diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,1805 +1,10754 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9987389659520807,
-  "global_step": 297,
+  "epoch": 1.0,
+  "global_step": 1784,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 4.767,
+      "learning_rate": 0.0,
+      "loss": 4.9491,
       "step": 1
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.8415,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.0663,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-07,
-      "loss": 4.9166,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.1788,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.8629,
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.7043,
       "step": 4
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.7452,
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.7991,
       "step": 5
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.8808,
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.8392,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-07,
-      "loss": 4.8603,
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.7255,
       "step": 7
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.2e-07,
-      "loss": 4.9323,
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.8788,
       "step": 8
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 4.8445,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 5.0804,
       "step": 9
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-07,
-      "loss": 4.7068,
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.7679,
       "step": 10
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.8473,
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.8014,
       "step": 11
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.599999999999999e-07,
-      "loss": 4.8768,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 4.766,
       "step": 12
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.2e-07,
-      "loss": 4.7908,
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.955,
       "step": 13
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.799999999999999e-07,
-      "loss": 4.7452,
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 4.5677,
       "step": 14
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.4e-07,
-      "loss": 4.7585,
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.6962,
       "step": 15
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9e-07,
-      "loss": 4.6847,
+      "epoch": 0.01,
+      "learning_rate": 2.6e-06,
+      "loss": 4.7875,
       "step": 16
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 4.7876,
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.7788,
       "step": 17
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.0200000000000002e-06,
-      "loss": 4.7936,
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 4.6534,
       "step": 18
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.08e-06,
-      "loss": 4.7411,
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.6281,
       "step": 19
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.14e-06,
-      "loss": 4.8271,
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.8938,
       "step": 20
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.828,
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 4.575,
       "step": 21
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.26e-06,
-      "loss": 4.7725,
+      "epoch": 0.01,
+      "learning_rate": 3.8e-06,
+      "loss": 4.9363,
       "step": 22
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3199999999999999e-06,
-      "loss": 4.6722,
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.7667,
       "step": 23
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.38e-06,
-      "loss": 4.8135,
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.4783,
       "step": 24
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.44e-06,
-      "loss": 4.6361,
+      "epoch": 0.01,
+      "learning_rate": 4.4e-06,
+      "loss": 4.6545,
       "step": 25
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5e-06,
-      "loss": 4.9185,
+      "epoch": 0.01,
+      "learning_rate": 4.6e-06,
+      "loss": 4.7421,
       "step": 26
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5599999999999999e-06,
-      "loss": 4.654,
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.5436,
       "step": 27
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.62e-06,
-      "loss": 4.6688,
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 4.2914,
       "step": 28
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.68e-06,
-      "loss": 4.5584,
+      "epoch": 0.02,
+      "learning_rate": 5.2e-06,
+      "loss": 4.6401,
       "step": 29
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.74e-06,
-      "loss": 4.7119,
+      "epoch": 0.02,
+      "learning_rate": 5.4e-06,
+      "loss": 4.4864,
       "step": 30
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.8e-06,
-      "loss": 4.6957,
+      "epoch": 0.02,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.378,
       "step": 31
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.86e-06,
-      "loss": 4.6116,
+      "epoch": 0.02,
+      "learning_rate": 5.8e-06,
+      "loss": 4.7201,
       "step": 32
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.6472,
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 4.4078,
       "step": 33
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.98e-06,
-      "loss": 4.5855,
+      "epoch": 0.02,
+      "learning_rate": 6.2e-06,
+      "loss": 4.3083,
       "step": 34
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.0400000000000004e-06,
-      "loss": 4.6201,
+      "epoch": 0.02,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.5288,
       "step": 35
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.5679,
+      "epoch": 0.02,
+      "learning_rate": 6.6e-06,
+      "loss": 4.7637,
       "step": 36
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.16e-06,
-      "loss": 4.6958,
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.4133,
       "step": 37
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.22e-06,
-      "loss": 4.6541,
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 4.5168,
       "step": 38
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.28e-06,
-      "loss": 4.7043,
+      "epoch": 0.02,
+      "learning_rate": 7.2e-06,
+      "loss": 4.6126,
       "step": 39
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.34e-06,
-      "loss": 4.5367,
+      "epoch": 0.02,
+      "learning_rate": 7.4e-06,
+      "loss": 4.6455,
       "step": 40
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.6178,
+      "epoch": 0.02,
+      "learning_rate": 7.6e-06,
+      "loss": 4.4193,
       "step": 41
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.46e-06,
-      "loss": 4.6936,
+      "epoch": 0.02,
+      "learning_rate": 7.8e-06,
+      "loss": 4.2379,
       "step": 42
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.52e-06,
-      "loss": 4.4532,
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.534,
       "step": 43
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.58e-06,
-      "loss": 4.7663,
+      "epoch": 0.02,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 4.339,
       "step": 44
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6399999999999997e-06,
-      "loss": 4.7255,
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.61,
       "step": 45
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.7e-06,
-      "loss": 4.7645,
+      "epoch": 0.03,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 4.5968,
       "step": 46
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.76e-06,
-      "loss": 4.7834,
+      "epoch": 0.03,
+      "learning_rate": 8.8e-06,
+      "loss": 4.6524,
       "step": 47
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.82e-06,
-      "loss": 4.8407,
+      "epoch": 0.03,
+      "learning_rate": 9e-06,
+      "loss": 5.0075,
       "step": 48
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.88e-06,
-      "loss": 4.9009,
+      "epoch": 0.03,
+      "learning_rate": 9.2e-06,
+      "loss": 5.0219,
       "step": 49
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.706,
+      "epoch": 0.03,
+      "learning_rate": 9.4e-06,
+      "loss": 4.9261,
       "step": 50
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3e-06,
-      "loss": 4.4758,
+      "epoch": 0.03,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.2912,
       "step": 51
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.06e-06,
-      "loss": 4.3677,
+      "epoch": 0.03,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.4304,
       "step": 52
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1199999999999998e-06,
-      "loss": 4.3162,
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 4.3948,
       "step": 53
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.18e-06,
-      "loss": 4.3327,
+      "epoch": 0.03,
+      "learning_rate": 1.02e-05,
+      "loss": 4.2804,
       "step": 54
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.24e-06,
-      "loss": 4.269,
+      "epoch": 0.03,
+      "learning_rate": 1.04e-05,
+      "loss": 4.3358,
       "step": 55
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3e-06,
-      "loss": 4.2999,
+      "epoch": 0.03,
+      "learning_rate": 1.06e-05,
+      "loss": 4.4619,
       "step": 56
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.36e-06,
-      "loss": 4.4273,
+      "epoch": 0.03,
+      "learning_rate": 1.08e-05,
+      "loss": 4.3818,
       "step": 57
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.4200000000000003e-06,
-      "loss": 4.4309,
+      "epoch": 0.03,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 4.3221,
       "step": 58
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.48e-06,
-      "loss": 4.3935,
+      "epoch": 0.03,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 4.3537,
       "step": 59
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.54e-06,
-      "loss": 4.3246,
+      "epoch": 0.03,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.2753,
       "step": 60
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6e-06,
-      "loss": 4.3995,
+      "epoch": 0.03,
+      "learning_rate": 1.16e-05,
+      "loss": 4.3486,
       "step": 61
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.66e-06,
-      "loss": 4.2675,
+      "epoch": 0.03,
+      "learning_rate": 1.18e-05,
+      "loss": 4.2601,
       "step": 62
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.72e-06,
-      "loss": 4.4125,
+      "epoch": 0.04,
+      "learning_rate": 1.2e-05,
+      "loss": 4.6203,
       "step": 63
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.3521,
+      "epoch": 0.04,
+      "learning_rate": 1.22e-05,
+      "loss": 4.1402,
       "step": 64
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.4045,
+      "epoch": 0.04,
+      "learning_rate": 1.24e-05,
+      "loss": 4.3295,
       "step": 65
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.9e-06,
-      "loss": 4.406,
+      "epoch": 0.04,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 4.4162,
       "step": 66
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.96e-06,
-      "loss": 4.3524,
+      "epoch": 0.04,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 4.4301,
       "step": 67
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.3942,
+      "epoch": 0.04,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 4.4517,
       "step": 68
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.080000000000001e-06,
-      "loss": 4.3343,
+      "epoch": 0.04,
+      "learning_rate": 1.32e-05,
+      "loss": 4.3618,
       "step": 69
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.14e-06,
-      "loss": 4.266,
+      "epoch": 0.04,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 4.2728,
       "step": 70
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.2738,
+      "epoch": 0.04,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 4.378,
       "step": 71
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.26e-06,
-      "loss": 4.3974,
+      "epoch": 0.04,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.1224,
       "step": 72
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.32e-06,
-      "loss": 4.3188,
+      "epoch": 0.04,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 4.375,
       "step": 73
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.3799999999999996e-06,
-      "loss": 4.3084,
+      "epoch": 0.04,
+      "learning_rate": 1.42e-05,
+      "loss": 4.1553,
       "step": 74
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.44e-06,
-      "loss": 4.4146,
+      "epoch": 0.04,
+      "learning_rate": 1.44e-05,
+      "loss": 4.6745,
       "step": 75
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.5e-06,
-      "loss": 4.2865,
+      "epoch": 0.04,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 4.4299,
       "step": 76
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.56e-06,
-      "loss": 4.2502,
+      "epoch": 0.04,
+      "learning_rate": 1.48e-05,
+      "loss": 4.255,
       "step": 77
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.62e-06,
-      "loss": 4.3103,
+      "epoch": 0.04,
+      "learning_rate": 1.5e-05,
+      "loss": 4.2102,
       "step": 78
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.68e-06,
-      "loss": 4.1937,
+      "epoch": 0.04,
+      "learning_rate": 1.52e-05,
+      "loss": 4.1986,
       "step": 79
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.74e-06,
-      "loss": 4.3765,
+      "epoch": 0.04,
+      "learning_rate": 1.54e-05,
+      "loss": 4.5409,
       "step": 80
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.3016,
+      "epoch": 0.05,
+      "learning_rate": 1.56e-05,
+      "loss": 4.1666,
       "step": 81
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.86e-06,
-      "loss": 4.3088,
+      "epoch": 0.05,
+      "learning_rate": 1.58e-05,
+      "loss": 4.3089,
       "step": 82
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.92e-06,
-      "loss": 4.2784,
+      "epoch": 0.05,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 4.1857,
       "step": 83
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.3069,
+      "epoch": 0.05,
+      "learning_rate": 1.62e-05,
+      "loss": 4.3097,
       "step": 84
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.04e-06,
-      "loss": 4.4127,
+      "epoch": 0.05,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 4.1882,
       "step": 85
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1e-06,
-      "loss": 4.2931,
+      "epoch": 0.05,
+      "learning_rate": 1.66e-05,
+      "loss": 4.1919,
       "step": 86
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.16e-06,
-      "loss": 4.3402,
+      "epoch": 0.05,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.5453,
       "step": 87
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.22e-06,
-      "loss": 4.3675,
+      "epoch": 0.05,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 4.1643,
       "step": 88
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.279999999999999e-06,
-      "loss": 4.3851,
+      "epoch": 0.05,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 4.5001,
       "step": 89
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.34e-06,
-      "loss": 4.4114,
+      "epoch": 0.05,
+      "learning_rate": 1.74e-05,
+      "loss": 4.0727,
       "step": 90
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.4e-06,
-      "loss": 4.4333,
+      "epoch": 0.05,
+      "learning_rate": 1.76e-05,
+      "loss": 4.4653,
       "step": 91
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.46e-06,
-      "loss": 4.4915,
+      "epoch": 0.05,
+      "learning_rate": 1.78e-05,
+      "loss": 4.1227,
       "step": 92
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.52e-06,
-      "loss": 4.4213,
+      "epoch": 0.05,
+      "learning_rate": 1.8e-05,
+      "loss": 3.9717,
       "step": 93
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.58e-06,
-      "loss": 4.3693,
+      "epoch": 0.05,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 4.4034,
       "step": 94
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.64e-06,
-      "loss": 4.3271,
+      "epoch": 0.05,
+      "learning_rate": 1.84e-05,
+      "loss": 4.6244,
       "step": 95
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 4.4958,
+      "epoch": 0.05,
+      "learning_rate": 1.86e-05,
+      "loss": 4.2337,
       "step": 96
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.76e-06,
-      "loss": 4.6801,
+      "epoch": 0.05,
+      "learning_rate": 1.88e-05,
+      "loss": 4.4171,
       "step": 97
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.82e-06,
-      "loss": 4.6378,
+      "epoch": 0.05,
+      "learning_rate": 1.9e-05,
+      "loss": 4.47,
       "step": 98
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.6153,
+      "epoch": 0.06,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.7971,
       "step": 99
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.940000000000001e-06,
-      "loss": 4.6683,
+      "epoch": 0.06,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.653,
       "step": 100
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6e-06,
-      "loss": 4.2246,
+      "epoch": 0.06,
+      "learning_rate": 1.94e-05,
+      "loss": 4.338,
       "step": 101
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 4.2737,
+      "epoch": 0.06,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 4.1293,
       "step": 102
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.12e-06,
-      "loss": 4.2711,
+      "epoch": 0.06,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 4.2627,
       "step": 103
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.18e-06,
-      "loss": 4.1781,
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 4.2222,
       "step": 104
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.2399999999999995e-06,
-      "loss": 4.2846,
+      "epoch": 0.06,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 4.2362,
       "step": 105
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.3e-06,
-      "loss": 4.2148,
+      "epoch": 0.06,
+      "learning_rate": 2.04e-05,
+      "loss": 4.2633,
       "step": 106
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.36e-06,
-      "loss": 4.1611,
+      "epoch": 0.06,
+      "learning_rate": 2.06e-05,
+      "loss": 4.3011,
       "step": 107
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.42e-06,
-      "loss": 4.2767,
+      "epoch": 0.06,
+      "learning_rate": 2.08e-05,
+      "loss": 4.3788,
       "step": 108
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.48e-06,
-      "loss": 4.2806,
+      "epoch": 0.06,
+      "learning_rate": 2.1e-05,
+      "loss": 4.2298,
       "step": 109
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.54e-06,
-      "loss": 4.2561,
+      "epoch": 0.06,
+      "learning_rate": 2.12e-05,
+      "loss": 4.2738,
       "step": 110
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.6e-06,
-      "loss": 4.3238,
+      "epoch": 0.06,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 4.2371,
       "step": 111
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 4.2542,
+      "epoch": 0.06,
+      "learning_rate": 2.16e-05,
+      "loss": 4.1407,
       "step": 112
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.72e-06,
-      "loss": 4.1891,
+      "epoch": 0.06,
+      "learning_rate": 2.18e-05,
+      "loss": 4.2748,
       "step": 113
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.78e-06,
-      "loss": 4.2587,
+      "epoch": 0.06,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.0144,
       "step": 114
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.840000000000001e-06,
-      "loss": 4.2102,
+      "epoch": 0.06,
+      "learning_rate": 2.22e-05,
+      "loss": 4.0539,
       "step": 115
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.900000000000001e-06,
-      "loss": 4.2241,
+      "epoch": 0.07,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 4.1864,
       "step": 116
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.96e-06,
-      "loss": 4.1748,
+      "epoch": 0.07,
+      "learning_rate": 2.26e-05,
+      "loss": 4.3539,
       "step": 117
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 4.2224,
+      "epoch": 0.07,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.1572,
       "step": 118
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.08e-06,
-      "loss": 4.2624,
+      "epoch": 0.07,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.2879,
       "step": 119
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.14e-06,
-      "loss": 4.2724,
+      "epoch": 0.07,
+      "learning_rate": 2.32e-05,
+      "loss": 4.3617,
       "step": 120
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.2e-06,
-      "loss": 4.2884,
+      "epoch": 0.07,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.3742,
       "step": 121
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.26e-06,
-      "loss": 4.3434,
+      "epoch": 0.07,
+      "learning_rate": 2.36e-05,
+      "loss": 4.2665,
       "step": 122
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.32e-06,
-      "loss": 4.1558,
+      "epoch": 0.07,
+      "learning_rate": 2.38e-05,
+      "loss": 4.3543,
       "step": 123
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 4.2972,
+      "epoch": 0.07,
+      "learning_rate": 2.4e-05,
+      "loss": 4.2733,
       "step": 124
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.44e-06,
-      "loss": 4.3484,
+      "epoch": 0.07,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 4.0012,
       "step": 125
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.5e-06,
-      "loss": 4.2669,
+      "epoch": 0.07,
+      "learning_rate": 2.44e-05,
+      "loss": 4.5502,
       "step": 126
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 4.3126,
+      "epoch": 0.07,
+      "learning_rate": 2.46e-05,
+      "loss": 4.3129,
       "step": 127
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.62e-06,
-      "loss": 4.168,
+      "epoch": 0.07,
+      "learning_rate": 2.48e-05,
+      "loss": 4.2663,
       "step": 128
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.680000000000001e-06,
-      "loss": 4.3161,
+      "epoch": 0.07,
+      "learning_rate": 2.5e-05,
+      "loss": 4.1351,
       "step": 129
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.74e-06,
-      "loss": 4.4017,
+      "epoch": 0.07,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 4.2632,
       "step": 130
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.8e-06,
-      "loss": 4.3215,
+      "epoch": 0.07,
+      "learning_rate": 2.54e-05,
+      "loss": 4.4973,
       "step": 131
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 4.1507,
+      "epoch": 0.07,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 4.238,
       "step": 132
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.92e-06,
-      "loss": 4.3154,
+      "epoch": 0.07,
+      "learning_rate": 2.58e-05,
+      "loss": 4.2848,
       "step": 133
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.98e-06,
-      "loss": 4.1907,
+      "epoch": 0.08,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.283,
       "step": 134
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 4.3897,
+      "epoch": 0.08,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 4.1799,
       "step": 135
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.1e-06,
-      "loss": 4.2992,
+      "epoch": 0.08,
+      "learning_rate": 2.64e-05,
+      "loss": 4.3688,
       "step": 136
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.160000000000001e-06,
-      "loss": 4.2309,
+      "epoch": 0.08,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 4.1213,
       "step": 137
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.220000000000001e-06,
-      "loss": 4.1562,
+      "epoch": 0.08,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 4.3644,
       "step": 138
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.28e-06,
-      "loss": 4.1999,
+      "epoch": 0.08,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.4876,
       "step": 139
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.340000000000001e-06,
-      "loss": 4.2273,
+      "epoch": 0.08,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 4.1429,
       "step": 140
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.4316,
+      "epoch": 0.08,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 4.3436,
       "step": 141
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.459999999999999e-06,
-      "loss": 4.197,
+      "epoch": 0.08,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.5859,
       "step": 142
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.52e-06,
-      "loss": 4.3164,
+      "epoch": 0.08,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 4.2765,
       "step": 143
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.58e-06,
-      "loss": 4.3355,
+      "epoch": 0.08,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.3337,
       "step": 144
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.64e-06,
-      "loss": 4.4157,
+      "epoch": 0.08,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.678,
       "step": 145
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.7e-06,
-      "loss": 4.2614,
+      "epoch": 0.08,
+      "learning_rate": 2.84e-05,
+      "loss": 4.4569,
       "step": 146
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.759999999999999e-06,
-      "loss": 4.3818,
+      "epoch": 0.08,
+      "learning_rate": 2.86e-05,
+      "loss": 4.4205,
       "step": 147
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 8.82e-06,
-      "loss": 4.5368,
+      "epoch": 0.08,
+      "learning_rate": 2.88e-05,
+      "loss": 4.5305,
       "step": 148
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 8.88e-06,
-      "loss": 4.534,
+      "epoch": 0.08,
+      "learning_rate": 2.9e-05,
+      "loss": 4.7369,
       "step": 149
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 8.939999999999999e-06,
-      "loss": 4.642,
+      "epoch": 0.08,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 4.0898,
       "step": 150
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 9e-06,
-      "loss": 4.26,
+      "epoch": 0.08,
+      "learning_rate": 2.94e-05,
+      "loss": 4.0178,
       "step": 151
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 9.06e-06,
-      "loss": 4.2318,
+      "epoch": 0.09,
+      "learning_rate": 2.96e-05,
+      "loss": 4.3968,
       "step": 152
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 9.12e-06,
-      "loss": 4.2651,
+      "epoch": 0.09,
+      "learning_rate": 2.98e-05,
+      "loss": 3.9944,
       "step": 153
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.18e-06,
-      "loss": 4.1934,
+      "epoch": 0.09,
+      "learning_rate": 3e-05,
+      "loss": 4.4267,
       "step": 154
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.24e-06,
-      "loss": 4.2266,
+      "epoch": 0.09,
+      "learning_rate": 3.02e-05,
+      "loss": 4.3041,
       "step": 155
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.3e-06,
-      "loss": 4.2535,
+      "epoch": 0.09,
+      "learning_rate": 3.04e-05,
+      "loss": 4.5498,
       "step": 156
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.36e-06,
-      "loss": 4.1637,
+      "epoch": 0.09,
+      "learning_rate": 3.06e-05,
+      "loss": 4.3296,
       "step": 157
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.42e-06,
-      "loss": 4.2322,
+      "epoch": 0.09,
+      "learning_rate": 3.08e-05,
+      "loss": 4.3957,
       "step": 158
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.48e-06,
-      "loss": 4.2244,
+      "epoch": 0.09,
+      "learning_rate": 3.1e-05,
+      "loss": 4.3101,
       "step": 159
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.54e-06,
-      "loss": 4.2042,
+      "epoch": 0.09,
+      "learning_rate": 3.12e-05,
+      "loss": 4.301,
       "step": 160
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2152,
+      "epoch": 0.09,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 4.3339,
       "step": 161
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.66e-06,
-      "loss": 4.1874,
+      "epoch": 0.09,
+      "learning_rate": 3.16e-05,
+      "loss": 4.4358,
       "step": 162
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.72e-06,
-      "loss": 4.2569,
+      "epoch": 0.09,
+      "learning_rate": 3.18e-05,
+      "loss": 4.2733,
       "step": 163
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 4.1769,
+      "epoch": 0.09,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.4713,
       "step": 164
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.84e-06,
-      "loss": 4.2435,
+      "epoch": 0.09,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 4.2859,
       "step": 165
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.9e-06,
-      "loss": 4.206,
+      "epoch": 0.09,
+      "learning_rate": 3.24e-05,
+      "loss": 4.4804,
       "step": 166
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 4.223,
+      "epoch": 0.09,
+      "learning_rate": 3.26e-05,
+      "loss": 4.2683,
       "step": 167
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.002e-05,
-      "loss": 4.2185,
+      "epoch": 0.09,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 4.0396,
       "step": 168
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.008e-05,
-      "loss": 4.1909,
+      "epoch": 0.09,
+      "learning_rate": 3.3e-05,
+      "loss": 4.1296,
       "step": 169
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.0140000000000001e-05,
-      "loss": 4.1373,
+      "epoch": 0.1,
+      "learning_rate": 3.32e-05,
+      "loss": 4.2608,
       "step": 170
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.02e-05,
-      "loss": 4.1232,
+      "epoch": 0.1,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 4.3222,
       "step": 171
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.0260000000000002e-05,
-      "loss": 4.2279,
+      "epoch": 0.1,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 4.3247,
       "step": 172
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.032e-05,
-      "loss": 4.0904,
+      "epoch": 0.1,
+      "learning_rate": 3.38e-05,
+      "loss": 4.2829,
       "step": 173
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.0379999999999999e-05,
-      "loss": 4.1484,
+      "epoch": 0.1,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.2406,
       "step": 174
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.044e-05,
-      "loss": 4.296,
+      "epoch": 0.1,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 4.4264,
       "step": 175
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.05e-05,
-      "loss": 4.1508,
+      "epoch": 0.1,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 4.159,
       "step": 176
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.0559999999999999e-05,
-      "loss": 4.2833,
+      "epoch": 0.1,
+      "learning_rate": 3.46e-05,
+      "loss": 4.1747,
       "step": 177
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.062e-05,
-      "loss": 4.1359,
+      "epoch": 0.1,
+      "learning_rate": 3.48e-05,
+      "loss": 4.1733,
       "step": 178
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.068e-05,
-      "loss": 4.1837,
+      "epoch": 0.1,
+      "learning_rate": 3.5e-05,
+      "loss": 3.989,
       "step": 179
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.074e-05,
-      "loss": 4.3284,
+      "epoch": 0.1,
+      "learning_rate": 3.52e-05,
+      "loss": 4.4171,
       "step": 180
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.08e-05,
-      "loss": 4.196,
+      "epoch": 0.1,
+      "learning_rate": 3.54e-05,
+      "loss": 4.3069,
       "step": 181
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.086e-05,
-      "loss": 4.1404,
+      "epoch": 0.1,
+      "learning_rate": 3.56e-05,
+      "loss": 4.2119,
       "step": 182
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.092e-05,
-      "loss": 4.1248,
+      "epoch": 0.1,
+      "learning_rate": 3.58e-05,
+      "loss": 4.4286,
       "step": 183
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.098e-05,
-      "loss": 4.2255,
+      "epoch": 0.1,
+      "learning_rate": 3.6e-05,
+      "loss": 4.3764,
       "step": 184
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.104e-05,
-      "loss": 4.2625,
+      "epoch": 0.1,
+      "learning_rate": 3.62e-05,
+      "loss": 4.3944,
       "step": 185
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.11e-05,
-      "loss": 4.2148,
+      "epoch": 0.1,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 4.2301,
       "step": 186
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.116e-05,
-      "loss": 4.2413,
+      "epoch": 0.1,
+      "learning_rate": 3.66e-05,
+      "loss": 4.2749,
       "step": 187
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.1220000000000001e-05,
-      "loss": 4.2147,
+      "epoch": 0.11,
+      "learning_rate": 3.68e-05,
+      "loss": 4.4429,
       "step": 188
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.128e-05,
-      "loss": 4.1846,
+      "epoch": 0.11,
+      "learning_rate": 3.7e-05,
+      "loss": 4.1818,
       "step": 189
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.134e-05,
-      "loss": 4.2799,
+      "epoch": 0.11,
+      "learning_rate": 3.72e-05,
+      "loss": 4.217,
       "step": 190
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.3126,
+      "epoch": 0.11,
+      "learning_rate": 3.74e-05,
+      "loss": 4.5048,
       "step": 191
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.146e-05,
-      "loss": 4.3646,
+      "epoch": 0.11,
+      "learning_rate": 3.76e-05,
+      "loss": 4.2005,
       "step": 192
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.152e-05,
-      "loss": 4.4346,
+      "epoch": 0.11,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 4.1811,
       "step": 193
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.1580000000000001e-05,
-      "loss": 4.3561,
+      "epoch": 0.11,
+      "learning_rate": 3.8e-05,
+      "loss": 4.2651,
       "step": 194
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.164e-05,
-      "loss": 4.3478,
+      "epoch": 0.11,
+      "learning_rate": 3.82e-05,
+      "loss": 4.5532,
       "step": 195
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.1700000000000001e-05,
-      "loss": 4.3004,
+      "epoch": 0.11,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 4.3434,
       "step": 196
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.1760000000000001e-05,
-      "loss": 4.1971,
+      "epoch": 0.11,
+      "learning_rate": 3.86e-05,
+      "loss": 4.3847,
       "step": 197
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.182e-05,
-      "loss": 4.6767,
+      "epoch": 0.11,
+      "learning_rate": 3.88e-05,
+      "loss": 4.3204,
       "step": 198
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.1880000000000001e-05,
-      "loss": 4.4133,
+      "epoch": 0.11,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 4.4078,
       "step": 199
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.1940000000000001e-05,
-      "loss": 4.6429,
+      "epoch": 0.11,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 4.2925,
       "step": 200
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.2e-05,
-      "loss": 4.2574,
+      "epoch": 0.11,
+      "learning_rate": 3.94e-05,
+      "loss": 4.3417,
       "step": 201
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.2060000000000001e-05,
-      "loss": 4.1977,
+      "epoch": 0.11,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 4.2594,
       "step": 202
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.2120000000000001e-05,
-      "loss": 4.1876,
+      "epoch": 0.11,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 4.3878,
       "step": 203
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.2180000000000002e-05,
-      "loss": 4.1156,
+      "epoch": 0.11,
+      "learning_rate": 4e-05,
+      "loss": 4.2064,
       "step": 204
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.224e-05,
-      "loss": 4.1713,
+      "epoch": 0.11,
+      "learning_rate": 4.02e-05,
+      "loss": 4.231,
       "step": 205
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.2299999999999999e-05,
-      "loss": 4.1696,
+      "epoch": 0.12,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 4.3418,
       "step": 206
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.236e-05,
-      "loss": 4.2188,
+      "epoch": 0.12,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 4.3995,
       "step": 207
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.242e-05,
-      "loss": 4.1895,
+      "epoch": 0.12,
+      "learning_rate": 4.08e-05,
+      "loss": 4.1203,
       "step": 208
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.2479999999999999e-05,
-      "loss": 4.1555,
+      "epoch": 0.12,
+      "learning_rate": 4.1e-05,
+      "loss": 4.119,
       "step": 209
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.254e-05,
-      "loss": 4.2614,
+      "epoch": 0.12,
+      "learning_rate": 4.12e-05,
+      "loss": 4.3351,
       "step": 210
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.26e-05,
-      "loss": 4.0539,
+      "epoch": 0.12,
+      "learning_rate": 4.14e-05,
+      "loss": 4.2445,
       "step": 211
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.2659999999999999e-05,
-      "loss": 4.1863,
+      "epoch": 0.12,
+      "learning_rate": 4.16e-05,
+      "loss": 4.2771,
       "step": 212
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.272e-05,
-      "loss": 4.2161,
+      "epoch": 0.12,
+      "learning_rate": 4.18e-05,
+      "loss": 4.2772,
       "step": 213
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.278e-05,
-      "loss": 4.2059,
+      "epoch": 0.12,
+      "learning_rate": 4.2e-05,
+      "loss": 4.1123,
       "step": 214
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.284e-05,
-      "loss": 4.2133,
+      "epoch": 0.12,
+      "learning_rate": 4.22e-05,
+      "loss": 4.5046,
       "step": 215
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.29e-05,
-      "loss": 4.1991,
+      "epoch": 0.12,
+      "learning_rate": 4.24e-05,
+      "loss": 3.8912,
       "step": 216
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.296e-05,
-      "loss": 4.1563,
+      "epoch": 0.12,
+      "learning_rate": 4.26e-05,
+      "loss": 4.1509,
       "step": 217
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.302e-05,
-      "loss": 4.1303,
+      "epoch": 0.12,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 4.2466,
       "step": 218
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.308e-05,
-      "loss": 4.1368,
+      "epoch": 0.12,
+      "learning_rate": 4.3e-05,
+      "loss": 4.2244,
       "step": 219
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.314e-05,
-      "loss": 4.1971,
+      "epoch": 0.12,
+      "learning_rate": 4.32e-05,
+      "loss": 4.5083,
       "step": 220
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.32e-05,
-      "loss": 4.1136,
+      "epoch": 0.12,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 4.2867,
       "step": 221
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.326e-05,
-      "loss": 4.1069,
+      "epoch": 0.12,
+      "learning_rate": 4.36e-05,
+      "loss": 4.409,
       "step": 222
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.3320000000000001e-05,
-      "loss": 4.1314,
+      "epoch": 0.12,
+      "learning_rate": 4.38e-05,
+      "loss": 4.1443,
       "step": 223
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.338e-05,
-      "loss": 4.1782,
+      "epoch": 0.13,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 4.291,
       "step": 224
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.344e-05,
-      "loss": 4.1584,
+      "epoch": 0.13,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 4.3287,
       "step": 225
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 4.1714,
+      "epoch": 0.13,
+      "learning_rate": 4.44e-05,
+      "loss": 4.5251,
       "step": 226
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.356e-05,
-      "loss": 4.1961,
+      "epoch": 0.13,
+      "learning_rate": 4.46e-05,
+      "loss": 4.3725,
       "step": 227
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.362e-05,
-      "loss": 4.1251,
+      "epoch": 0.13,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 4.0831,
       "step": 228
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.3680000000000001e-05,
-      "loss": 4.1515,
+      "epoch": 0.13,
+      "learning_rate": 4.5e-05,
+      "loss": 4.2753,
       "step": 229
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.374e-05,
-      "loss": 4.1599,
+      "epoch": 0.13,
+      "learning_rate": 4.52e-05,
+      "loss": 4.288,
       "step": 230
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.2587,
+      "epoch": 0.13,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 4.3239,
       "step": 231
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3860000000000001e-05,
-      "loss": 4.1317,
+      "epoch": 0.13,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 4.2988,
       "step": 232
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.392e-05,
-      "loss": 4.2434,
+      "epoch": 0.13,
+      "learning_rate": 4.58e-05,
+      "loss": 4.4565,
       "step": 233
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3980000000000002e-05,
-      "loss": 4.2162,
+      "epoch": 0.13,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.3211,
       "step": 234
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.4040000000000001e-05,
-      "loss": 4.2583,
+      "epoch": 0.13,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 4.1615,
       "step": 235
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.4099999999999999e-05,
-      "loss": 4.0917,
+      "epoch": 0.13,
+      "learning_rate": 4.64e-05,
+      "loss": 4.753,
       "step": 236
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.416e-05,
-      "loss": 4.1941,
+      "epoch": 0.13,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 4.3354,
       "step": 237
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.422e-05,
-      "loss": 4.3278,
+      "epoch": 0.13,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 4.4139,
       "step": 238
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.428e-05,
-      "loss": 4.1576,
+      "epoch": 0.13,
+      "learning_rate": 4.7e-05,
+      "loss": 4.3243,
       "step": 239
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.434e-05,
-      "loss": 4.2114,
+      "epoch": 0.13,
+      "learning_rate": 4.72e-05,
+      "loss": 4.3752,
       "step": 240
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.44e-05,
-      "loss": 4.3157,
+      "epoch": 0.14,
+      "learning_rate": 4.74e-05,
+      "loss": 4.4773,
       "step": 241
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.446e-05,
-      "loss": 4.2419,
+      "epoch": 0.14,
+      "learning_rate": 4.76e-05,
+      "loss": 4.2556,
       "step": 242
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.452e-05,
-      "loss": 4.1953,
+      "epoch": 0.14,
+      "learning_rate": 4.78e-05,
+      "loss": 4.3593,
       "step": 243
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.458e-05,
-      "loss": 4.2287,
+      "epoch": 0.14,
+      "learning_rate": 4.8e-05,
+      "loss": 4.6138,
       "step": 244
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.464e-05,
-      "loss": 4.2023,
+      "epoch": 0.14,
+      "learning_rate": 4.82e-05,
+      "loss": 4.5959,
       "step": 245
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.47e-05,
-      "loss": 4.3791,
+      "epoch": 0.14,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 4.5063,
       "step": 246
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.4760000000000001e-05,
-      "loss": 4.3207,
+      "epoch": 0.14,
+      "learning_rate": 4.86e-05,
+      "loss": 4.2961,
       "step": 247
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.482e-05,
-      "loss": 4.323,
+      "epoch": 0.14,
+      "learning_rate": 4.88e-05,
+      "loss": 4.4591,
       "step": 248
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.488e-05,
-      "loss": 4.4303,
+      "epoch": 0.14,
+      "learning_rate": 4.9e-05,
+      "loss": 4.342,
       "step": 249
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.4940000000000001e-05,
-      "loss": 4.4593,
+      "epoch": 0.14,
+      "learning_rate": 4.92e-05,
+      "loss": 4.6868,
       "step": 250
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.5e-05,
-      "loss": 4.2375,
+      "epoch": 0.14,
+      "learning_rate": 4.94e-05,
+      "loss": 4.2804,
       "step": 251
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.506e-05,
-      "loss": 4.1519,
+      "epoch": 0.14,
+      "learning_rate": 4.96e-05,
+      "loss": 4.6598,
       "step": 252
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.5120000000000001e-05,
-      "loss": 4.115,
+      "epoch": 0.14,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 4.286,
       "step": 253
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.518e-05,
-      "loss": 4.1146,
+      "epoch": 0.14,
+      "learning_rate": 5e-05,
+      "loss": 4.4268,
       "step": 254
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.524e-05,
-      "loss": 4.1397,
+      "epoch": 0.14,
+      "learning_rate": 5.02e-05,
+      "loss": 4.3185,
       "step": 255
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.53e-05,
-      "loss": 4.1209,
+      "epoch": 0.14,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 4.2675,
       "step": 256
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.5360000000000002e-05,
-      "loss": 4.2665,
+      "epoch": 0.14,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 4.272,
       "step": 257
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.542e-05,
-      "loss": 4.1351,
+      "epoch": 0.14,
+      "learning_rate": 5.08e-05,
+      "loss": 4.4397,
       "step": 258
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.548e-05,
-      "loss": 4.1725,
+      "epoch": 0.15,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.5586,
       "step": 259
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.554e-05,
-      "loss": 4.0934,
+      "epoch": 0.15,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 3.9949,
       "step": 260
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.56e-05,
-      "loss": 4.1963,
+      "epoch": 0.15,
+      "learning_rate": 5.14e-05,
+      "loss": 4.2235,
       "step": 261
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.5660000000000003e-05,
-      "loss": 4.2262,
+      "epoch": 0.15,
+      "learning_rate": 5.16e-05,
+      "loss": 4.3789,
       "step": 262
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.5720000000000002e-05,
-      "loss": 4.0871,
+      "epoch": 0.15,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 4.4334,
       "step": 263
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.578e-05,
-      "loss": 4.1276,
+      "epoch": 0.15,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 4.1581,
       "step": 264
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.584e-05,
-      "loss": 4.137,
+      "epoch": 0.15,
+      "learning_rate": 5.22e-05,
+      "loss": 4.2716,
       "step": 265
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.59e-05,
-      "loss": 4.2361,
+      "epoch": 0.15,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 4.0698,
       "step": 266
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.596e-05,
-      "loss": 4.1436,
+      "epoch": 0.15,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 4.3449,
       "step": 267
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.6020000000000002e-05,
-      "loss": 4.1081,
+      "epoch": 0.15,
+      "learning_rate": 5.28e-05,
+      "loss": 4.4088,
       "step": 268
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.6080000000000002e-05,
-      "loss": 4.0701,
+      "epoch": 0.15,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 4.4824,
       "step": 269
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.614e-05,
-      "loss": 4.2026,
+      "epoch": 0.15,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 4.0367,
       "step": 270
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.62e-05,
-      "loss": 4.0701,
+      "epoch": 0.15,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 4.3449,
       "step": 271
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.626e-05,
-      "loss": 4.0502,
+      "epoch": 0.15,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 4.2108,
       "step": 272
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.6320000000000003e-05,
-      "loss": 4.1113,
+      "epoch": 0.15,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 4.0506,
       "step": 273
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.6380000000000002e-05,
-      "loss": 4.0909,
+      "epoch": 0.15,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 4.3988,
       "step": 274
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.6440000000000002e-05,
-      "loss": 4.2243,
+      "epoch": 0.15,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 4.3266,
       "step": 275
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.65e-05,
-      "loss": 4.1479,
+      "epoch": 0.15,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 4.4264,
       "step": 276
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.656e-05,
-      "loss": 4.1566,
+      "epoch": 0.16,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 4.2787,
       "step": 277
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.6620000000000004e-05,
-      "loss": 4.0795,
+      "epoch": 0.16,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 4.4066,
       "step": 278
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.6680000000000003e-05,
-      "loss": 4.1255,
+      "epoch": 0.16,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.1534,
       "step": 279
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.6740000000000002e-05,
-      "loss": 4.0884,
+      "epoch": 0.16,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 4.2387,
       "step": 280
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.1399,
+      "epoch": 0.16,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 4.5592,
       "step": 281
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.686e-05,
-      "loss": 4.1306,
+      "epoch": 0.16,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 4.5102,
       "step": 282
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.6919999999999997e-05,
-      "loss": 4.2975,
+      "epoch": 0.16,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 4.4963,
       "step": 283
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.698e-05,
-      "loss": 4.0595,
+      "epoch": 0.16,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 4.1404,
       "step": 284
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.704e-05,
-      "loss": 4.1804,
+      "epoch": 0.16,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 4.3225,
       "step": 285
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.71e-05,
-      "loss": 4.1973,
+      "epoch": 0.16,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.2083,
       "step": 286
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.716e-05,
-      "loss": 4.1023,
+      "epoch": 0.16,
+      "learning_rate": 5.66e-05,
+      "loss": 4.4062,
       "step": 287
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.7219999999999998e-05,
-      "loss": 4.1242,
+      "epoch": 0.16,
+      "learning_rate": 5.68e-05,
+      "loss": 4.5078,
       "step": 288
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.728e-05,
-      "loss": 4.1294,
+      "epoch": 0.16,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.2776,
       "step": 289
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.734e-05,
-      "loss": 4.2188,
+      "epoch": 0.16,
+      "learning_rate": 5.72e-05,
+      "loss": 4.4046,
       "step": 290
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.74e-05,
-      "loss": 4.2035,
+      "epoch": 0.16,
+      "learning_rate": 5.74e-05,
+      "loss": 4.3294,
       "step": 291
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.746e-05,
-      "loss": 4.2955,
+      "epoch": 0.16,
+      "learning_rate": 5.76e-05,
+      "loss": 4.7494,
       "step": 292
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.7519999999999998e-05,
-      "loss": 4.262,
+      "epoch": 0.16,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 4.5344,
       "step": 293
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.758e-05,
-      "loss": 4.4573,
+      "epoch": 0.16,
+      "learning_rate": 5.8e-05,
+      "loss": 4.5209,
       "step": 294
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.764e-05,
-      "loss": 4.227,
+      "epoch": 0.17,
+      "learning_rate": 5.82e-05,
+      "loss": 4.3708,
       "step": 295
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.77e-05,
-      "loss": 4.5969,
+      "epoch": 0.17,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 5.1234,
       "step": 296
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.776e-05,
-      "loss": 4.4712,
+      "epoch": 0.17,
+      "learning_rate": 5.86e-05,
+      "loss": 4.3939,
       "step": 297
     },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.88e-05,
+      "loss": 4.5994,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9e-05,
+      "loss": 4.4041,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.92e-05,
+      "loss": 4.6423,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.94e-05,
+      "loss": 4.4683,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.96e-05,
+      "loss": 4.3763,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 4.4272,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6e-05,
+      "loss": 4.2001,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.02e-05,
+      "loss": 4.755,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.04e-05,
+      "loss": 4.2649,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.06e-05,
+      "loss": 4.4684,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.08e-05,
+      "loss": 5.1254,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.1e-05,
+      "loss": 4.4714,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.12e-05,
+      "loss": 4.5853,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.14e-05,
+      "loss": 4.4693,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.16e-05,
+      "loss": 4.5472,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.18e-05,
+      "loss": 4.4195,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.2e-05,
+      "loss": 4.5668,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 4.4117,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.24e-05,
+      "loss": 4.2694,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.26e-05,
+      "loss": 4.2984,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 4.1355,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.3e-05,
+      "loss": 4.3497,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.32e-05,
+      "loss": 4.6001,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 4.6639,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.36e-05,
+      "loss": 4.7958,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.38e-05,
+      "loss": 4.5806,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 4.2765,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.42e-05,
+      "loss": 4.4485,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 4.574,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 4.3824,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.48e-05,
+      "loss": 4.5099,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.2511,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.52e-05,
+      "loss": 4.5127,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.54e-05,
+      "loss": 4.4232,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 4.4945,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.58e-05,
+      "loss": 4.8722,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.6e-05,
+      "loss": 4.5344,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 4.6432,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.64e-05,
+      "loss": 4.4751,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.66e-05,
+      "loss": 4.3603,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 4.1594,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.7e-05,
+      "loss": 4.2968,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 4.4818,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 4.3059,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.76e-05,
+      "loss": 4.4054,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 4.7326,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.1529,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.82e-05,
+      "loss": 4.8932,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 4.2191,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 4.7467,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 4.7302,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.9e-05,
+      "loss": 4.6994,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.92e-05,
+      "loss": 4.3508,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 4.4927,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.96e-05,
+      "loss": 4.6246,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.98e-05,
+      "loss": 4.502,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7e-05,
+      "loss": 4.2909,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.02e-05,
+      "loss": 4.0331,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.04e-05,
+      "loss": 4.4068,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.06e-05,
+      "loss": 4.566,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.08e-05,
+      "loss": 4.4114,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.1e-05,
+      "loss": 4.8375,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.12e-05,
+      "loss": 4.2689,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.14e-05,
+      "loss": 4.2941,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.16e-05,
+      "loss": 4.1669,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.18e-05,
+      "loss": 4.4077,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.2e-05,
+      "loss": 4.6177,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.22e-05,
+      "loss": 5.0609,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.24e-05,
+      "loss": 4.7905,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.26e-05,
+      "loss": 4.3293,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 4.2958,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.3e-05,
+      "loss": 4.3222,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.32e-05,
+      "loss": 4.5063,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 4.2663,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.36e-05,
+      "loss": 4.7706,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.38e-05,
+      "loss": 4.3834,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4e-05,
+      "loss": 4.4925,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.42e-05,
+      "loss": 4.0847,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.44e-05,
+      "loss": 4.2565,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.46e-05,
+      "loss": 4.2806,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.48e-05,
+      "loss": 4.1096,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.3248,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.52e-05,
+      "loss": 4.6304,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.54e-05,
+      "loss": 4.0771,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 4.4156,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.58e-05,
+      "loss": 4.3534,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.6e-05,
+      "loss": 4.4788,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 4.4498,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.64e-05,
+      "loss": 4.4509,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.66e-05,
+      "loss": 4.9955,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 4.1628,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.7e-05,
+      "loss": 4.7865,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.72e-05,
+      "loss": 5.5212,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 4.1773,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.76e-05,
+      "loss": 4.7625,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 4.6582,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 4.5779,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.82e-05,
+      "loss": 4.6164,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 5.2347,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 4.489,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.88e-05,
+      "loss": 5.0273,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 8.0616,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 4.9529,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.94e-05,
+      "loss": 4.5552,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 4.3541,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.98e-05,
+      "loss": 4.3047,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8e-05,
+      "loss": 4.4997,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 4.2333,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.04e-05,
+      "loss": 4.4826,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 4.3035,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 4.4925,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.1e-05,
+      "loss": 4.697,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 4.379,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.14e-05,
+      "loss": 4.4468,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.16e-05,
+      "loss": 4.698,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.18e-05,
+      "loss": 4.5262,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.2e-05,
+      "loss": 4.4332,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.22e-05,
+      "loss": 4.5571,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.24e-05,
+      "loss": 4.3238,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.26e-05,
+      "loss": 4.7802,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.28e-05,
+      "loss": 4.5399,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.3e-05,
+      "loss": 4.405,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.32e-05,
+      "loss": 4.5178,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.34e-05,
+      "loss": 4.2381,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.36e-05,
+      "loss": 4.546,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.38e-05,
+      "loss": 4.2236,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.4e-05,
+      "loss": 4.7548,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.42e-05,
+      "loss": 4.5744,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.44e-05,
+      "loss": 4.4304,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.46e-05,
+      "loss": 5.0038,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.48e-05,
+      "loss": 4.8463,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.5e-05,
+      "loss": 4.3219,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.52e-05,
+      "loss": 4.5205,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.54e-05,
+      "loss": 4.6564,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 4.4411,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.58e-05,
+      "loss": 4.4333,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6e-05,
+      "loss": 4.4752,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 4.4606,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.64e-05,
+      "loss": 4.1038,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.66e-05,
+      "loss": 4.3064,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 4.5611,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7e-05,
+      "loss": 4.2406,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.72e-05,
+      "loss": 4.4713,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 4.8589,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76e-05,
+      "loss": 4.6789,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78e-05,
+      "loss": 4.4164,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 4.7035,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82e-05,
+      "loss": 4.4509,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 5.7712,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.86e-05,
+      "loss": 5.3732,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.88e-05,
+      "loss": 5.1341,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 5.6573,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.92e-05,
+      "loss": 5.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.94e-05,
+      "loss": 4.1658,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 4.4291,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.98e-05,
+      "loss": 4.7395,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9e-05,
+      "loss": 4.1336,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 4.2356,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.04e-05,
+      "loss": 4.337,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.06e-05,
+      "loss": 4.3844,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 4.287,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.1e-05,
+      "loss": 4.1492,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 4.5343,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 4.6228,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.16e-05,
+      "loss": 4.1707,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 4.577,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 4.2326,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.22e-05,
+      "loss": 4.1823,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 4.6568,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 4.4755,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.28e-05,
+      "loss": 4.519,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 4.5803,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 4.4706,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 4.6845,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 4.2703,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.38e-05,
+      "loss": 4.313,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.4e-05,
+      "loss": 4.7141,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.42e-05,
+      "loss": 4.3869,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.44e-05,
+      "loss": 4.5496,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.46e-05,
+      "loss": 4.7068,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.48e-05,
+      "loss": 4.407,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.5e-05,
+      "loss": 4.599,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.52e-05,
+      "loss": 4.5897,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.54e-05,
+      "loss": 4.169,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.56e-05,
+      "loss": 4.8109,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.58e-05,
+      "loss": 4.7983,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.6e-05,
+      "loss": 4.301,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 4.8136,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.64e-05,
+      "loss": 5.1314,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.66e-05,
+      "loss": 4.4838,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 4.593,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.7e-05,
+      "loss": 4.7619,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.72e-05,
+      "loss": 4.4978,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.74e-05,
+      "loss": 4.7094,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.76e-05,
+      "loss": 5.1335,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.78e-05,
+      "loss": 4.5353,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.8e-05,
+      "loss": 5.1835,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.82e-05,
+      "loss": 5.0476,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.84e-05,
+      "loss": 4.587,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.86e-05,
+      "loss": 4.2607,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.88e-05,
+      "loss": 5.3366,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 4.8184,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.92e-05,
+      "loss": 4.3147,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 4.7075324058532715,
+      "eval_runtime": 947.3,
+      "eval_samples_per_second": 2.789,
+      "eval_steps_per_second": 0.349,
+      "eval_wer": 1.9290162633875447,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.94e-05,
+      "loss": 5.3015,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 4.2163,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.98e-05,
+      "loss": 4.5817,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001,
+      "loss": 4.7673,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.992211838006231e-05,
+      "loss": 4.583,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.984423676012462e-05,
+      "loss": 4.5462,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.976635514018692e-05,
+      "loss": 4.4602,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.968847352024923e-05,
+      "loss": 4.5587,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.961059190031154e-05,
+      "loss": 4.8726,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.953271028037385e-05,
+      "loss": 5.0269,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.945482866043614e-05,
+      "loss": 4.5041,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.937694704049845e-05,
+      "loss": 4.5437,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.929906542056076e-05,
+      "loss": 4.697,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.922118380062307e-05,
+      "loss": 4.193,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.914330218068536e-05,
+      "loss": 4.2665,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.906542056074767e-05,
+      "loss": 5.52,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.898753894080998e-05,
+      "loss": 4.8,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.890965732087229e-05,
+      "loss": 4.9779,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.883177570093459e-05,
+      "loss": 4.8114,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.87538940809969e-05,
+      "loss": 4.4001,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.867601246105919e-05,
+      "loss": 4.1522,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.85981308411215e-05,
+      "loss": 4.578,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.852024922118381e-05,
+      "loss": 4.6367,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.84423676012461e-05,
+      "loss": 4.7124,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.836448598130841e-05,
+      "loss": 5.6669,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.828660436137072e-05,
+      "loss": 4.6662,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.820872274143302e-05,
+      "loss": 4.8018,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.813084112149533e-05,
+      "loss": 4.1469,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.805295950155764e-05,
+      "loss": 4.5179,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.797507788161995e-05,
+      "loss": 4.9164,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.789719626168224e-05,
+      "loss": 4.2611,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.781931464174455e-05,
+      "loss": 4.1499,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.774143302180686e-05,
+      "loss": 4.4709,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.766355140186917e-05,
+      "loss": 4.2037,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.758566978193146e-05,
+      "loss": 4.8989,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.750778816199377e-05,
+      "loss": 4.4561,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.742990654205608e-05,
+      "loss": 4.3328,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.735202492211839e-05,
+      "loss": 4.5576,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.727414330218069e-05,
+      "loss": 4.6243,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.7196261682243e-05,
+      "loss": 4.6593,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.71183800623053e-05,
+      "loss": 5.0028,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.704049844236761e-05,
+      "loss": 4.5634,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.696261682242991e-05,
+      "loss": 6.364,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.688473520249222e-05,
+      "loss": 4.4907,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.680685358255453e-05,
+      "loss": 4.3936,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.672897196261684e-05,
+      "loss": 5.4682,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.665109034267913e-05,
+      "loss": 5.1027,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.657320872274144e-05,
+      "loss": 6.0538,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.649532710280375e-05,
+      "loss": 5.389,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.641744548286606e-05,
+      "loss": 4.7322,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.633956386292835e-05,
+      "loss": 4.7192,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.626168224299066e-05,
+      "loss": 5.1824,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.618380062305297e-05,
+      "loss": 5.1587,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.610591900311527e-05,
+      "loss": 4.6674,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.602803738317758e-05,
+      "loss": 4.3485,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.595015576323989e-05,
+      "loss": 4.9679,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.587227414330218e-05,
+      "loss": 4.5833,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.579439252336449e-05,
+      "loss": 4.3588,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.571651090342679e-05,
+      "loss": 4.4552,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.56386292834891e-05,
+      "loss": 4.3219,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.55607476635514e-05,
+      "loss": 4.6793,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.548286604361371e-05,
+      "loss": 4.7544,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.540498442367601e-05,
+      "loss": 4.1464,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.532710280373832e-05,
+      "loss": 4.9414,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.524922118380063e-05,
+      "loss": 4.9265,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.517133956386294e-05,
+      "loss": 4.3055,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.509345794392523e-05,
+      "loss": 4.476,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.501557632398754e-05,
+      "loss": 4.8398,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.493769470404985e-05,
+      "loss": 4.4756,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.485981308411216e-05,
+      "loss": 4.6567,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.478193146417445e-05,
+      "loss": 4.2824,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.470404984423676e-05,
+      "loss": 4.6007,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.462616822429907e-05,
+      "loss": 5.1757,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.454828660436138e-05,
+      "loss": 4.4601,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.447040498442368e-05,
+      "loss": 4.2596,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.439252336448599e-05,
+      "loss": 4.4326,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.43146417445483e-05,
+      "loss": 4.5379,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.42367601246106e-05,
+      "loss": 4.4634,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.41588785046729e-05,
+      "loss": 4.6301,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.408099688473521e-05,
+      "loss": 5.2573,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.400311526479752e-05,
+      "loss": 4.3973,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.392523364485983e-05,
+      "loss": 4.7662,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.384735202492212e-05,
+      "loss": 4.42,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.376947040498443e-05,
+      "loss": 4.7101,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.369158878504674e-05,
+      "loss": 5.3389,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.361370716510905e-05,
+      "loss": 4.3448,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.353582554517135e-05,
+      "loss": 4.9993,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.345794392523365e-05,
+      "loss": 4.4442,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.338006230529596e-05,
+      "loss": 4.8559,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.330218068535826e-05,
+      "loss": 4.6335,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.322429906542057e-05,
+      "loss": 4.4581,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.314641744548286e-05,
+      "loss": 4.9841,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.306853582554517e-05,
+      "loss": 4.4821,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.299065420560748e-05,
+      "loss": 4.7013,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.291277258566978e-05,
+      "loss": 4.5624,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.283489096573209e-05,
+      "loss": 4.7267,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.27570093457944e-05,
+      "loss": 6.08,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.26791277258567e-05,
+      "loss": 6.4348,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.2601246105919e-05,
+      "loss": 4.9671,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.252336448598131e-05,
+      "loss": 4.987,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.244548286604362e-05,
+      "loss": 4.1211,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.236760124610593e-05,
+      "loss": 4.6718,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.228971962616822e-05,
+      "loss": 4.3752,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.221183800623053e-05,
+      "loss": 4.3506,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.213395638629284e-05,
+      "loss": 4.3198,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.205607476635515e-05,
+      "loss": 4.2261,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.197819314641744e-05,
+      "loss": 4.7796,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.190031152647975e-05,
+      "loss": 4.6005,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.182242990654206e-05,
+      "loss": 4.6845,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.174454828660437e-05,
+      "loss": 4.228,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 4.8007,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.158878504672898e-05,
+      "loss": 4.3532,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.151090342679129e-05,
+      "loss": 4.6751,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.14330218068536e-05,
+      "loss": 4.6625,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.135514018691589e-05,
+      "loss": 4.6042,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.12772585669782e-05,
+      "loss": 4.2013,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.119937694704051e-05,
+      "loss": 4.8334,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.112149532710282e-05,
+      "loss": 4.2135,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.104361370716511e-05,
+      "loss": 4.6628,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.096573208722742e-05,
+      "loss": 4.8568,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.088785046728973e-05,
+      "loss": 4.577,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.080996884735204e-05,
+      "loss": 4.5132,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.073208722741434e-05,
+      "loss": 4.5329,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.065420560747664e-05,
+      "loss": 5.0965,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.057632398753894e-05,
+      "loss": 4.5763,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.049844236760125e-05,
+      "loss": 4.4123,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.042056074766356e-05,
+      "loss": 4.2595,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.034267912772585e-05,
+      "loss": 5.0914,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.026479750778816e-05,
+      "loss": 5.3159,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.018691588785047e-05,
+      "loss": 4.3541,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.010903426791277e-05,
+      "loss": 4.4803,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.003115264797508e-05,
+      "loss": 4.7626,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.995327102803739e-05,
+      "loss": 4.5064,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.98753894080997e-05,
+      "loss": 4.3488,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.979750778816199e-05,
+      "loss": 4.4202,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.97196261682243e-05,
+      "loss": 4.1421,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.964174454828661e-05,
+      "loss": 4.4058,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.956386292834892e-05,
+      "loss": 4.1725,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.948598130841121e-05,
+      "loss": 4.5331,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.940809968847352e-05,
+      "loss": 5.6425,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.933021806853583e-05,
+      "loss": 4.8571,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.925233644859814e-05,
+      "loss": 4.5585,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.917445482866044e-05,
+      "loss": 4.7614,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.909657320872274e-05,
+      "loss": 4.5678,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.901869158878505e-05,
+      "loss": 4.4327,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.894080996884736e-05,
+      "loss": 4.7126,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.886292834890966e-05,
+      "loss": 5.3126,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.878504672897197e-05,
+      "loss": 6.2914,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.870716510903428e-05,
+      "loss": 4.7907,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.862928348909659e-05,
+      "loss": 4.8894,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.855140186915888e-05,
+      "loss": 4.3384,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.847352024922119e-05,
+      "loss": 4.3788,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.83956386292835e-05,
+      "loss": 4.38,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.831775700934581e-05,
+      "loss": 4.3244,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.82398753894081e-05,
+      "loss": 4.9414,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.816199376947041e-05,
+      "loss": 4.2382,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.808411214953272e-05,
+      "loss": 4.3228,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.800623052959502e-05,
+      "loss": 4.87,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.792834890965733e-05,
+      "loss": 4.6152,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.785046728971964e-05,
+      "loss": 4.4443,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.777258566978193e-05,
+      "loss": 4.8306,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.769470404984424e-05,
+      "loss": 4.584,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.761682242990654e-05,
+      "loss": 4.4139,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.753894080996884e-05,
+      "loss": 4.8834,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.746105919003115e-05,
+      "loss": 4.3759,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.738317757009346e-05,
+      "loss": 4.7428,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.730529595015576e-05,
+      "loss": 4.4282,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.722741433021807e-05,
+      "loss": 4.5489,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.714953271028038e-05,
+      "loss": 4.2051,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.707165109034269e-05,
+      "loss": 4.5607,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.699376947040498e-05,
+      "loss": 4.7091,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.691588785046729e-05,
+      "loss": 5.1176,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.68380062305296e-05,
+      "loss": 4.3949,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.676012461059191e-05,
+      "loss": 4.2356,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.66822429906542e-05,
+      "loss": 4.4917,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.660436137071651e-05,
+      "loss": 4.3925,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.652647975077882e-05,
+      "loss": 4.2797,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.644859813084113e-05,
+      "loss": 4.2677,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.637071651090343e-05,
+      "loss": 4.7455,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.629283489096574e-05,
+      "loss": 4.9401,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.621495327102804e-05,
+      "loss": 4.3894,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.613707165109035e-05,
+      "loss": 4.4105,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.605919003115265e-05,
+      "loss": 4.7056,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.598130841121496e-05,
+      "loss": 4.8374,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.590342679127727e-05,
+      "loss": 4.7006,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.582554517133958e-05,
+      "loss": 4.5107,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.574766355140187e-05,
+      "loss": 4.2182,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.566978193146418e-05,
+      "loss": 4.1902,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.559190031152649e-05,
+      "loss": 4.6584,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.55140186915888e-05,
+      "loss": 4.4036,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.54361370716511e-05,
+      "loss": 4.1882,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.53582554517134e-05,
+      "loss": 4.7875,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.528037383177571e-05,
+      "loss": 4.6571,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.520249221183801e-05,
+      "loss": 4.3997,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.512461059190032e-05,
+      "loss": 4.5902,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.504672897196261e-05,
+      "loss": 4.6124,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.496884735202492e-05,
+      "loss": 4.0131,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.489096573208723e-05,
+      "loss": 4.4198,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.481308411214953e-05,
+      "loss": 5.5593,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.473520249221184e-05,
+      "loss": 4.4242,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.465732087227414e-05,
+      "loss": 4.1628,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.457943925233645e-05,
+      "loss": 4.3002,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.450155763239875e-05,
+      "loss": 4.4323,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.442367601246106e-05,
+      "loss": 5.2875,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.434579439252337e-05,
+      "loss": 5.337,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.426791277258568e-05,
+      "loss": 4.8201,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.419003115264797e-05,
+      "loss": 4.734,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.411214953271028e-05,
+      "loss": 4.6463,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.403426791277259e-05,
+      "loss": 4.3115,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.39563862928349e-05,
+      "loss": 4.4657,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.38785046728972e-05,
+      "loss": 4.6514,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.38006230529595e-05,
+      "loss": 4.2752,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.372274143302181e-05,
+      "loss": 4.2493,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.364485981308412e-05,
+      "loss": 4.8864,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.356697819314642e-05,
+      "loss": 4.8047,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.348909657320873e-05,
+      "loss": 4.1735,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.341121495327104e-05,
+      "loss": 4.4169,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 4.6063,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.325545171339564e-05,
+      "loss": 4.3406,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.317757009345795e-05,
+      "loss": 4.5759,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.309968847352026e-05,
+      "loss": 4.7111,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.302180685358257e-05,
+      "loss": 4.4709,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.294392523364486e-05,
+      "loss": 4.3149,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.286604361370717e-05,
+      "loss": 4.8441,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.278816199376948e-05,
+      "loss": 4.4673,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.271028037383179e-05,
+      "loss": 4.7216,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.263239875389408e-05,
+      "loss": 4.0092,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.25545171339564e-05,
+      "loss": 4.7319,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.24766355140187e-05,
+      "loss": 4.3046,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.2398753894081e-05,
+      "loss": 4.4357,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.232087227414331e-05,
+      "loss": 4.3488,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.22429906542056e-05,
+      "loss": 4.2819,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.216510903426791e-05,
+      "loss": 5.4103,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.208722741433022e-05,
+      "loss": 4.3532,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.200934579439252e-05,
+      "loss": 4.5757,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.193146417445483e-05,
+      "loss": 4.3834,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.185358255451713e-05,
+      "loss": 4.238,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.177570093457944e-05,
+      "loss": 4.4355,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.169781931464174e-05,
+      "loss": 4.497,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.161993769470405e-05,
+      "loss": 4.3891,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.154205607476636e-05,
+      "loss": 4.9447,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.146417445482867e-05,
+      "loss": 4.244,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.138629283489096e-05,
+      "loss": 5.2218,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.130841121495327e-05,
+      "loss": 4.4696,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.123052959501558e-05,
+      "loss": 5.3885,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.115264797507789e-05,
+      "loss": 4.5004,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.107476635514018e-05,
+      "loss": 4.5987,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.09968847352025e-05,
+      "loss": 4.6396,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.09190031152648e-05,
+      "loss": 4.3363,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.084112149532711e-05,
+      "loss": 4.7165,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.076323987538941e-05,
+      "loss": 4.4349,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.068535825545172e-05,
+      "loss": 4.3899,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.060747663551403e-05,
+      "loss": 4.5635,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.052959501557633e-05,
+      "loss": 4.5619,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.045171339563863e-05,
+      "loss": 4.3701,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.037383177570094e-05,
+      "loss": 4.6714,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.029595015576325e-05,
+      "loss": 4.7692,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.021806853582556e-05,
+      "loss": 4.4807,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.014018691588785e-05,
+      "loss": 4.4829,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.006230529595016e-05,
+      "loss": 4.3783,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.998442367601247e-05,
+      "loss": 4.6417,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.990654205607478e-05,
+      "loss": 4.372,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.982866043613708e-05,
+      "loss": 4.3511,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.975077881619938e-05,
+      "loss": 4.177,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.967289719626168e-05,
+      "loss": 4.1005,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.959501557632399e-05,
+      "loss": 4.5744,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.95171339563863e-05,
+      "loss": 4.4858,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.94392523364486e-05,
+      "loss": 4.1914,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.93613707165109e-05,
+      "loss": 5.0006,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.928348909657321e-05,
+      "loss": 4.4004,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.920560747663551e-05,
+      "loss": 4.4175,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.912772585669782e-05,
+      "loss": 4.3401,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.904984423676013e-05,
+      "loss": 4.2514,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.897196261682243e-05,
+      "loss": 4.9949,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.889408099688473e-05,
+      "loss": 4.2924,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.881619937694704e-05,
+      "loss": 4.3603,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.873831775700935e-05,
+      "loss": 4.2146,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.866043613707166e-05,
+      "loss": 4.4843,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.858255451713395e-05,
+      "loss": 4.5589,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.850467289719626e-05,
+      "loss": 4.4346,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.842679127725857e-05,
+      "loss": 4.5059,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.834890965732088e-05,
+      "loss": 4.6551,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.827102803738318e-05,
+      "loss": 4.2883,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.819314641744548e-05,
+      "loss": 4.4217,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.81152647975078e-05,
+      "loss": 4.4488,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.80373831775701e-05,
+      "loss": 4.8291,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.79595015576324e-05,
+      "loss": 4.2201,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.788161993769471e-05,
+      "loss": 4.9014,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.780373831775702e-05,
+      "loss": 4.6765,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.772585669781933e-05,
+      "loss": 4.8622,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.764797507788162e-05,
+      "loss": 4.5598,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.757009345794393e-05,
+      "loss": 4.3493,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.749221183800624e-05,
+      "loss": 4.4674,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.741433021806855e-05,
+      "loss": 4.3265,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.733644859813084e-05,
+      "loss": 5.1845,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.725856697819315e-05,
+      "loss": 5.3613,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.718068535825546e-05,
+      "loss": 5.0543,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.710280373831776e-05,
+      "loss": 4.5626,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.702492211838007e-05,
+      "loss": 4.5347,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.694704049844238e-05,
+      "loss": 4.5914,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.686915887850467e-05,
+      "loss": 4.6287,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.679127725856698e-05,
+      "loss": 4.5135,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.671339563862928e-05,
+      "loss": 4.5834,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.663551401869158e-05,
+      "loss": 4.4724,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.65576323987539e-05,
+      "loss": 4.4331,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.64797507788162e-05,
+      "loss": 4.4964,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.64018691588785e-05,
+      "loss": 4.5353,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.632398753894081e-05,
+      "loss": 4.038,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.624610591900312e-05,
+      "loss": 4.5097,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.616822429906543e-05,
+      "loss": 4.9234,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.609034267912772e-05,
+      "loss": 4.5265,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.601246105919003e-05,
+      "loss": 4.8665,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.593457943925234e-05,
+      "loss": 4.5298,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.585669781931465e-05,
+      "loss": 4.1562,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.577881619937694e-05,
+      "loss": 4.3682,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.570093457943925e-05,
+      "loss": 4.4111,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.562305295950156e-05,
+      "loss": 4.6573,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.554517133956387e-05,
+      "loss": 4.483,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.546728971962617e-05,
+      "loss": 4.3528,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.538940809968848e-05,
+      "loss": 4.3377,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.531152647975078e-05,
+      "loss": 4.4368,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.52336448598131e-05,
+      "loss": 4.2774,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.515576323987539e-05,
+      "loss": 4.8133,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.50778816199377e-05,
+      "loss": 4.5117,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.6504,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.492211838006232e-05,
+      "loss": 4.3947,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.484423676012463e-05,
+      "loss": 4.8765,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.476635514018692e-05,
+      "loss": 4.2875,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.468847352024923e-05,
+      "loss": 4.6465,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.461059190031154e-05,
+      "loss": 4.3284,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.453271028037383e-05,
+      "loss": 4.5927,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.445482866043614e-05,
+      "loss": 4.7698,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.437694704049845e-05,
+      "loss": 4.7391,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.429906542056075e-05,
+      "loss": 4.0485,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.422118380062306e-05,
+      "loss": 4.7104,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.414330218068535e-05,
+      "loss": 4.3291,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.406542056074766e-05,
+      "loss": 4.4961,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.398753894080997e-05,
+      "loss": 4.2818,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.390965732087228e-05,
+      "loss": 4.3396,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.383177570093458e-05,
+      "loss": 4.5339,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.375389408099688e-05,
+      "loss": 5.0919,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.36760124610592e-05,
+      "loss": 4.2608,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.35981308411215e-05,
+      "loss": 4.2489,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.35202492211838e-05,
+      "loss": 4.6593,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.344236760124611e-05,
+      "loss": 4.0924,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.336448598130842e-05,
+      "loss": 4.2453,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.328660436137072e-05,
+      "loss": 4.1412,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.320872274143302e-05,
+      "loss": 4.5701,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.313084112149533e-05,
+      "loss": 4.7639,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.305295950155764e-05,
+      "loss": 5.9174,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.297507788161995e-05,
+      "loss": 4.4482,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.289719626168224e-05,
+      "loss": 4.6065,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.281931464174455e-05,
+      "loss": 4.3837,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.274143302180686e-05,
+      "loss": 4.4017,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.266355140186917e-05,
+      "loss": 4.4166,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.258566978193147e-05,
+      "loss": 4.5272,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.250778816199377e-05,
+      "loss": 4.2663,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.242990654205608e-05,
+      "loss": 4.6292,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.235202492211839e-05,
+      "loss": 4.6275,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.227414330218069e-05,
+      "loss": 4.7781,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.2196261682243e-05,
+      "loss": 4.244,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.21183800623053e-05,
+      "loss": 4.508,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.204049844236762e-05,
+      "loss": 4.6605,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.196261682242991e-05,
+      "loss": 4.3986,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.188473520249222e-05,
+      "loss": 4.7589,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.180685358255453e-05,
+      "loss": 4.3773,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.172897196261682e-05,
+      "loss": 4.508,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.165109034267913e-05,
+      "loss": 4.5915,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.157320872274143e-05,
+      "loss": 4.4711,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.149532710280374e-05,
+      "loss": 4.611,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.141744548286605e-05,
+      "loss": 4.3636,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.133956386292834e-05,
+      "loss": 4.6357,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.126168224299065e-05,
+      "loss": 4.8975,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.118380062305296e-05,
+      "loss": 4.4576,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.110591900311527e-05,
+      "loss": 4.565,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.102803738317757e-05,
+      "loss": 4.7625,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.095015576323987e-05,
+      "loss": 4.6593,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.087227414330218e-05,
+      "loss": 4.3407,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.079439252336449e-05,
+      "loss": 4.3585,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.071651090342679e-05,
+      "loss": 4.4165,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.06386292834891e-05,
+      "loss": 4.7313,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.05607476635514e-05,
+      "loss": 4.3207,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.048286604361372e-05,
+      "loss": 4.6034,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.040498442367601e-05,
+      "loss": 4.8277,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.032710280373832e-05,
+      "loss": 4.6736,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.024922118380063e-05,
+      "loss": 4.6905,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.017133956386294e-05,
+      "loss": 4.9285,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.009345794392523e-05,
+      "loss": 4.1706,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.001557632398754e-05,
+      "loss": 4.7265,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.993769470404985e-05,
+      "loss": 4.7344,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.985981308411216e-05,
+      "loss": 4.3939,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.978193146417446e-05,
+      "loss": 4.4702,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.970404984423677e-05,
+      "loss": 5.9223,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.962616822429907e-05,
+      "loss": 4.4691,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.954828660436138e-05,
+      "loss": 4.4626,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.947040498442368e-05,
+      "loss": 4.2363,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.939252336448599e-05,
+      "loss": 4.3435,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.93146417445483e-05,
+      "loss": 5.1803,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.92367601246106e-05,
+      "loss": 5.6149,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.91588785046729e-05,
+      "loss": 5.3223,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.908099688473521e-05,
+      "loss": 4.4559,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.90031152647975e-05,
+      "loss": 4.3908,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.892523364485982e-05,
+      "loss": 4.8824,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.884735202492212e-05,
+      "loss": 4.205,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.876947040498442e-05,
+      "loss": 4.4449,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.869158878504673e-05,
+      "loss": 4.8105,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.861370716510904e-05,
+      "loss": 4.6456,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.853582554517133e-05,
+      "loss": 4.2413,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.845794392523364e-05,
+      "loss": 4.4511,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.838006230529595e-05,
+      "loss": 4.3313,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.830218068535826e-05,
+      "loss": 4.4102,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.822429906542056e-05,
+      "loss": 4.3799,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.814641744548287e-05,
+      "loss": 4.5986,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.806853582554517e-05,
+      "loss": 4.3488,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.799065420560748e-05,
+      "loss": 4.6507,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.791277258566978e-05,
+      "loss": 4.3653,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.783489096573209e-05,
+      "loss": 4.1537,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.77570093457944e-05,
+      "loss": 4.8184,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.76791277258567e-05,
+      "loss": 4.4075,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.7601246105919e-05,
+      "loss": 4.0893,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.752336448598131e-05,
+      "loss": 4.3137,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.744548286604362e-05,
+      "loss": 4.3618,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.736760124610593e-05,
+      "loss": 4.1979,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.728971962616822e-05,
+      "loss": 4.292,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.721183800623053e-05,
+      "loss": 4.6538,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.713395638629284e-05,
+      "loss": 4.9366,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.705607476635515e-05,
+      "loss": 4.2921,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.697819314641745e-05,
+      "loss": 4.3445,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.690031152647976e-05,
+      "loss": 4.194,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.682242990654207e-05,
+      "loss": 4.5145,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.674454828660437e-05,
+      "loss": 4.7051,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 4.5534,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.658878504672898e-05,
+      "loss": 4.5466,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.651090342679129e-05,
+      "loss": 4.4489,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.64330218068536e-05,
+      "loss": 4.7372,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.635514018691589e-05,
+      "loss": 4.5042,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.62772585669782e-05,
+      "loss": 4.3994,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.61993769470405e-05,
+      "loss": 4.0729,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.61214953271028e-05,
+      "loss": 4.1231,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.60436137071651e-05,
+      "loss": 4.7086,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.596573208722741e-05,
+      "loss": 4.489,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.588785046728972e-05,
+      "loss": 4.3496,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.580996884735203e-05,
+      "loss": 5.2228,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.573208722741432e-05,
+      "loss": 4.671,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.565420560747663e-05,
+      "loss": 4.4398,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.557632398753894e-05,
+      "loss": 4.5184,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.549844236760125e-05,
+      "loss": 4.1917,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.542056074766355e-05,
+      "loss": 4.4664,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.534267912772586e-05,
+      "loss": 5.1438,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.526479750778817e-05,
+      "loss": 3.9848,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.518691588785047e-05,
+      "loss": 4.3759,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.510903426791277e-05,
+      "loss": 4.3488,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.503115264797508e-05,
+      "loss": 4.3789,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.495327102803739e-05,
+      "loss": 4.0584,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.48753894080997e-05,
+      "loss": 4.4669,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.479750778816199e-05,
+      "loss": 4.372,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.47196261682243e-05,
+      "loss": 4.5326,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.464174454828661e-05,
+      "loss": 4.118,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.456386292834892e-05,
+      "loss": 4.6913,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.448598130841122e-05,
+      "loss": 4.7556,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.440809968847352e-05,
+      "loss": 4.3374,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.433021806853583e-05,
+      "loss": 4.3574,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.425233644859814e-05,
+      "loss": 4.8321,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.417445482866044e-05,
+      "loss": 4.3267,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.409657320872275e-05,
+      "loss": 4.2422,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.401869158878506e-05,
+      "loss": 4.2042,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.394080996884737e-05,
+      "loss": 4.6413,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.386292834890966e-05,
+      "loss": 4.2778,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.378504672897197e-05,
+      "loss": 4.5639,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.370716510903428e-05,
+      "loss": 4.1926,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.362928348909657e-05,
+      "loss": 4.2094,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.355140186915888e-05,
+      "loss": 4.3524,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.347352024922119e-05,
+      "loss": 4.3888,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.339563862928349e-05,
+      "loss": 4.8345,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.33177570093458e-05,
+      "loss": 4.4651,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.323987538940809e-05,
+      "loss": 4.3839,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.31619937694704e-05,
+      "loss": 4.9496,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.308411214953271e-05,
+      "loss": 4.4045,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.300623052959502e-05,
+      "loss": 4.2761,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.292834890965732e-05,
+      "loss": 4.2605,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.285046728971962e-05,
+      "loss": 3.9765,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.277258566978193e-05,
+      "loss": 3.976,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.269470404984424e-05,
+      "loss": 4.6331,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.261682242990654e-05,
+      "loss": 4.5048,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.253894080996885e-05,
+      "loss": 4.0882,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.246105919003116e-05,
+      "loss": 4.9075,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.238317757009346e-05,
+      "loss": 4.514,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.230529595015576e-05,
+      "loss": 4.3101,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.222741433021807e-05,
+      "loss": 4.487,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.214953271028038e-05,
+      "loss": 4.2208,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.207165109034269e-05,
+      "loss": 4.6124,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.199376947040498e-05,
+      "loss": 4.6812,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.191588785046729e-05,
+      "loss": 4.7564,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.18380062305296e-05,
+      "loss": 4.4887,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.176012461059191e-05,
+      "loss": 4.3399,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.16822429906542e-05,
+      "loss": 3.8386,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.160436137071651e-05,
+      "loss": 4.2554,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.152647975077882e-05,
+      "loss": 4.9028,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.144859813084113e-05,
+      "loss": 5.3328,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.137071651090343e-05,
+      "loss": 4.4733,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 4.615284442901611,
+      "eval_runtime": 941.2047,
+      "eval_samples_per_second": 2.807,
+      "eval_steps_per_second": 0.352,
+      "eval_wer": 1.9654105513685045,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.129283489096574e-05,
+      "loss": 4.6829,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.121495327102805e-05,
+      "loss": 4.3793,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.113707165109036e-05,
+      "loss": 4.5775,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.105919003115265e-05,
+      "loss": 4.6767,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.0981308411214953e-05,
+      "loss": 4.4274,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.090342679127726e-05,
+      "loss": 4.3074,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.082554517133957e-05,
+      "loss": 4.5089,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.074766355140187e-05,
+      "loss": 4.5339,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.0669781931464176e-05,
+      "loss": 4.5604,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.0591900311526485e-05,
+      "loss": 4.1409,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.0514018691588794e-05,
+      "loss": 4.4076,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.043613707165109e-05,
+      "loss": 4.3945,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.03582554517134e-05,
+      "loss": 4.5356,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.028037383177571e-05,
+      "loss": 4.4349,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.020249221183801e-05,
+      "loss": 5.0375,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.012461059190031e-05,
+      "loss": 4.2961,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.0046728971962615e-05,
+      "loss": 4.4754,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9968847352024924e-05,
+      "loss": 4.3219,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.989096573208723e-05,
+      "loss": 4.5984,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.981308411214953e-05,
+      "loss": 4.7447,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.973520249221184e-05,
+      "loss": 4.5289,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9657320872274146e-05,
+      "loss": 4.5183,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9579439252336456e-05,
+      "loss": 4.4976,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.950155763239875e-05,
+      "loss": 4.5859,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.942367601246106e-05,
+      "loss": 4.4019,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.934579439252337e-05,
+      "loss": 4.3922,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.926791277258568e-05,
+      "loss": 4.3908,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9190031152647974e-05,
+      "loss": 4.6303,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.911214953271028e-05,
+      "loss": 4.3819,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.903426791277259e-05,
+      "loss": 4.4018,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.89563862928349e-05,
+      "loss": 4.532,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8878504672897196e-05,
+      "loss": 4.2605,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8800623052959505e-05,
+      "loss": 4.5421,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.872274143302181e-05,
+      "loss": 4.0662,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.864485981308412e-05,
+      "loss": 4.246,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.856697819314641e-05,
+      "loss": 4.6471,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.848909657320872e-05,
+      "loss": 4.4306,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.841121495327103e-05,
+      "loss": 4.3853,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 4.3287,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8255451713395635e-05,
+      "loss": 4.9969,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8177570093457944e-05,
+      "loss": 4.6844,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.809968847352025e-05,
+      "loss": 4.455,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.802180685358256e-05,
+      "loss": 4.7139,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.794392523364486e-05,
+      "loss": 4.5008,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.786604361370717e-05,
+      "loss": 5.0222,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7788161993769476e-05,
+      "loss": 4.475,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7710280373831785e-05,
+      "loss": 4.396,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.763239875389408e-05,
+      "loss": 4.3261,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.755451713395639e-05,
+      "loss": 4.5298,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.74766355140187e-05,
+      "loss": 5.2768,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7398753894081e-05,
+      "loss": 4.9073,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.73208722741433e-05,
+      "loss": 4.2247,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7242990654205605e-05,
+      "loss": 4.2371,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7165109034267914e-05,
+      "loss": 4.1627,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7087227414330223e-05,
+      "loss": 4.3326,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.700934579439252e-05,
+      "loss": 4.2372,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.693146417445483e-05,
+      "loss": 4.6594,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685358255451714e-05,
+      "loss": 4.352,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6775700934579446e-05,
+      "loss": 4.8343,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.669781931464174e-05,
+      "loss": 4.138,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.661993769470405e-05,
+      "loss": 4.4469,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.654205607476636e-05,
+      "loss": 4.303,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.646417445482867e-05,
+      "loss": 4.5367,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6386292834890964e-05,
+      "loss": 3.9678,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6308411214953273e-05,
+      "loss": 4.3845,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.623052959501558e-05,
+      "loss": 4.0831,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6152647975077885e-05,
+      "loss": 4.369,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.607476635514019e-05,
+      "loss": 4.2208,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.5996884735202496e-05,
+      "loss": 4.5197,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.59190031152648e-05,
+      "loss": 4.2602,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.584112149532711e-05,
+      "loss": 4.2613,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.57632398753894e-05,
+      "loss": 4.0353,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.568535825545171e-05,
+      "loss": 4.1289,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.560747663551402e-05,
+      "loss": 4.5645,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.552959501557633e-05,
+      "loss": 4.334,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.5451713395638626e-05,
+      "loss": 4.7573,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.5373831775700935e-05,
+      "loss": 4.3613,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.5295950155763244e-05,
+      "loss": 4.2563,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.521806853582555e-05,
+      "loss": 4.3157,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.514018691588785e-05,
+      "loss": 4.6432,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.506230529595016e-05,
+      "loss": 4.468,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4984423676012466e-05,
+      "loss": 4.56,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4906542056074776e-05,
+      "loss": 4.012,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.482866043613707e-05,
+      "loss": 4.3352,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.475077881619938e-05,
+      "loss": 4.3486,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.467289719626168e-05,
+      "loss": 4.5342,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.459501557632399e-05,
+      "loss": 4.5278,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4517133956386294e-05,
+      "loss": 4.5199,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4439252336448596e-05,
+      "loss": 4.7292,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4361370716510905e-05,
+      "loss": 4.0688,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4283489096573214e-05,
+      "loss": 4.2223,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.420560747663551e-05,
+      "loss": 4.3992,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.412772585669782e-05,
+      "loss": 4.4786,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.404984423676013e-05,
+      "loss": 4.7942,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.397196261682244e-05,
+      "loss": 4.4041,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.389408099688473e-05,
+      "loss": 4.7723,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.381619937694704e-05,
+      "loss": 4.6,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.373831775700935e-05,
+      "loss": 5.2447,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.366043613707166e-05,
+      "loss": 4.6042,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3582554517133955e-05,
+      "loss": 4.8501,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3504672897196264e-05,
+      "loss": 4.1882,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.342679127725857e-05,
+      "loss": 4.4367,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3348909657320875e-05,
+      "loss": 4.4387,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.327102803738318e-05,
+      "loss": 4.0813,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.319314641744548e-05,
+      "loss": 4.1411,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.311526479750779e-05,
+      "loss": 4.1148,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.30373831775701e-05,
+      "loss": 3.9163,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.2959501557632394e-05,
+      "loss": 4.0676,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.28816199376947e-05,
+      "loss": 4.3732,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.280373831775701e-05,
+      "loss": 4.4373,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.272585669781932e-05,
+      "loss": 4.2394,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.2647975077881616e-05,
+      "loss": 4.4812,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.2570093457943925e-05,
+      "loss": 4.1389,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.2492211838006234e-05,
+      "loss": 4.6663,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.2414330218068543e-05,
+      "loss": 4.8786,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.233644859813084e-05,
+      "loss": 4.3887,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.225856697819315e-05,
+      "loss": 4.4003,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.218068535825546e-05,
+      "loss": 4.3238,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.210280373831776e-05,
+      "loss": 4.2319,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.202492211838006e-05,
+      "loss": 4.1697,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.194704049844237e-05,
+      "loss": 4.3429,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.186915887850467e-05,
+      "loss": 4.2449,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.179127725856698e-05,
+      "loss": 4.0401,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.171339563862928e-05,
+      "loss": 4.1221,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.163551401869159e-05,
+      "loss": 4.7439,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1557632398753896e-05,
+      "loss": 4.2685,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1479750778816205e-05,
+      "loss": 4.238,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.14018691588785e-05,
+      "loss": 4.7109,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.132398753894081e-05,
+      "loss": 3.9161,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.124610591900312e-05,
+      "loss": 4.652,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.116822429906543e-05,
+      "loss": 4.4683,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.109034267912772e-05,
+      "loss": 4.2397,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.101246105919003e-05,
+      "loss": 4.5041,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.093457943925234e-05,
+      "loss": 4.2838,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.085669781931465e-05,
+      "loss": 4.6411,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.0778816199376946e-05,
+      "loss": 4.2182,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.0700934579439255e-05,
+      "loss": 4.3461,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.062305295950156e-05,
+      "loss": 4.3968,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.0545171339563866e-05,
+      "loss": 4.1808,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.046728971962617e-05,
+      "loss": 4.6302,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.038940809968847e-05,
+      "loss": 4.2863,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.031152647975078e-05,
+      "loss": 4.635,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.023364485981309e-05,
+      "loss": 4.3922,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.0155763239875384e-05,
+      "loss": 4.6227,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.007788161993769e-05,
+      "loss": 4.4209,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-05,
+      "loss": 4.5097,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.992211838006231e-05,
+      "loss": 4.4401,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9844236760124614e-05,
+      "loss": 4.2722,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.976635514018692e-05,
+      "loss": 4.6484,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9688473520249225e-05,
+      "loss": 4.2247,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9610591900311534e-05,
+      "loss": 4.3661,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9532710280373836e-05,
+      "loss": 4.4102,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9454828660436145e-05,
+      "loss": 4.4548,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.937694704049845e-05,
+      "loss": 4.4076,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.929906542056075e-05,
+      "loss": 4.3881,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.922118380062305e-05,
+      "loss": 4.3717,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.914330218068536e-05,
+      "loss": 4.136,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9065420560747664e-05,
+      "loss": 4.2634,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.898753894080997e-05,
+      "loss": 4.0908,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8909657320872275e-05,
+      "loss": 4.2626,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8831775700934584e-05,
+      "loss": 4.4041,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8753894080996886e-05,
+      "loss": 4.2338,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8676012461059195e-05,
+      "loss": 4.3419,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.85981308411215e-05,
+      "loss": 4.1212,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.852024922118381e-05,
+      "loss": 4.1652,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.844236760124611e-05,
+      "loss": 4.5154,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.836448598130842e-05,
+      "loss": 4.2805,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.828660436137072e-05,
+      "loss": 4.0478,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.820872274143303e-05,
+      "loss": 4.3595,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.813084112149533e-05,
+      "loss": 4.4268,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8052959501557634e-05,
+      "loss": 4.1907,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.797507788161994e-05,
+      "loss": 4.0994,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7897196261682245e-05,
+      "loss": 4.4798,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.781931464174455e-05,
+      "loss": 4.1039,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.774143302180686e-05,
+      "loss": 4.243,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.766355140186916e-05,
+      "loss": 4.6006,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.758566978193147e-05,
+      "loss": 4.3773,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.750778816199377e-05,
+      "loss": 4.3939,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.742990654205608e-05,
+      "loss": 3.9865,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.735202492211838e-05,
+      "loss": 4.221,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.727414330218069e-05,
+      "loss": 4.1083,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.719626168224299e-05,
+      "loss": 4.5258,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.71183800623053e-05,
+      "loss": 4.3894,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7040498442367604e-05,
+      "loss": 4.1279,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6962616822429913e-05,
+      "loss": 4.9217,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6884735202492216e-05,
+      "loss": 4.0905,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6806853582554525e-05,
+      "loss": 4.5697,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.672897196261683e-05,
+      "loss": 4.2326,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.665109034267913e-05,
+      "loss": 4.2444,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.657320872274143e-05,
+      "loss": 4.0309,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.649532710280374e-05,
+      "loss": 4.3294,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.641744548286604e-05,
+      "loss": 4.4713,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.633956386292835e-05,
+      "loss": 4.7732,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6261682242990654e-05,
+      "loss": 4.2318,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6183800623052963e-05,
+      "loss": 5.0241,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6105919003115266e-05,
+      "loss": 4.2359,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6028037383177575e-05,
+      "loss": 4.0989,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.595015576323988e-05,
+      "loss": 4.5598,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5872274143302186e-05,
+      "loss": 4.4083,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579439252336449e-05,
+      "loss": 4.0506,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.57165109034268e-05,
+      "loss": 4.3192,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.56386292834891e-05,
+      "loss": 4.3377,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.556074766355141e-05,
+      "loss": 4.1608,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.548286604361371e-05,
+      "loss": 4.0382,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.540498442367602e-05,
+      "loss": 3.9645,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.532710280373832e-05,
+      "loss": 4.3084,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5249221183800625e-05,
+      "loss": 4.507,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.517133956386293e-05,
+      "loss": 4.2311,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5093457943925236e-05,
+      "loss": 4.437,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.501557632398754e-05,
+      "loss": 4.1853,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.493769470404985e-05,
+      "loss": 4.3222,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.485981308411215e-05,
+      "loss": 4.4712,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.478193146417446e-05,
+      "loss": 4.3735,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.470404984423676e-05,
+      "loss": 4.1071,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.462616822429907e-05,
+      "loss": 4.188,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.454828660436137e-05,
+      "loss": 4.3132,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.447040498442368e-05,
+      "loss": 4.233,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4392523364485984e-05,
+      "loss": 4.0084,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.431464174454829e-05,
+      "loss": 3.9989,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4236760124610595e-05,
+      "loss": 4.3191,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4158878504672904e-05,
+      "loss": 4.2507,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4080996884735206e-05,
+      "loss": 3.9392,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.400311526479751e-05,
+      "loss": 4.4491,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.392523364485982e-05,
+      "loss": 4.4103,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.384735202492212e-05,
+      "loss": 4.849,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.376947040498442e-05,
+      "loss": 4.3933,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.369158878504673e-05,
+      "loss": 4.0786,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3613707165109034e-05,
+      "loss": 4.1735,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.353582554517134e-05,
+      "loss": 4.1158,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3457943925233645e-05,
+      "loss": 4.614,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3380062305295954e-05,
+      "loss": 4.3099,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3302180685358256e-05,
+      "loss": 4.3833,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3224299065420565e-05,
+      "loss": 4.3767,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.314641744548287e-05,
+      "loss": 4.0846,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.306853582554518e-05,
+      "loss": 4.5863,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.299065420560748e-05,
+      "loss": 4.2614,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.291277258566979e-05,
+      "loss": 4.3551,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.283489096573209e-05,
+      "loss": 4.0372,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.27570093457944e-05,
+      "loss": 4.2832,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26791277258567e-05,
+      "loss": 4.6511,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2601246105919004e-05,
+      "loss": 4.3879,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2523364485981306e-05,
+      "loss": 4.6423,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2445482866043615e-05,
+      "loss": 4.7758,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236760124610592e-05,
+      "loss": 4.3009,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228971962616823e-05,
+      "loss": 3.9736,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.221183800623053e-05,
+      "loss": 4.4961,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.213395638629284e-05,
+      "loss": 4.4246,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.205607476635514e-05,
+      "loss": 4.2787,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.197819314641745e-05,
+      "loss": 4.4894,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.190031152647975e-05,
+      "loss": 4.5764,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.182242990654206e-05,
+      "loss": 4.5264,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174454828660436e-05,
+      "loss": 4.4131,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 4.4158,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1588785046728974e-05,
+      "loss": 4.2926,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1510903426791283e-05,
+      "loss": 4.6846,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1433021806853586e-05,
+      "loss": 4.2237,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1355140186915895e-05,
+      "loss": 4.1456,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12772585669782e-05,
+      "loss": 4.0339,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11993769470405e-05,
+      "loss": 4.3679,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11214953271028e-05,
+      "loss": 4.5789,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104361370716511e-05,
+      "loss": 4.1616,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096573208722741e-05,
+      "loss": 4.3772,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088785046728972e-05,
+      "loss": 4.3432,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0809968847352024e-05,
+      "loss": 4.2312,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073208722741433e-05,
+      "loss": 4.9508,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0654205607476636e-05,
+      "loss": 4.4536,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0576323987538945e-05,
+      "loss": 4.3635,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049844236760125e-05,
+      "loss": 4.1063,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0420560747663556e-05,
+      "loss": 4.4203,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034267912772586e-05,
+      "loss": 4.1518,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026479750778817e-05,
+      "loss": 4.3135,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018691588785047e-05,
+      "loss": 4.1203,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010903426791278e-05,
+      "loss": 4.4608,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003115264797508e-05,
+      "loss": 4.1746,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995327102803739e-05,
+      "loss": 4.2147,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.987538940809969e-05,
+      "loss": 4.6276,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9797507788161995e-05,
+      "loss": 4.368,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.97196261682243e-05,
+      "loss": 4.6408,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9641744548286606e-05,
+      "loss": 4.3779,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956386292834891e-05,
+      "loss": 4.3127,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948598130841122e-05,
+      "loss": 4.2948,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940809968847352e-05,
+      "loss": 4.2811,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933021806853583e-05,
+      "loss": 4.4045,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925233644859813e-05,
+      "loss": 4.1635,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917445482866044e-05,
+      "loss": 4.2613,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909657320872274e-05,
+      "loss": 4.2453,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901869158878505e-05,
+      "loss": 4.3448,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8940809968847354e-05,
+      "loss": 4.1962,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886292834890966e-05,
+      "loss": 4.3113,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8785046728971965e-05,
+      "loss": 4.0098,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8707165109034274e-05,
+      "loss": 4.4021,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8629283489096576e-05,
+      "loss": 4.796,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855140186915888e-05,
+      "loss": 5.2543,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.847352024922119e-05,
+      "loss": 4.0699,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.839563862928349e-05,
+      "loss": 4.0832,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.831775700934579e-05,
+      "loss": 3.9291,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.82398753894081e-05,
+      "loss": 4.1121,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8161993769470404e-05,
+      "loss": 3.7655,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.808411214953271e-05,
+      "loss": 5.036,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8006230529595015e-05,
+      "loss": 3.9566,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7928348909657324e-05,
+      "loss": 4.376,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7850467289719626e-05,
+      "loss": 4.4706,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7772585669781935e-05,
+      "loss": 4.3632,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.769470404984424e-05,
+      "loss": 4.2771,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.761682242990655e-05,
+      "loss": 4.161,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.753894080996885e-05,
+      "loss": 4.2584,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.746105919003116e-05,
+      "loss": 4.3258,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.738317757009346e-05,
+      "loss": 4.0207,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.730529595015577e-05,
+      "loss": 4.1747,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.722741433021807e-05,
+      "loss": 4.0289,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7149532710280374e-05,
+      "loss": 4.0952,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7071651090342676e-05,
+      "loss": 4.476,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6993769470404985e-05,
+      "loss": 4.2005,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.691588785046729e-05,
+      "loss": 4.0494,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.68380062305296e-05,
+      "loss": 4.5126,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.67601246105919e-05,
+      "loss": 4.2933,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.668224299065421e-05,
+      "loss": 4.0876,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.660436137071651e-05,
+      "loss": 4.8204,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.652647975077882e-05,
+      "loss": 4.1393,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.644859813084112e-05,
+      "loss": 3.8391,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.637071651090343e-05,
+      "loss": 3.9118,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.629283489096573e-05,
+      "loss": 4.1447,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.621495327102804e-05,
+      "loss": 4.1678,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6137071651090344e-05,
+      "loss": 4.1614,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.605919003115265e-05,
+      "loss": 4.2597,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5981308411214956e-05,
+      "loss": 3.9822,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5903426791277265e-05,
+      "loss": 3.8439,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.582554517133957e-05,
+      "loss": 4.1795,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.574766355140187e-05,
+      "loss": 4.4386,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.566978193146417e-05,
+      "loss": 4.5739,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.559190031152648e-05,
+      "loss": 4.1422,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.551401869158878e-05,
+      "loss": 4.3292,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.543613707165109e-05,
+      "loss": 4.2655,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5358255451713394e-05,
+      "loss": 4.3378,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.52803738317757e-05,
+      "loss": 4.0101,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5202492211838006e-05,
+      "loss": 4.3536,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5124610591900315e-05,
+      "loss": 4.3953,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.504672897196262e-05,
+      "loss": 4.4666,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4968847352024926e-05,
+      "loss": 4.5317,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.489096573208723e-05,
+      "loss": 4.578,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.481308411214954e-05,
+      "loss": 4.4997,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.473520249221184e-05,
+      "loss": 4.3168,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.465732087227415e-05,
+      "loss": 4.4219,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.457943925233645e-05,
+      "loss": 4.8721,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.450155763239875e-05,
+      "loss": 4.5965,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.442367601246106e-05,
+      "loss": 4.4048,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4345794392523365e-05,
+      "loss": 5.1924,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.426791277258567e-05,
+      "loss": 4.1583,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4190031152647976e-05,
+      "loss": 4.5196,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.411214953271028e-05,
+      "loss": 4.4903,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.403426791277259e-05,
+      "loss": 4.3965,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.395638629283489e-05,
+      "loss": 4.4282,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.38785046728972e-05,
+      "loss": 4.3888,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.38006230529595e-05,
+      "loss": 4.5135,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.372274143302181e-05,
+      "loss": 4.1862,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.364485981308411e-05,
+      "loss": 4.6593,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.356697819314642e-05,
+      "loss": 3.9444,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3489096573208724e-05,
+      "loss": 4.4367,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.341121495327103e-05,
+      "loss": 4.4877,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 4.7337,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3255451713395644e-05,
+      "loss": 4.2004,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3177570093457946e-05,
+      "loss": 4.0808,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.309968847352025e-05,
+      "loss": 4.007,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.302180685358255e-05,
+      "loss": 4.7419,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.294392523364486e-05,
+      "loss": 4.2006,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.286604361370716e-05,
+      "loss": 3.9478,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.278816199376947e-05,
+      "loss": 4.3813,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2710280373831774e-05,
+      "loss": 4.0479,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.263239875389408e-05,
+      "loss": 4.3773,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2554517133956385e-05,
+      "loss": 4.3997,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2476635514018694e-05,
+      "loss": 4.3293,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2398753894080996e-05,
+      "loss": 3.9908,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2320872274143305e-05,
+      "loss": 4.2449,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.224299065420561e-05,
+      "loss": 4.3402,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.216510903426792e-05,
+      "loss": 3.9674,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.208722741433022e-05,
+      "loss": 3.9451,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.200934579439253e-05,
+      "loss": 4.2769,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.193146417445483e-05,
+      "loss": 4.1836,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.185358255451714e-05,
+      "loss": 4.1871,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.177570093457944e-05,
+      "loss": 4.4366,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1697819314641744e-05,
+      "loss": 3.8294,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1619937694704046e-05,
+      "loss": 4.4796,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.1542056074766355e-05,
+      "loss": 4.0509,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.146417445482866e-05,
+      "loss": 4.3866,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.1386292834890967e-05,
+      "loss": 4.4418,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.130841121495327e-05,
+      "loss": 4.4472,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.123052959501558e-05,
+      "loss": 4.4474,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.115264797507788e-05,
+      "loss": 4.269,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.107476635514019e-05,
+      "loss": 4.1612,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.099688473520249e-05,
+      "loss": 4.4594,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.09190031152648e-05,
+      "loss": 4.1118,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.08411214953271e-05,
+      "loss": 4.0935,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.076323987538941e-05,
+      "loss": 4.1921,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0685358255451714e-05,
+      "loss": 4.399,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.060747663551402e-05,
+      "loss": 4.8534,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0529595015576326e-05,
+      "loss": 4.1574,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.045171339563863e-05,
+      "loss": 4.4921,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0373831775700934e-05,
+      "loss": 4.1745,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0295950155763243e-05,
+      "loss": 4.2879,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0218068535825545e-05,
+      "loss": 4.3816,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0140186915887854e-05,
+      "loss": 4.0071,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0062305295950156e-05,
+      "loss": 4.3023,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9984423676012462e-05,
+      "loss": 4.0883,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9906542056074764e-05,
+      "loss": 4.1687,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9828660436137073e-05,
+      "loss": 4.0627,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9750778816199376e-05,
+      "loss": 4.8978,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9672897196261685e-05,
+      "loss": 4.2016,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9595015576323987e-05,
+      "loss": 4.1435,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9517133956386296e-05,
+      "loss": 4.091,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9439252336448598e-05,
+      "loss": 4.0738,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9361370716510904e-05,
+      "loss": 3.9693,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9283489096573206e-05,
+      "loss": 4.0482,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9205607476635515e-05,
+      "loss": 4.4403,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9127725856697818e-05,
+      "loss": 4.5655,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9049844236760127e-05,
+      "loss": 4.2177,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.897196261682243e-05,
+      "loss": 4.4108,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8894080996884738e-05,
+      "loss": 4.2348,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.881619937694704e-05,
+      "loss": 4.3636,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.873831775700935e-05,
+      "loss": 4.1983,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.866043613707165e-05,
+      "loss": 4.0838,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8582554517133957e-05,
+      "loss": 4.2093,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.850467289719626e-05,
+      "loss": 4.3004,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.842679127725857e-05,
+      "loss": 4.2148,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.834890965732087e-05,
+      "loss": 4.2197,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.827102803738318e-05,
+      "loss": 4.4343,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8193146417445482e-05,
+      "loss": 4.4103,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.811526479750779e-05,
+      "loss": 4.0669,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8037383177570094e-05,
+      "loss": 4.0813,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.79595015576324e-05,
+      "loss": 4.2913,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.78816199376947e-05,
+      "loss": 4.5403,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.780373831775701e-05,
+      "loss": 4.474,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7725856697819313e-05,
+      "loss": 4.2164,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7647975077881622e-05,
+      "loss": 4.6647,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7570093457943924e-05,
+      "loss": 4.132,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7492211838006233e-05,
+      "loss": 3.9877,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7414330218068536e-05,
+      "loss": 4.1841,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.733644859813084e-05,
+      "loss": 4.3926,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7258566978193147e-05,
+      "loss": 4.1512,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7180685358255453e-05,
+      "loss": 4.3345,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7102803738317755e-05,
+      "loss": 4.2993,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7024922118380064e-05,
+      "loss": 4.5895,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6947040498442366e-05,
+      "loss": 4.0648,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6869158878504675e-05,
+      "loss": 4.1287,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6791277258566978e-05,
+      "loss": 4.1253,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6713395638629287e-05,
+      "loss": 4.2013,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.663551401869159e-05,
+      "loss": 4.4121,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6557632398753895e-05,
+      "loss": 4.308,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6479750778816197e-05,
+      "loss": 4.3283,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6401869158878506e-05,
+      "loss": 4.492,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6323987538940808e-05,
+      "loss": 4.0404,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6246105919003117e-05,
+      "loss": 4.1482,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.616822429906542e-05,
+      "loss": 4.2487,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.609034267912773e-05,
+      "loss": 4.1202,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.601246105919003e-05,
+      "loss": 4.0152,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5934579439252337e-05,
+      "loss": 4.1621,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.585669781931464e-05,
+      "loss": 4.1045,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5778816199376948e-05,
+      "loss": 4.5036,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.570093457943925e-05,
+      "loss": 4.3859,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.562305295950156e-05,
+      "loss": 4.0852,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.554517133956386e-05,
+      "loss": 4.6093,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.546728971962617e-05,
+      "loss": 4.2873,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5389408099688473e-05,
+      "loss": 4.2908,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.531152647975078e-05,
+      "loss": 4.1769,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5233644859813084e-05,
+      "loss": 4.3563,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.515576323987539e-05,
+      "loss": 4.5103,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5077881619937692e-05,
+      "loss": 4.0124,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5e-05,
+      "loss": 4.1602,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4922118380062307e-05,
+      "loss": 4.1945,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4844236760124613e-05,
+      "loss": 3.9754,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4766355140186918e-05,
+      "loss": 3.8922,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4688473520249224e-05,
+      "loss": 4.1642,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4610591900311526e-05,
+      "loss": 3.9212,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4532710280373832e-05,
+      "loss": 4.0488,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4454828660436138e-05,
+      "loss": 4.1397,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4376947040498443e-05,
+      "loss": 4.2296,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.429906542056075e-05,
+      "loss": 4.2054,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4221183800623055e-05,
+      "loss": 4.3405,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.414330218068536e-05,
+      "loss": 3.8345,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4065420560747666e-05,
+      "loss": 4.2334,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.398753894080997e-05,
+      "loss": 4.0399,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3909657320872274e-05,
+      "loss": 4.143,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.383177570093458e-05,
+      "loss": 4.7482,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3753894080996885e-05,
+      "loss": 4.241,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.367601246105919e-05,
+      "loss": 3.8591,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3598130841121497e-05,
+      "loss": 4.0871,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3520249221183802e-05,
+      "loss": 4.3268,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3442367601246108e-05,
+      "loss": 4.3531,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3364485981308414e-05,
+      "loss": 4.2356,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3286604361370716e-05,
+      "loss": 4.2386,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.320872274143302e-05,
+      "loss": 3.9994,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3130841121495327e-05,
+      "loss": 4.696,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3052959501557633e-05,
+      "loss": 4.2167,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.297507788161994e-05,
+      "loss": 3.8667,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2897196261682244e-05,
+      "loss": 4.0731,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.281931464174455e-05,
+      "loss": 4.0753,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2741433021806856e-05,
+      "loss": 5.3092,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.266355140186916e-05,
+      "loss": 4.3613,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2585669781931463e-05,
+      "loss": 4.418,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.250778816199377e-05,
+      "loss": 4.001,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2429906542056075e-05,
+      "loss": 4.2575,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 4.330733299255371,
+      "eval_runtime": 943.0842,
+      "eval_samples_per_second": 2.801,
+      "eval_steps_per_second": 0.351,
+      "eval_wer": 1.92959143197144,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.235202492211838e-05,
+      "loss": 4.2634,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2274143302180686e-05,
+      "loss": 4.2131,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2196261682242992e-05,
+      "loss": 4.378,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2118380062305298e-05,
+      "loss": 4.245,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2040498442367603e-05,
+      "loss": 4.2706,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.196261682242991e-05,
+      "loss": 4.1096,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.188473520249221e-05,
+      "loss": 4.0582,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1806853582554517e-05,
+      "loss": 4.1962,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1728971962616822e-05,
+      "loss": 4.2645,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1651090342679128e-05,
+      "loss": 4.1499,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1573208722741434e-05,
+      "loss": 3.9418,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.149532710280374e-05,
+      "loss": 4.3031,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1417445482866045e-05,
+      "loss": 3.9088,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.133956386292835e-05,
+      "loss": 4.0311,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1261682242990653e-05,
+      "loss": 4.1776,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.118380062305296e-05,
+      "loss": 4.4708,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1105919003115264e-05,
+      "loss": 4.0567,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.102803738317757e-05,
+      "loss": 4.3632,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0950155763239876e-05,
+      "loss": 4.3634,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.087227414330218e-05,
+      "loss": 4.0794,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0794392523364487e-05,
+      "loss": 4.2829,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0716510903426793e-05,
+      "loss": 4.241,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.06386292834891e-05,
+      "loss": 4.6959,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.05607476635514e-05,
+      "loss": 3.9899,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0482866043613706e-05,
+      "loss": 4.0553,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0404984423676012e-05,
+      "loss": 3.9625,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0327102803738318e-05,
+      "loss": 4.0764,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0249221183800623e-05,
+      "loss": 4.0961,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.017133956386293e-05,
+      "loss": 3.8403,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0093457943925235e-05,
+      "loss": 3.9684,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.001557632398754e-05,
+      "loss": 4.0612,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9937694704049846e-05,
+      "loss": 4.0636,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.985981308411215e-05,
+      "loss": 4.1004,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9781931464174454e-05,
+      "loss": 4.1802,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.970404984423676e-05,
+      "loss": 4.1397,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9626168224299065e-05,
+      "loss": 3.8772,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.954828660436137e-05,
+      "loss": 4.1737,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9470404984423677e-05,
+      "loss": 4.3925,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9392523364485982e-05,
+      "loss": 4.0429,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9314641744548288e-05,
+      "loss": 4.3418,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9236760124610594e-05,
+      "loss": 4.4488,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9158878504672896e-05,
+      "loss": 3.8899,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9080996884735202e-05,
+      "loss": 4.3971,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9003115264797507e-05,
+      "loss": 4.3679,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925233644859813e-05,
+      "loss": 4.8267,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.884735202492212e-05,
+      "loss": 3.8938,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8769470404984424e-05,
+      "loss": 4.3807,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.869158878504673e-05,
+      "loss": 4.2499,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8613707165109036e-05,
+      "loss": 4.6344,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8535825545171338e-05,
+      "loss": 3.897,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8457943925233644e-05,
+      "loss": 4.4322,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.838006230529595e-05,
+      "loss": 4.294,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8302180685358255e-05,
+      "loss": 4.0467,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.822429906542056e-05,
+      "loss": 3.9704,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8146417445482866e-05,
+      "loss": 4.3733,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8068535825545172e-05,
+      "loss": 4.5383,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7990654205607478e-05,
+      "loss": 4.2512,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7912772585669783e-05,
+      "loss": 4.3976,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7834890965732086e-05,
+      "loss": 4.1185,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.775700934579439e-05,
+      "loss": 4.3184,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7679127725856697e-05,
+      "loss": 4.3355,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7601246105919003e-05,
+      "loss": 4.1799,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.752336448598131e-05,
+      "loss": 4.2742,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7445482866043614e-05,
+      "loss": 4.0244,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.736760124610592e-05,
+      "loss": 4.2809,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7289719626168225e-05,
+      "loss": 3.986,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.721183800623053e-05,
+      "loss": 3.8743,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7133956386292833e-05,
+      "loss": 4.2197,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.705607476635514e-05,
+      "loss": 4.239,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6978193146417445e-05,
+      "loss": 4.0877,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.690031152647975e-05,
+      "loss": 4.0946,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6822429906542056e-05,
+      "loss": 4.3854,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6744548286604362e-05,
+      "loss": 4.2533,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 3.9763,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588785046728973e-05,
+      "loss": 3.8658,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6510903426791275e-05,
+      "loss": 4.2685,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.643302180685358e-05,
+      "loss": 4.0244,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6355140186915887e-05,
+      "loss": 4.2041,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6277258566978192e-05,
+      "loss": 4.2631,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6199376947040498e-05,
+      "loss": 4.3845,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6121495327102804e-05,
+      "loss": 4.2794,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.604361370716511e-05,
+      "loss": 4.2896,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5965732087227415e-05,
+      "loss": 4.0986,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.588785046728972e-05,
+      "loss": 4.0236,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5809968847352023e-05,
+      "loss": 4.0008,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.573208722741433e-05,
+      "loss": 4.1111,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5654205607476634e-05,
+      "loss": 4.0878,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.557632398753894e-05,
+      "loss": 4.2799,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5498442367601246e-05,
+      "loss": 4.164,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.542056074766355e-05,
+      "loss": 3.9739,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5342679127725857e-05,
+      "loss": 4.0746,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5264797507788163e-05,
+      "loss": 4.1098,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5186915887850467e-05,
+      "loss": 3.9665,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5109034267912772e-05,
+      "loss": 4.1269,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5031152647975078e-05,
+      "loss": 4.0904,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4953271028037382e-05,
+      "loss": 4.2817,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4875389408099688e-05,
+      "loss": 4.2481,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4797507788161993e-05,
+      "loss": 4.3851,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4719626168224299e-05,
+      "loss": 5.3225,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4641744548286603e-05,
+      "loss": 4.5594,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4563862928348909e-05,
+      "loss": 4.2251,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4485981308411214e-05,
+      "loss": 3.9512,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.440809968847352e-05,
+      "loss": 4.0625,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4330218068535826e-05,
+      "loss": 4.166,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.425233644859813e-05,
+      "loss": 3.9243,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4174454828660435e-05,
+      "loss": 4.2362,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4096573208722741e-05,
+      "loss": 4.1658,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4018691588785047e-05,
+      "loss": 4.1772,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.394080996884735e-05,
+      "loss": 4.3123,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3862928348909656e-05,
+      "loss": 4.0615,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3785046728971962e-05,
+      "loss": 4.3045,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3707165109034268e-05,
+      "loss": 4.095,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3629283489096573e-05,
+      "loss": 4.2796,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3551401869158877e-05,
+      "loss": 3.9485,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3473520249221183e-05,
+      "loss": 4.1645,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3395638629283489e-05,
+      "loss": 4.0592,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3317757009345794e-05,
+      "loss": 4.2829,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3239875389408098e-05,
+      "loss": 4.0828,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3161993769470404e-05,
+      "loss": 4.1241,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.308411214953271e-05,
+      "loss": 4.1715,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3006230529595015e-05,
+      "loss": 4.3639,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.292834890965732e-05,
+      "loss": 3.869,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2850467289719625e-05,
+      "loss": 3.9788,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.277258566978193e-05,
+      "loss": 3.9047,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2694704049844236e-05,
+      "loss": 4.11,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2616822429906542e-05,
+      "loss": 4.4782,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2538940809968846e-05,
+      "loss": 4.1596,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2461059190031153e-05,
+      "loss": 4.0248,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2383177570093459e-05,
+      "loss": 4.248,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2305295950155763e-05,
+      "loss": 4.1175,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2227414330218069e-05,
+      "loss": 4.1657,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2149532710280374e-05,
+      "loss": 4.1843,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.207165109034268e-05,
+      "loss": 4.116,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1993769470404986e-05,
+      "loss": 4.0794,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.191588785046729e-05,
+      "loss": 4.1596,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1838006230529595e-05,
+      "loss": 4.0767,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1760124610591901e-05,
+      "loss": 4.0131,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1682242990654207e-05,
+      "loss": 4.2488,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.160436137071651e-05,
+      "loss": 3.9894,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1526479750778816e-05,
+      "loss": 4.3284,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1448598130841122e-05,
+      "loss": 4.2368,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1370716510903428e-05,
+      "loss": 4.1453,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1292834890965732e-05,
+      "loss": 4.0778,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1214953271028037e-05,
+      "loss": 4.2252,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1137071651090343e-05,
+      "loss": 4.5659,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1059190031152649e-05,
+      "loss": 4.5888,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0981308411214954e-05,
+      "loss": 3.8661,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0903426791277258e-05,
+      "loss": 5.3569,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0825545171339564e-05,
+      "loss": 4.2981,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.074766355140187e-05,
+      "loss": 3.9199,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0669781931464175e-05,
+      "loss": 4.186,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.059190031152648e-05,
+      "loss": 4.0958,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0514018691588785e-05,
+      "loss": 4.1248,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.043613707165109e-05,
+      "loss": 4.0707,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0358255451713396e-05,
+      "loss": 3.9772,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.02803738317757e-05,
+      "loss": 4.089,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0202492211838006e-05,
+      "loss": 4.333,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0124610591900312e-05,
+      "loss": 4.2365,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0046728971962617e-05,
+      "loss": 3.9973,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.968847352024923e-06,
+      "loss": 4.1452,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.890965732087227e-06,
+      "loss": 4.2778,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813084112149533e-06,
+      "loss": 4.215,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.735202492211838e-06,
+      "loss": 3.8217,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.657320872274144e-06,
+      "loss": 4.2197,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.579439252336448e-06,
+      "loss": 3.958,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.501557632398754e-06,
+      "loss": 4.1512,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.42367601246106e-06,
+      "loss": 4.0015,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.345794392523365e-06,
+      "loss": 3.9713,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.267912772585669e-06,
+      "loss": 4.0309,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.190031152647975e-06,
+      "loss": 4.0808,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.11214953271028e-06,
+      "loss": 4.2054,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.034267912772586e-06,
+      "loss": 4.4306,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.956386292834892e-06,
+      "loss": 4.02,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.878504672897196e-06,
+      "loss": 4.3602,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.800623052959501e-06,
+      "loss": 4.4308,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.722741433021807e-06,
+      "loss": 4.0521,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.644859813084113e-06,
+      "loss": 3.877,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.566978193146417e-06,
+      "loss": 4.1146,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.489096573208722e-06,
+      "loss": 4.3121,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.411214953271028e-06,
+      "loss": 4.4493,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 3.8074,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.255451713395638e-06,
+      "loss": 3.8759,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.177570093457943e-06,
+      "loss": 4.3885,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.099688473520249e-06,
+      "loss": 3.8546,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.021806853582555e-06,
+      "loss": 4.2389,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.94392523364486e-06,
+      "loss": 4.1319,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.866043613707164e-06,
+      "loss": 4.0019,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.78816199376947e-06,
+      "loss": 4.2848,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.710280373831776e-06,
+      "loss": 4.1605,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.632398753894081e-06,
+      "loss": 3.9612,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.554517133956386e-06,
+      "loss": 4.3611,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.476635514018691e-06,
+      "loss": 4.2704,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.398753894080997e-06,
+      "loss": 4.3371,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.3208722741433015e-06,
+      "loss": 4.439,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.242990654205607e-06,
+      "loss": 4.4614,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.165109034267913e-06,
+      "loss": 5.0503,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.087227414330218e-06,
+      "loss": 4.1794,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.009345794392523e-06,
+      "loss": 4.5423,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.931464174454828e-06,
+      "loss": 4.4314,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.853582554517134e-06,
+      "loss": 4.4201,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.775700934579439e-06,
+      "loss": 4.5935,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.697819314641744e-06,
+      "loss": 4.1363,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.619937694704049e-06,
+      "loss": 4.0374,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.542056074766355e-06,
+      "loss": 4.1211,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.46417445482866e-06,
+      "loss": 4.0976,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.386292834890965e-06,
+      "loss": 4.005,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.308411214953271e-06,
+      "loss": 4.0011,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.230529595015577e-06,
+      "loss": 4.2884,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.1526479750778815e-06,
+      "loss": 4.5252,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.074766355140187e-06,
+      "loss": 4.3139,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.996884735202493e-06,
+      "loss": 4.0151,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.919003115264798e-06,
+      "loss": 4.2045,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.841121495327103e-06,
+      "loss": 4.4532,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.763239875389408e-06,
+      "loss": 4.054,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.685358255451714e-06,
+      "loss": 4.1351,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.607476635514019e-06,
+      "loss": 4.36,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.529595015576324e-06,
+      "loss": 4.5952,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.451713395638629e-06,
+      "loss": 4.6016,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.373831775700935e-06,
+      "loss": 3.85,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.29595015576324e-06,
+      "loss": 4.0359,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.218068535825545e-06,
+      "loss": 4.2021,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.14018691588785e-06,
+      "loss": 4.2977,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.062305295950156e-06,
+      "loss": 4.5216,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9844236760124615e-06,
+      "loss": 4.1776,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.906542056074766e-06,
+      "loss": 4.1341,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.828660436137072e-06,
+      "loss": 4.1123,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.750778816199377e-06,
+      "loss": 4.2287,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6728971962616825e-06,
+      "loss": 4.2999,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.595015576323987e-06,
+      "loss": 4.4255,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517133956386293e-06,
+      "loss": 4.216,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.439252336448598e-06,
+      "loss": 3.8769,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3613707165109035e-06,
+      "loss": 4.078,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.283489096573208e-06,
+      "loss": 4.0669,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.205607476635514e-06,
+      "loss": 4.2556,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.127725856697819e-06,
+      "loss": 4.7089,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0498442367601245e-06,
+      "loss": 4.545,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.97196261682243e-06,
+      "loss": 3.9075,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.894080996884735e-06,
+      "loss": 4.1991,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.816199376947041e-06,
+      "loss": 4.0854,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7383177570093455e-06,
+      "loss": 4.0528,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6604361370716508e-06,
+      "loss": 4.2288,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5825545171339564e-06,
+      "loss": 4.1263,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5046728971962617e-06,
+      "loss": 3.996,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.426791277258567e-06,
+      "loss": 4.0549,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.348909657320872e-06,
+      "loss": 4.117,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2710280373831774e-06,
+      "loss": 4.3191,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1931464174454827e-06,
+      "loss": 4.2075,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1152647975077884e-06,
+      "loss": 4.1135,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0373831775700936e-06,
+      "loss": 4.1765,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.959501557632399e-06,
+      "loss": 4.2682,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.881619937694704e-06,
+      "loss": 3.9246,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8037383177570094e-06,
+      "loss": 4.062,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7258566978193146e-06,
+      "loss": 4.2809,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.64797507788162e-06,
+      "loss": 4.2596,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.570093457943925e-06,
+      "loss": 4.2439,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4922118380062308e-06,
+      "loss": 4.0405,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.414330218068536e-06,
+      "loss": 4.0753,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3364485981308413e-06,
+      "loss": 4.0266,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2585669781931465e-06,
+      "loss": 4.3627,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1806853582554518e-06,
+      "loss": 4.0553,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.102803738317757e-06,
+      "loss": 4.412,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0249221183800623e-06,
+      "loss": 3.9167,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9470404984423675e-06,
+      "loss": 4.0662,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8691588785046728e-06,
+      "loss": 4.3687,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7912772585669782e-06,
+      "loss": 3.9177,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7133956386292835e-06,
+      "loss": 4.0368,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6355140186915887e-06,
+      "loss": 4.0202,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5576323987538942e-06,
+      "loss": 4.0052,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4797507788161994e-06,
+      "loss": 4.0511,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4018691588785047e-06,
+      "loss": 4.0027,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.32398753894081e-06,
+      "loss": 4.1855,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2461059190031154e-06,
+      "loss": 4.0499,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1682242990654206e-06,
+      "loss": 4.1041,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0903426791277259e-06,
+      "loss": 4.1928,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0124610591900311e-06,
+      "loss": 4.0084,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.345794392523364e-07,
+      "loss": 4.1502,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.566978193146417e-07,
+      "loss": 4.1795,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.788161993769471e-07,
+      "loss": 4.1516,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.009345794392523e-07,
+      "loss": 4.6246,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.230529595015577e-07,
+      "loss": 4.0178,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.451713395638629e-07,
+      "loss": 4.4176,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.672897196261682e-07,
+      "loss": 3.9512,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8940809968847354e-07,
+      "loss": 4.6122,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1152647975077885e-07,
+      "loss": 3.4805,
+      "step": 1784
+    },
     {
       "epoch": 1.0,
-      "step": 297,
+      "step": 1784,
       "total_flos": 0.0,
-      "train_loss": 4.3407110846805255,
-      "train_runtime": 4813.5869,
-      "train_samples_per_second": 5.929,
-      "train_steps_per_second": 0.062
+      "train_loss": 4.4116124524129345,
+      "train_runtime": 8505.8878,
+      "train_samples_per_second": 3.355,
+      "train_steps_per_second": 0.21
     }
   ],
-  "max_steps": 297,
+  "max_steps": 1784,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,