      "loss": 4.0975,
+      "step": 403
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 2.4120000000000003e-05,
-      "loss": 4.2928,
-      "step": 405
+      "loss": 4.0795,
+      "step": 404
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 2.4180000000000002e-05,
-      "loss": 4.1763,
-      "step": 406
+      "loss": 4.0454,
+      "step": 405
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 2.4240000000000002e-05,
-      "loss": 4.1616,
-      "step": 407
+      "loss": 4.1319,
+      "step": 406
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 2.43e-05,
-      "loss": 4.0906,
-      "step": 408
+      "loss": 4.1041,
+      "step": 407
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 2.4360000000000004e-05,
-      "loss": 4.3761,
-      "step": 409
+      "loss": 4.1619,
+      "step": 408
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 2.442e-05,
-      "loss": 4.1907,
-      "step": 410
+      "loss": 4.1178,
+      "step": 409
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 2.448e-05,
-      "loss": 4.0623,
-      "step": 411
+      "loss": 4.0931,
+      "step": 410
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 2.454e-05,
-      "loss": 4.1642,
-      "step": 412
+      "loss": 4.314,
+      "step": 411
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 2.4599999999999998e-05,
-      "loss": 4.2181,
-      "step": 413
+      "loss": 4.0508,
+      "step": 412
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 2.4659999999999998e-05,
-      "loss": 4.0526,
-      "step": 414
+      "loss": 4.003,
+      "step": 413
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 2.472e-05,
-      "loss": 4.0498,
-      "step": 415
+      "loss": 4.1375,
+      "step": 414
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 2.478e-05,
-      "loss": 4.2762,
-      "step": 416
+      "loss": 4.2206,
+      "step": 415
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 2.484e-05,
-      "loss": 4.0794,
-      "step": 417
+      "loss": 4.0785,
+      "step": 416
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 2.49e-05,
-      "loss": 4.2228,
-      "step": 418
+      "loss": 4.1078,
+      "step": 417
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 2.4959999999999998e-05,
-      "loss": 4.1585,
-      "step": 419
+      "loss": 4.0014,
+      "step": 418
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 2.502e-05,
-      "loss": 4.1818,
-      "step": 420
+      "loss": 4.0941,
+      "step": 419
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 2.508e-05,
-      "loss": 4.2265,
-      "step": 421
+      "loss": 4.0181,
+      "step": 420
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 2.514e-05,
-      "loss": 4.2701,
-      "step": 422
+      "loss": 4.1846,
+      "step": 421
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 2.52e-05,
-      "loss": 4.2883,
-      "step": 423
+      "loss": 4.0393,
+      "step": 422
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 2.526e-05,
-      "loss": 4.1966,
-      "step": 424
+      "loss": 4.1369,
+      "step": 423
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 2.5319999999999998e-05,
-      "loss": 4.2658,
-      "step": 425
+      "loss": 4.1978,
+      "step": 424
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 2.538e-05,
-      "loss": 4.286,
-      "step": 426
+      "loss": 4.1363,
+      "step": 425
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 2.544e-05,
-      "loss": 4.1286,
-      "step": 427
+      "loss": 4.1793,
+      "step": 426
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 2.55e-05,
-      "loss": 4.3255,
-      "step": 428
+      "loss": 4.1114,
+      "step": 427
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 2.556e-05,
-      "loss": 4.1596,
-      "step": 429
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.562e-05,
-      "loss": 4.3055,
-      "step": 430
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.568e-05,
-      "loss": 4.3901,
-      "step": 431
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.574e-05,
-      "loss": 4.2223,
-      "step": 432
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.58e-05,
-      "loss": 3.9652,
-      "step": 433
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.586e-05,
-      "loss": 4.1636,
-      "step": 434
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.592e-05,
-      "loss": 4.1553,
-      "step": 435
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.5980000000000002e-05,
-      "loss": 4.3838,
-      "step": 436
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.604e-05,
-      "loss": 4.3877,
-      "step": 437
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.61e-05,
-      "loss": 4.2501,
-      "step": 438
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.616e-05,
-      "loss": 4.0833,
-      "step": 439
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.622e-05,
-      "loss": 4.1429,
-      "step": 440
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.628e-05,
-      "loss": 4.2947,
-      "step": 441
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.6340000000000002e-05,
-      "loss": 4.1863,
-      "step": 442
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.64e-05,
-      "loss": 4.2328,
-      "step": 443
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.646e-05,
-      "loss": 4.4469,
-      "step": 444
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.652e-05,
-      "loss": 4.2335,
-      "step": 445
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.658e-05,
-      "loss": 4.0421,
-      "step": 446
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.6640000000000002e-05,
-      "loss": 4.2282,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.6700000000000002e-05,
-      "loss": 4.4805,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.676e-05,
-      "loss": 4.3042,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.682e-05,
-      "loss": 4.8199,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.688e-05,
-      "loss": 4.2232,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.6940000000000003e-05,
-      "loss": 4.2092,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 4.352,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.7060000000000002e-05,
-      "loss": 4.0295,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.712e-05,
-      "loss": 4.3023,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.718e-05,
-      "loss": 4.2455,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.724e-05,
-      "loss": 4.1829,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.7300000000000003e-05,
-      "loss": 4.1425,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.7360000000000002e-05,
-      "loss": 4.243,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.7420000000000002e-05,
-      "loss": 4.3147,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.748e-05,
-      "loss": 4.0656,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.754e-05,
-      "loss": 4.103,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 4.1576,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.7660000000000003e-05,
-      "loss": 4.0859,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.7720000000000002e-05,
-      "loss": 4.2193,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.778e-05,
-      "loss": 4.2194,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.784e-05,
-      "loss": 4.1411,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.79e-05,
-      "loss": 4.0721,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.7960000000000003e-05,
-      "loss": 4.1481,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.8020000000000003e-05,
-      "loss": 4.2244,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.8080000000000002e-05,
-      "loss": 4.0628,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.8139999999999998e-05,
-      "loss": 4.0549,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.243,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.826e-05,
-      "loss": 4.2145,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.832e-05,
-      "loss": 4.1422,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.838e-05,
-      "loss": 4.3287,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.844e-05,
-      "loss": 4.3965,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.8499999999999998e-05,
-      "loss": 4.0726,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.856e-05,
-      "loss": 3.9443,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.862e-05,
-      "loss": 4.264,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.868e-05,
-      "loss": 4.0944,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.874e-05,
-      "loss": 4.0911,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.88e-05,
-      "loss": 4.2576,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.8859999999999998e-05,
-      "loss": 4.1463,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.892e-05,
-      "loss": 4.3166,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.898e-05,
-      "loss": 4.0712,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.904e-05,
-      "loss": 3.9972,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.91e-05,
-      "loss": 4.127,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.916e-05,
-      "loss": 4.1673,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.922e-05,
-      "loss": 4.303,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.928e-05,
-      "loss": 4.0828,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.934e-05,
-      "loss": 4.3801,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.94e-05,
-      "loss": 4.3874,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.946e-05,
-      "loss": 4.1561,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 2.9520000000000002e-05,
-      "loss": 4.2939,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.958e-05,
-      "loss": 4.0775,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.964e-05,
-      "loss": 4.0271,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.97e-05,
-      "loss": 4.7244,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.976e-05,
-      "loss": 4.6151,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.982e-05,
-      "loss": 4.3507,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 4.249406337738037,
-      "eval_runtime": 954.2128,
-      "eval_samples_per_second": 2.769,
-      "eval_steps_per_second": 0.347,
-      "eval_wer": 1.931356604522015,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.9880000000000002e-05,
-      "loss": 4.4717,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.994e-05,
-      "loss": 4.1174,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 3e-05,
-      "loss": 4.3457,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.9923469387755103e-05,
-      "loss": 4.1425,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.9846938775510208e-05,
-      "loss": 4.0916,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.9770408163265306e-05,
-      "loss": 4.23,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.969387755102041e-05,
-      "loss": 4.1879,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.961734693877551e-05,
-      "loss": 4.2693,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.9540816326530616e-05,
-      "loss": 4.1314,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.9464285714285714e-05,
-      "loss": 4.2637,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.9387755102040816e-05,
-      "loss": 4.2189,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 2.9311224489795918e-05,
-      "loss": 4.2026,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.9234693877551023e-05,
-      "loss": 4.4774,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.9158163265306125e-05,
-      "loss": 4.1445,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.9081632653061224e-05,
-      "loss": 4.1927,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.9005102040816326e-05,
-      "loss": 4.0574,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.892857142857143e-05,
-      "loss": 4.2716,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.8852040816326533e-05,
-      "loss": 4.1753,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.877551020408163e-05,
-      "loss": 4.2727,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.8698979591836733e-05,
-      "loss": 4.1709,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.862244897959184e-05,
-      "loss": 4.1588,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.854591836734694e-05,
-      "loss": 4.104,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.8469387755102042e-05,
-      "loss": 4.0468,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.8392857142857144e-05,
-      "loss": 4.2316,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.8316326530612246e-05,
-      "loss": 4.3244,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.8239795918367348e-05,
-      "loss": 4.0525,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.816326530612245e-05,
-      "loss": 4.2284,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.8086734693877552e-05,
-      "loss": 4.1786,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.8010204081632654e-05,
-      "loss": 4.14,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 2.7933673469387756e-05,
-      "loss": 4.4132,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.7857142857142858e-05,
-      "loss": 4.273,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.778061224489796e-05,
-      "loss": 3.9821,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.770408163265306e-05,
-      "loss": 4.188,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.7627551020408163e-05,
-      "loss": 4.0577,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.7551020408163265e-05,
-      "loss": 4.2338,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.747448979591837e-05,
-      "loss": 4.2538,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.739795918367347e-05,
-      "loss": 4.3179,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.732142857142857e-05,
-      "loss": 4.3678,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 2.7244897959183673e-05,
-      "loss": 4.1675,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.7168367346938778e-05,
-      "loss": 4.2616,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.7091836734693877e-05,
-      "loss": 4.2889,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.701530612244898e-05,
-      "loss": 4.3385,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.6938775510204084e-05,
-      "loss": 4.4943,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.6862244897959186e-05,
-      "loss": 4.3588,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.6785714285714288e-05,
-      "loss": 4.386,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.6709183673469386e-05,
-      "loss": 4.4699,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.663265306122449e-05,
-      "loss": 4.3611,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 2.6556122448979593e-05,
-      "loss": 4.6624,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.6479591836734695e-05,
-      "loss": 4.1314,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.6403061224489794e-05,
-      "loss": 5.1441,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.63265306122449e-05,
-      "loss": 4.1062,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.625e-05,
-      "loss": 4.3416,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.6173469387755103e-05,
-      "loss": 4.2483,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.6096938775510205e-05,
-      "loss": 4.1344,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.6020408163265307e-05,
-      "loss": 4.1248,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.594387755102041e-05,
-      "loss": 4.1874,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 2.586734693877551e-05,
-      "loss": 4.0656,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.5790816326530613e-05,
-      "loss": 4.128,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.5714285714285714e-05,
-      "loss": 4.1983,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.5637755102040816e-05,
-      "loss": 4.0579,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.5561224489795918e-05,
-      "loss": 4.2611,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.548469387755102e-05,
-      "loss": 4.094,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.5408163265306125e-05,
-      "loss": 4.0976,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.5331632653061224e-05,
-      "loss": 4.1275,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.5255102040816326e-05,
-      "loss": 4.2178,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 2.517857142857143e-05,
-      "loss": 4.2822,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.5102040816326533e-05,
-      "loss": 4.1375,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.502551020408163e-05,
-      "loss": 4.0841,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.4948979591836734e-05,
-      "loss": 4.2421,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.487244897959184e-05,
-      "loss": 4.067,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.479591836734694e-05,
-      "loss": 4.0541,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.4719387755102043e-05,
-      "loss": 4.2383,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.464285714285714e-05,
-      "loss": 4.0874,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.4566326530612246e-05,
-      "loss": 4.0371,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 2.448979591836735e-05,
-      "loss": 4.2899,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.441326530612245e-05,
-      "loss": 4.0796,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.433673469387755e-05,
-      "loss": 4.1797,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.4260204081632654e-05,
-      "loss": 4.3384,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.4183673469387756e-05,
-      "loss": 3.8088,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.4107142857142858e-05,
-      "loss": 4.3933,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.403061224489796e-05,
-      "loss": 4.2551,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.3954081632653062e-05,
-      "loss": 4.0488,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.3877551020408164e-05,
-      "loss": 4.1074,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.3801020408163266e-05,
-      "loss": 4.1928,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.372448979591837e-05,
-      "loss": 4.3976,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.364795918367347e-05,
-      "loss": 4.2446,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.357142857142857e-05,
-      "loss": 4.255,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.3494897959183673e-05,
-      "loss": 4.2302,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.341836734693878e-05,
-      "loss": 3.9942,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.3341836734693877e-05,
-      "loss": 4.1137,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.326530612244898e-05,
-      "loss": 4.2067,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.318877551020408e-05,
-      "loss": 4.3063,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.3112244897959186e-05,
-      "loss": 4.4976,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.3035714285714288e-05,
-      "loss": 4.2891,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.2959183673469387e-05,
-      "loss": 4.2596,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.288265306122449e-05,
-      "loss": 4.0942,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.2806122448979594e-05,
-      "loss": 4.2176,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.2729591836734696e-05,
-      "loss": 4.3389,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.2653061224489794e-05,
-      "loss": 4.223,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.2576530612244896e-05,
-      "loss": 4.1904,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.25e-05,
-      "loss": 4.2231,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.2423469387755103e-05,
-      "loss": 4.0458,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.2346938775510205e-05,
-      "loss": 4.1966,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.2270408163265307e-05,
-      "loss": 4.1522,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.219387755102041e-05,
-      "loss": 4.1061,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.211734693877551e-05,
-      "loss": 4.1197,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.2040816326530613e-05,
-      "loss": 4.2229,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.1964285714285715e-05,
-      "loss": 4.1381,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.1887755102040817e-05,
-      "loss": 4.1293,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.181122448979592e-05,
-      "loss": 4.14,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.173469387755102e-05,
-      "loss": 4.1027,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1658163265306126e-05,
-      "loss": 4.2896,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1581632653061224e-05,
-      "loss": 4.1789,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1505102040816326e-05,
-      "loss": 4.3021,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1428571428571428e-05,
-      "loss": 4.1265,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1352040816326533e-05,
-      "loss": 4.1473,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1275510204081632e-05,
-      "loss": 4.0719,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1198979591836734e-05,
-      "loss": 4.1128,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.1122448979591836e-05,
-      "loss": 4.0491,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.104591836734694e-05,
-      "loss": 4.1953,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.0969387755102043e-05,
-      "loss": 4.0822,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.089285714285714e-05,
-      "loss": 4.0029,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.0816326530612247e-05,
-      "loss": 4.2648,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.073979591836735e-05,
-      "loss": 4.0931,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.066326530612245e-05,
-      "loss": 4.3205,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.058673469387755e-05,
-      "loss": 4.2901,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.0510204081632654e-05,
-      "loss": 4.181,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.0433673469387756e-05,
-      "loss": 4.1243,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.0357142857142858e-05,
-      "loss": 4.1229,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.0280612244897957e-05,
-      "loss": 4.2929,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.0204081632653062e-05,
-      "loss": 4.2063,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.0127551020408164e-05,
-      "loss": 4.0951,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.0051020408163266e-05,
-      "loss": 4.1848,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.9974489795918368e-05,
-      "loss": 4.1729,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.989795918367347e-05,
-      "loss": 4.281,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.982142857142857e-05,
-      "loss": 4.2894,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.9744897959183674e-05,
-      "loss": 4.1699,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9668367346938775e-05,
-      "loss": 4.0088,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9591836734693877e-05,
-      "loss": 4.2616,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.951530612244898e-05,
-      "loss": 4.0165,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.943877551020408e-05,
-      "loss": 4.1598,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9362244897959186e-05,
-      "loss": 4.5368,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.928571428571429e-05,
-      "loss": 4.3574,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9209183673469387e-05,
-      "loss": 4.1234,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.913265306122449e-05,
-      "loss": 3.962,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9056122448979594e-05,
-      "loss": 4.2601,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8979591836734696e-05,
-      "loss": 4.1748,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8903061224489795e-05,
-      "loss": 4.1067,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8826530612244896e-05,
-      "loss": 4.4876,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8750000000000002e-05,
-      "loss": 4.2195,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8673469387755104e-05,
-      "loss": 4.2406,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8596938775510206e-05,
-      "loss": 4.2498,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8520408163265304e-05,
-      "loss": 4.2465,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.844387755102041e-05,
-      "loss": 4.2118,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.836734693877551e-05,
-      "loss": 4.1716,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.8290816326530613e-05,
-      "loss": 4.1002,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.8214285714285712e-05,
-      "loss": 4.1314,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.8137755102040817e-05,
-      "loss": 4.3314,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.806122448979592e-05,
-      "loss": 4.148,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.798469387755102e-05,
-      "loss": 4.2266,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7908163265306126e-05,
-      "loss": 3.9961,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7831632653061225e-05,
-      "loss": 3.9725,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7755102040816327e-05,
-      "loss": 4.325,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.767857142857143e-05,
-      "loss": 4.0709,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.7602040816326534e-05,
-      "loss": 4.0242,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.7525510204081632e-05,
-      "loss": 4.2876,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.7448979591836734e-05,
-      "loss": 4.0623,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.7372448979591836e-05,
-      "loss": 4.2775,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.729591836734694e-05,
-      "loss": 4.237,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.7219387755102043e-05,
-      "loss": 4.2058,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.7142857142857142e-05,
-      "loss": 4.0794,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.7066326530612244e-05,
-      "loss": 3.9794,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.698979591836735e-05,
-      "loss": 3.9593,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.691326530612245e-05,
-      "loss": 4.0232,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.683673469387755e-05,
-      "loss": 4.0226,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.676020408163265e-05,
-      "loss": 4.1974,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.6683673469387757e-05,
-      "loss": 3.9916,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.660714285714286e-05,
-      "loss": 3.9367,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.6530612244897957e-05,
-      "loss": 3.9962,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.6454081632653062e-05,
-      "loss": 4.0845,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.6377551020408164e-05,
-      "loss": 4.3325,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.6301020408163266e-05,
-      "loss": 4.0054,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.6224489795918368e-05,
-      "loss": 4.1694,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.614795918367347e-05,
-      "loss": 4.1841,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.6071428571428572e-05,
-      "loss": 4.2737,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.5994897959183674e-05,
-      "loss": 4.1188,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.5918367346938776e-05,
-      "loss": 4.1621,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.5841836734693878e-05,
-      "loss": 4.2523,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.576530612244898e-05,
-      "loss": 4.272,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.568877551020408e-05,
-      "loss": 4.3658,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.5612244897959183e-05,
-      "loss": 4.0139,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.553571428571429e-05,
-      "loss": 4.1947,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.5459183673469387e-05,
-      "loss": 4.0925,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.538265306122449e-05,
-      "loss": 4.4659,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.530612244897959e-05,
-      "loss": 4.269,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.5229591836734695e-05,
-      "loss": 4.2835,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.5153061224489795e-05,
-      "loss": 4.3569,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.5076530612244897e-05,
-      "loss": 4.3019,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.5e-05,
-      "loss": 4.3235,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.4923469387755104e-05,
-      "loss": 4.4127,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4846938775510204e-05,
-      "loss": 4.0454,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4770408163265308e-05,
-      "loss": 4.1379,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4693877551020408e-05,
-      "loss": 4.0424,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4617346938775512e-05,
-      "loss": 3.9108,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4540816326530612e-05,
-      "loss": 4.0812,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4464285714285715e-05,
-      "loss": 4.2139,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4387755102040816e-05,
-      "loss": 4.23,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.431122448979592e-05,
-      "loss": 4.1661,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.4234693877551021e-05,
-      "loss": 4.0422,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.4158163265306123e-05,
-      "loss": 4.1259,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.4081632653061225e-05,
-      "loss": 4.0219,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.4005102040816327e-05,
-      "loss": 4.0935,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.3928571428571429e-05,
-      "loss": 4.0343,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.385204081632653e-05,
-      "loss": 4.2476,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.3775510204081633e-05,
-      "loss": 4.2593,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.3698979591836735e-05,
-      "loss": 4.0613,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.3622448979591836e-05,
-      "loss": 4.2907,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.3545918367346938e-05,
-      "loss": 4.057,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.3469387755102042e-05,
-      "loss": 3.8796,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.3392857142857144e-05,
-      "loss": 4.1651,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.3316326530612246e-05,
-      "loss": 3.965,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.3239795918367348e-05,
-      "loss": 4.017,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.316326530612245e-05,
-      "loss": 4.0912,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.3086734693877551e-05,
-      "loss": 4.1649,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.3010204081632653e-05,
-      "loss": 4.0581,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.2933673469387755e-05,
-      "loss": 4.1833,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2857142857142857e-05,
-      "loss": 4.1175,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2780612244897959e-05,
-      "loss": 4.0868,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2704081632653063e-05,
-      "loss": 3.9839,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2627551020408163e-05,
-      "loss": 4.2042,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2551020408163267e-05,
-      "loss": 4.3366,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2474489795918367e-05,
-      "loss": 4.1627,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.239795918367347e-05,
-      "loss": 4.0717,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.232142857142857e-05,
-      "loss": 4.2349,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2244897959183674e-05,
-      "loss": 3.9495,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2168367346938774e-05,
-      "loss": 4.072,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2091836734693878e-05,
-      "loss": 4.2674,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.201530612244898e-05,
-      "loss": 4.1501,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.1938775510204082e-05,
-      "loss": 4.0789,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.1862244897959185e-05,
-      "loss": 4.2417,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.1785714285714286e-05,
-      "loss": 4.2802,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.170918367346939e-05,
-      "loss": 4.066,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.163265306122449e-05,
-      "loss": 3.953,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.1556122448979593e-05,
-      "loss": 4.0866,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1479591836734693e-05,
-      "loss": 4.2401,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1403061224489797e-05,
-      "loss": 4.3716,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1326530612244897e-05,
-      "loss": 4.2219,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.125e-05,
-      "loss": 4.3632,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1173469387755103e-05,
-      "loss": 4.5002,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1096938775510205e-05,
-      "loss": 4.2068,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1020408163265306e-05,
-      "loss": 4.2042,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.0943877551020408e-05,
-      "loss": 4.1899,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.086734693877551e-05,
-      "loss": 4.1472,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0790816326530612e-05,
-      "loss": 4.037,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0714285714285714e-05,
-      "loss": 4.1912,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0637755102040816e-05,
-      "loss": 4.2355,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0561224489795918e-05,
-      "loss": 4.1236,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0484693877551021e-05,
-      "loss": 4.2208,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0408163265306123e-05,
-      "loss": 4.0635,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0331632653061225e-05,
-      "loss": 4.1248,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0255102040816327e-05,
-      "loss": 4.0792,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0178571428571429e-05,
-      "loss": 4.1373,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.0102040816326531e-05,
-      "loss": 4.0515,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.0025510204081633e-05,
-      "loss": 4.0731,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.948979591836735e-06,
-      "loss": 4.1974,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.872448979591837e-06,
-      "loss": 4.0851,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.795918367346939e-06,
-      "loss": 4.0982,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.71938775510204e-06,
-      "loss": 4.1605,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.642857142857144e-06,
-      "loss": 4.1659,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.566326530612244e-06,
-      "loss": 4.0684,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.489795918367348e-06,
-      "loss": 4.0657,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.413265306122448e-06,
-      "loss": 4.238,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.336734693877552e-06,
-      "loss": 4.2754,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.260204081632652e-06,
-      "loss": 3.915,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.183673469387756e-06,
-      "loss": 3.9955,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.107142857142856e-06,
-      "loss": 4.1319,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.03061224489796e-06,
-      "loss": 3.9204,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.954081632653063e-06,
-      "loss": 4.1897,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.877551020408163e-06,
-      "loss": 4.0072,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.801020408163267e-06,
-      "loss": 4.0125,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.724489795918367e-06,
-      "loss": 3.9928,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.64795918367347e-06,
-      "loss": 4.1519,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.571428571428571e-06,
-      "loss": 4.1489,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.494897959183675e-06,
-      "loss": 4.0451,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.418367346938775e-06,
-      "loss": 4.0635,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.341836734693878e-06,
-      "loss": 3.9523,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.265306122448979e-06,
-      "loss": 4.1608,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.188775510204082e-06,
-      "loss": 4.1575,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.112244897959184e-06,
-      "loss": 4.0074,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 8.035714285714286e-06,
-      "loss": 4.1326,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.959183673469388e-06,
-      "loss": 4.1211,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.88265306122449e-06,
-      "loss": 3.9565,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.806122448979592e-06,
-      "loss": 4.1213,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.729591836734694e-06,
-      "loss": 4.2507,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.653061224489796e-06,
-      "loss": 4.2829,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.5765306122448974e-06,
-      "loss": 4.1432,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.5e-06,
-      "loss": 4.2565,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.423469387755102e-06,
-      "loss": 4.3092,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.346938775510204e-06,
-      "loss": 4.6345,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.270408163265306e-06,
-      "loss": 4.5129,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.193877551020408e-06,
-      "loss": 4.1561,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.117346938775511e-06,
-      "loss": 4.0536,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.0408163265306125e-06,
-      "loss": 4.0601,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.964285714285714e-06,
-      "loss": 4.122,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.887755102040816e-06,
-      "loss": 3.9336,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.811224489795918e-06,
-      "loss": 4.1128,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.734693877551021e-06,
-      "loss": 4.183,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.658163265306123e-06,
-      "loss": 3.9993,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.581632653061225e-06,
-      "loss": 4.2294,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.505102040816327e-06,
-      "loss": 3.9998,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.428571428571429e-06,
-      "loss": 4.2926,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.352040816326531e-06,
-      "loss": 4.1365,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.275510204081633e-06,
-      "loss": 3.9426,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.198979591836735e-06,
-      "loss": 4.0211,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.122448979591837e-06,
-      "loss": 4.0301,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.045918367346939e-06,
-      "loss": 4.017,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.969387755102041e-06,
-      "loss": 4.0372,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.892857142857143e-06,
-      "loss": 4.0688,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.816326530612245e-06,
-      "loss": 4.0503,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.739795918367347e-06,
-      "loss": 4.0608,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.6632653061224486e-06,
-      "loss": 3.9885,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.586734693877551e-06,
-      "loss": 4.0087,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.510204081632653e-06,
-      "loss": 3.9787,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.433673469387755e-06,
-      "loss": 4.0661,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.357142857142857e-06,
-      "loss": 4.3232,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.280612244897959e-06,
-      "loss": 4.2064,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.204081632653062e-06,
-      "loss": 4.1131,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.127551020408164e-06,
-      "loss": 4.0079,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.0510204081632655e-06,
-      "loss": 3.988,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.9744897959183674e-06,
-      "loss": 4.1855,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.897959183673469e-06,
-      "loss": 3.9875,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.821428571428572e-06,
-      "loss": 3.9255,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.744897959183674e-06,
-      "loss": 4.2085,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.668367346938776e-06,
-      "loss": 4.0791,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.591836734693878e-06,
-      "loss": 4.1169,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.51530612244898e-06,
-      "loss": 4.096,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.438775510204082e-06,
-      "loss": 4.1276,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.3622448979591836e-06,
-      "loss": 4.1881,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.2857142857142855e-06,
-      "loss": 3.9964,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.209183673469387e-06,
-      "loss": 4.1282,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.132653061224489e-06,
-      "loss": 4.2978,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.056122448979592e-06,
-      "loss": 4.1185,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.979591836734694e-06,
-      "loss": 4.298,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.903061224489796e-06,
-      "loss": 4.2988,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.826530612244898e-06,
-      "loss": 4.3789,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.75e-06,
-      "loss": 4.5702,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.673469387755102e-06,
-      "loss": 4.1955,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.596938775510204e-06,
-      "loss": 4.5673,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.5204081632653062e-06,
-      "loss": 4.5096,
-      "step": 849
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.443877551020408e-06,
-      "loss": 4.2798,
-      "step": 850
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.3673469387755105e-06,
-      "loss": 4.0646,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.2908163265306124e-06,
-      "loss": 3.978,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.2142857142857143e-06,
-      "loss": 4.0582,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.1377551020408166e-06,
-      "loss": 4.0127,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.0612244897959185e-06,
-      "loss": 3.9874,
-      "step": 855
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.9846938775510205e-06,
-      "loss": 4.054,
-      "step": 856
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.9081632653061224e-06,
-      "loss": 4.1327,
-      "step": 857
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.8316326530612243e-06,
-      "loss": 4.1236,
-      "step": 858
+      "loss": 4.1015,
+      "step": 428
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.7551020408163266e-06,
-      "loss": 3.9831,
-      "step": 859
+      "learning_rate": 2.562e-05,
+      "loss": 4.0305,
+      "step": 429
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.6785714285714285e-06,
-      "loss": 4.0765,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.602040816326531e-06,
-      "loss": 4.2494,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.5255102040816328e-06,
-      "loss": 3.9901,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.4489795918367347e-06,
-      "loss": 4.1015,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.372448979591837e-06,
-      "loss": 4.2779,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.295918367346939e-06,
-      "loss": 4.1575,
-      "step": 865
+      "learning_rate": 2.568e-05,
+      "loss": 4.1179,
+      "step": 430
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.219387755102041e-06,
-      "loss": 3.8855,
-      "step": 866
+      "learning_rate": 2.574e-05,
+      "loss": 4.1957,
+      "step": 431
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.1428571428571427e-06,
-      "loss": 4.0976,
-      "step": 867
+      "learning_rate": 2.58e-05,
+      "loss": 4.3161,
+      "step": 432
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.0663265306122446e-06,
-      "loss": 3.9664,
-      "step": 868
+      "learning_rate": 2.586e-05,
+      "loss": 4.1238,
+      "step": 433
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.989795918367347e-06,
-      "loss": 4.0142,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.913265306122449e-06,
-      "loss": 4.1759,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.836734693877551e-06,
-      "loss": 3.9869,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.7602040816326531e-06,
-      "loss": 4.153,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.6836734693877552e-06,
-      "loss": 4.1535,
-      "step": 873
+      "learning_rate": 2.592e-05,
+      "loss": 4.1152,
+      "step": 434
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.6071428571428572e-06,
-      "loss": 4.1515,
-      "step": 874
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 4.1593,
+      "step": 435
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.5306122448979593e-06,
-      "loss": 4.2312,
-      "step": 875
+      "learning_rate": 2.604e-05,
+      "loss": 4.0797,
+      "step": 436
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.4540816326530612e-06,
-      "loss": 3.9983,
-      "step": 876
+      "learning_rate": 2.61e-05,
+      "loss": 4.0625,
+      "step": 437
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.3775510204081633e-06,
-      "loss": 4.0585,
-      "step": 877
+      "learning_rate": 2.616e-05,
+      "loss": 4.2553,
+      "step": 438
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.3010204081632654e-06,
-      "loss": 4.1122,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.2244897959183673e-06,
-      "loss": 4.2815,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.1479591836734695e-06,
-      "loss": 4.3714,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.0714285714285714e-06,
-      "loss": 3.9905,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 9.948979591836735e-07,
-      "loss": 4.1339,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 9.183673469387755e-07,
-      "loss": 4.0394,
-      "step": 883
+      "learning_rate": 2.622e-05,
+      "loss": 4.1998,
+      "step": 439
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.418367346938776e-07,
-      "loss": 4.0357,
-      "step": 884
+      "learning_rate": 2.628e-05,
+      "loss": 4.2637,
+      "step": 440
     },
     {
       "epoch": 0.99,
-      "learning_rate": 7.653061224489796e-07,
-      "loss": 4.096,
-      "step": 885
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 4.2109,
+      "step": 441
     },
     {
       "epoch": 0.99,
-      "learning_rate": 6.887755102040817e-07,
-      "loss": 4.1527,
-      "step": 886
+      "learning_rate": 2.64e-05,
+      "loss": 4.4059,
+      "step": 442
     },
     {
       "epoch": 0.99,
-      "learning_rate": 6.122448979591837e-07,
-      "loss": 4.0574,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.357142857142857e-07,
-      "loss": 4.2464,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 4.5918367346938775e-07,
-      "loss": 4.2688,
-      "step": 889
+      "learning_rate": 2.646e-05,
+      "loss": 4.3216,
+      "step": 443
     },
     {
       "epoch": 1.0,
-      "learning_rate": 3.826530612244898e-07,
-      "loss": 4.1339,
-      "step": 890
+      "learning_rate": 2.652e-05,
+      "loss": 4.2793,
+      "step": 444
     },
     {
       "epoch": 1.0,
-      "learning_rate": 3.0612244897959183e-07,
-      "loss": 4.3171,
-      "step": 891
+      "learning_rate": 2.658e-05,
+      "loss": 4.5485,
+      "step": 445
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.2959183673469388e-07,
-      "loss": 3.8865,
-      "step": 892
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 4.2807,
+      "step": 446
     },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 4.249312744012328,
-      "train_runtime": 6250.8757,
-      "train_samples_per_second": 4.565,
-      "train_steps_per_second": 0.143
+      "train_loss": 4.298295111399595,
+      "train_runtime": 5163.9194,
+      "train_samples_per_second": 5.526,
+      "train_steps_per_second": 0.086
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,