diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,5383 +2,2698 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 4.7935,
+      "learning_rate": 6e-07,
+      "loss": 4.7769,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 6e-07,
-      "loss": 5.0643,
+      "learning_rate": 1.2e-06,
+      "loss": 4.9164,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.2e-06,
-      "loss": 4.9205,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 4.8809,
       "step": 3
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 1.8e-06,
-      "loss": 4.9075,
+      "loss": 4.8612,
       "step": 4
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.4e-06,
-      "loss": 4.7267,
+      "loss": 4.7997,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.9999999999999997e-06,
-      "loss": 4.8143,
+      "loss": 4.8452,
       "step": 6
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 3.6e-06,
-      "loss": 4.7767,
+      "loss": 4.7936,
       "step": 7
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 4.2e-06,
-      "loss": 4.6701,
+      "loss": 4.8186,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.2e-06,
-      "loss": 4.7857,
+      "epoch": 0.02,
+      "learning_rate": 4.8e-06,
+      "loss": 4.7825,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.8e-06,
-      "loss": 4.6694,
+      "epoch": 0.02,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 4.621,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 5.399999999999999e-06,
-      "loss": 4.7207,
+      "epoch": 0.02,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 4.7091,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 5.999999999999999e-06,
-      "loss": 4.6639,
+      "epoch": 0.03,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 4.6398,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.599999999999999e-06,
-      "loss": 4.6055,
+      "epoch": 0.03,
+      "learning_rate": 7.2e-06,
+      "loss": 4.5605,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.2e-06,
-      "loss": 4.5577,
+      "epoch": 0.03,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 4.5522,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.799999999999998e-06,
-      "loss": 4.4804,
+      "epoch": 0.03,
+      "learning_rate": 8.4e-06,
+      "loss": 4.4202,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.4e-06,
-      "loss": 4.5974,
+      "epoch": 0.04,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 4.5517,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.999999999999999e-06,
-      "loss": 4.5684,
+      "epoch": 0.04,
+      "learning_rate": 9.6e-06,
+      "loss": 4.4522,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 9.6e-06,
-      "loss": 4.678,
+      "epoch": 0.04,
+      "learning_rate": 1.02e-05,
+      "loss": 4.5369,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.02e-05,
-      "loss": 4.3706,
+      "epoch": 0.04,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 4.3402,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.0799999999999998e-05,
-      "loss": 4.5311,
+      "epoch": 0.04,
+      "learning_rate": 1.14e-05,
+      "loss": 4.4829,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.14e-05,
-      "loss": 4.4249,
+      "epoch": 0.05,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 4.4559,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.1999999999999999e-05,
-      "loss": 4.415,
+      "epoch": 0.05,
+      "learning_rate": 1.26e-05,
+      "loss": 4.4034,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.26e-05,
-      "loss": 4.433,
+      "epoch": 0.05,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 4.4785,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.3199999999999997e-05,
-      "loss": 4.5236,
+      "epoch": 0.05,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 4.4421,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.3799999999999998e-05,
-      "loss": 4.4252,
+      "epoch": 0.06,
+      "learning_rate": 1.44e-05,
+      "loss": 4.2612,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.44e-05,
-      "loss": 4.5983,
+      "epoch": 0.06,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 4.515,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.4999999999999999e-05,
-      "loss": 4.3082,
+      "epoch": 0.06,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 4.382,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.5599999999999996e-05,
-      "loss": 4.2028,
+      "epoch": 0.06,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 4.217,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.6199999999999997e-05,
-      "loss": 4.2437,
+      "epoch": 0.07,
+      "learning_rate": 1.68e-05,
+      "loss": 4.1962,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.68e-05,
-      "loss": 4.45,
+      "epoch": 0.07,
+      "learning_rate": 1.74e-05,
+      "loss": 4.4288,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.74e-05,
-      "loss": 4.4441,
+      "epoch": 0.07,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 4.3932,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.7999999999999997e-05,
-      "loss": 4.2113,
+      "epoch": 0.07,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 4.2549,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.8599999999999998e-05,
-      "loss": 4.3137,
+      "epoch": 0.07,
+      "learning_rate": 1.92e-05,
+      "loss": 4.3319,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.92e-05,
-      "loss": 4.2111,
+      "epoch": 0.08,
+      "learning_rate": 1.98e-05,
+      "loss": 4.2797,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.98e-05,
-      "loss": 4.4028,
+      "epoch": 0.08,
+      "learning_rate": 2.04e-05,
+      "loss": 4.3458,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.04e-05,
-      "loss": 4.3463,
+      "epoch": 0.08,
+      "learning_rate": 2.1e-05,
+      "loss": 4.3281,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.1e-05,
-      "loss": 4.3769,
+      "epoch": 0.08,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 4.3676,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.1599999999999996e-05,
-      "loss": 4.3872,
+      "epoch": 0.09,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 4.304,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.2199999999999998e-05,
-      "loss": 4.5496,
+      "epoch": 0.09,
+      "learning_rate": 2.28e-05,
+      "loss": 4.4121,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.28e-05,
-      "loss": 4.3342,
+      "epoch": 0.09,
+      "learning_rate": 2.34e-05,
+      "loss": 4.2497,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.34e-05,
-      "loss": 4.2224,
+      "epoch": 0.09,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 4.3074,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.3999999999999997e-05,
-      "loss": 4.3825,
+      "epoch": 0.09,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 4.388,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.4599999999999998e-05,
-      "loss": 4.1646,
+      "epoch": 0.1,
+      "learning_rate": 2.52e-05,
+      "loss": 4.1538,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.52e-05,
-      "loss": 4.3405,
+      "epoch": 0.1,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 4.3297,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.5799999999999997e-05,
-      "loss": 4.3905,
+      "epoch": 0.1,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 4.4247,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.6399999999999995e-05,
-      "loss": 4.5426,
+      "epoch": 0.1,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 4.4755,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.6999999999999996e-05,
-      "loss": 4.4031,
+      "epoch": 0.11,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 4.3917,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.7599999999999997e-05,
-      "loss": 4.7509,
+      "epoch": 0.11,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.617,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.9197,
+      "epoch": 0.11,
+      "learning_rate": 2.88e-05,
+      "loss": 4.6925,
       "step": 49
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 2.88e-05,
-      "loss": 4.562,
+      "loss": 4.4051,
       "step": 50
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 2.94e-05,
-      "loss": 4.1801,
+      "loss": 4.2783,
       "step": 51
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 2.9999999999999997e-05,
-      "loss": 4.3261,
+      "loss": 4.1599,
       "step": 52
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 3.06e-05,
-      "loss": 4.1817,
+      "loss": 4.3254,
       "step": 53
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 3.119999999999999e-05,
-      "loss": 4.2294,
+      "loss": 4.2387,
       "step": 54
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 3.1799999999999994e-05,
-      "loss": 4.2412,
+      "loss": 4.0646,
       "step": 55
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.13,
       "learning_rate": 3.2399999999999995e-05,
-      "loss": 4.319,
+      "loss": 4.265,
       "step": 56
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.13,
       "learning_rate": 3.2999999999999996e-05,
-      "loss": 4.3087,
+      "loss": 4.286,
       "step": 57
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 3.36e-05,
-      "loss": 4.5114,
+      "loss": 4.3781,
       "step": 58
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 3.42e-05,
-      "loss": 4.2629,
+      "loss": 4.295,
       "step": 59
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 3.48e-05,
-      "loss": 4.136,
+      "loss": 4.1682,
       "step": 60
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 3.539999999999999e-05,
-      "loss": 4.3773,
+      "loss": 4.2437,
       "step": 61
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 3.5999999999999994e-05,
-      "loss": 4.3272,
+      "loss": 4.188,
       "step": 62
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 3.6599999999999995e-05,
-      "loss": 4.1657,
+      "loss": 4.1751,
       "step": 63
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 3.7199999999999996e-05,
-      "loss": 4.321,
+      "loss": 4.2774,
       "step": 64
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 3.78e-05,
-      "loss": 4.1184,
+      "loss": 4.2418,
       "step": 65
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 3.84e-05,
-      "loss": 4.348,
+      "loss": 4.3547,
       "step": 66
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 3.9e-05,
-      "loss": 4.1546,
+      "loss": 4.3256,
       "step": 67
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 3.96e-05,
-      "loss": 4.1783,
+      "loss": 4.1705,
       "step": 68
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 4.02e-05,
-      "loss": 4.1314,
+      "loss": 4.2013,
       "step": 69
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 4.08e-05,
-      "loss": 4.2552,
+      "loss": 4.1742,
       "step": 70
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 4.14e-05,
-      "loss": 4.4319,
+      "loss": 4.2953,
       "step": 71
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 4.2e-05,
-      "loss": 4.3253,
+      "loss": 4.3439,
       "step": 72
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 4.259999999999999e-05,
-      "loss": 4.2573,
+      "loss": 4.1583,
       "step": 73
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 4.319999999999999e-05,
-      "loss": 4.2875,
+      "loss": 4.2408,
       "step": 74
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 4.3799999999999994e-05,
-      "loss": 4.1211,
+      "loss": 4.2264,
       "step": 75
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 4.4399999999999995e-05,
-      "loss": 4.3568,
+      "loss": 4.2005,
       "step": 76
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 4.4999999999999996e-05,
-      "loss": 4.2824,
+      "loss": 4.2334,
       "step": 77
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 4.56e-05,
-      "loss": 4.2062,
+      "loss": 4.1885,
       "step": 78
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 4.62e-05,
-      "loss": 4.0674,
+      "loss": 4.1813,
       "step": 79
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 4.68e-05,
-      "loss": 4.3285,
+      "loss": 4.3409,
       "step": 80
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 4.7399999999999993e-05,
-      "loss": 4.3773,
+      "loss": 4.262,
       "step": 81
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 4.7999999999999994e-05,
-      "loss": 4.1549,
+      "loss": 4.2151,
       "step": 82
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 4.8599999999999995e-05,
-      "loss": 4.4001,
+      "loss": 4.321,
       "step": 83
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 4.9199999999999997e-05,
-      "loss": 4.3308,
+      "loss": 4.1941,
       "step": 84
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.19,
       "learning_rate": 4.98e-05,
-      "loss": 4.2605,
+      "loss": 4.2677,
       "step": 85
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.19,
       "learning_rate": 5.04e-05,
-      "loss": 4.2488,
+      "loss": 4.2252,
       "step": 86
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 5.1e-05,
-      "loss": 4.255,
+      "loss": 4.2876,
       "step": 87
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 5.1599999999999994e-05,
-      "loss": 4.2311,
+      "loss": 4.1668,
       "step": 88
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 5.2199999999999995e-05,
-      "loss": 4.3115,
+      "loss": 4.3233,
       "step": 89
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 5.279999999999999e-05,
-      "loss": 4.246,
+      "loss": 4.3072,
       "step": 90
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 5.339999999999999e-05,
-      "loss": 4.4423,
+      "loss": 4.1785,
       "step": 91
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.21,
       "learning_rate": 5.399999999999999e-05,
-      "loss": 4.389,
+      "loss": 4.428,
       "step": 92
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.21,
       "learning_rate": 5.459999999999999e-05,
-      "loss": 4.113,
+      "loss": 4.3544,
       "step": 93
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.21,
       "learning_rate": 5.519999999999999e-05,
-      "loss": 4.3143,
+      "loss": 4.3594,
       "step": 94
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.21,
       "learning_rate": 5.5799999999999994e-05,
-      "loss": 4.6075,
+      "loss": 4.2516,
       "step": 95
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.4951,
+      "loss": 4.4909,
       "step": 96
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.319,
+      "loss": 4.3805,
       "step": 97
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 5.76e-05,
-      "loss": 4.5888,
+      "loss": 4.2646,
       "step": 98
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 5.82e-05,
-      "loss": 4.3275,
+      "loss": 4.6111,
       "step": 99
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 5.88e-05,
-      "loss": 4.3674,
+      "loss": 4.4095,
       "step": 100
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.23,
       "learning_rate": 5.94e-05,
-      "loss": 4.2166,
+      "loss": 4.2562,
       "step": 101
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.23,
       "learning_rate": 5.9999999999999995e-05,
-      "loss": 4.262,
+      "loss": 4.1922,
       "step": 102
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.23,
       "learning_rate": 6.0599999999999996e-05,
-      "loss": 4.342,
+      "loss": 4.0757,
       "step": 103
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.23,
       "learning_rate": 6.12e-05,
-      "loss": 4.2719,
+      "loss": 4.1528,
       "step": 104
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 6.18e-05,
-      "loss": 4.1079,
+      "loss": 4.0813,
       "step": 105
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 6.239999999999999e-05,
-      "loss": 4.3223,
+      "loss": 4.1241,
       "step": 106
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 6.299999999999999e-05,
-      "loss": 4.3472,
+      "loss": 4.183,
       "step": 107
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 6.359999999999999e-05,
-      "loss": 4.4634,
+      "loss": 4.0929,
       "step": 108
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 6.419999999999999e-05,
-      "loss": 4.0303,
+      "loss": 4.1176,
       "step": 109
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.25,
       "learning_rate": 6.479999999999999e-05,
-      "loss": 4.1659,
+      "loss": 4.1593,
       "step": 110
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.25,
       "learning_rate": 6.539999999999999e-05,
-      "loss": 4.3407,
+      "loss": 4.1886,
       "step": 111
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.25,
       "learning_rate": 6.599999999999999e-05,
-      "loss": 4.2267,
+      "loss": 4.1987,
       "step": 112
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.25,
       "learning_rate": 6.659999999999999e-05,
-      "loss": 4.1988,
+      "loss": 4.1617,
       "step": 113
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 6.72e-05,
-      "loss": 4.1583,
+      "loss": 4.1273,
       "step": 114
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 6.78e-05,
-      "loss": 4.2206,
+      "loss": 4.1707,
       "step": 115
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 6.84e-05,
-      "loss": 3.9665,
+      "loss": 4.2915,
       "step": 116
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 6.9e-05,
-      "loss": 4.2874,
+      "loss": 4.1509,
       "step": 117
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 6.96e-05,
-      "loss": 4.1614,
+      "loss": 4.2995,
       "step": 118
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.27,
       "learning_rate": 7.02e-05,
-      "loss": 4.278,
+      "loss": 4.2146,
       "step": 119
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.27,
       "learning_rate": 7.079999999999999e-05,
-      "loss": 4.33,
+      "loss": 4.2528,
       "step": 120
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.27,
       "learning_rate": 7.139999999999999e-05,
-      "loss": 4.2586,
+      "loss": 4.0505,
       "step": 121
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.27,
       "learning_rate": 7.199999999999999e-05,
-      "loss": 4.2284,
+      "loss": 4.1719,
       "step": 122
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 7.259999999999999e-05,
-      "loss": 4.0917,
+      "loss": 4.1813,
       "step": 123
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 7.319999999999999e-05,
-      "loss": 4.3557,
+      "loss": 4.1625,
       "step": 124
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 7.379999999999999e-05,
-      "loss": 4.2795,
+      "loss": 4.2766,
       "step": 125
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 7.439999999999999e-05,
-      "loss": 4.3276,
+      "loss": 4.2836,
       "step": 126
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 7.5e-05,
-      "loss": 4.3446,
+      "loss": 4.1918,
       "step": 127
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.29,
       "learning_rate": 7.56e-05,
-      "loss": 4.2295,
+      "loss": 4.1589,
       "step": 128
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.29,
       "learning_rate": 7.62e-05,
-      "loss": 4.1818,
+      "loss": 4.085,
       "step": 129
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 7.68e-05,
-      "loss": 4.3607,
+      "loss": 4.2634,
       "step": 130
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 7.74e-05,
-      "loss": 4.2731,
+      "loss": 4.1471,
       "step": 131
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.8e-05,
-      "loss": 4.3785,
+      "loss": 4.2167,
       "step": 132
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.86e-05,
-      "loss": 4.3502,
+      "loss": 4.1086,
       "step": 133
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.92e-05,
-      "loss": 4.3146,
+      "loss": 4.2122,
       "step": 134
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.98e-05,
-      "loss": 4.1596,
+      "loss": 4.3218,
       "step": 135
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 8.04e-05,
-      "loss": 4.2668,
+      "loss": 4.1705,
       "step": 136
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 8.1e-05,
-      "loss": 4.5135,
+      "loss": 4.2621,
       "step": 137
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 8.16e-05,
-      "loss": 4.4041,
+      "loss": 4.2883,
       "step": 138
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 8.22e-05,
-      "loss": 4.3569,
+      "loss": 4.3078,
       "step": 139
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 8.28e-05,
-      "loss": 4.1605,
+      "loss": 4.2738,
       "step": 140
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.34e-05,
-      "loss": 4.4837,
+      "loss": 4.3946,
       "step": 141
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.4e-05,
-      "loss": 4.5306,
+      "loss": 4.3515,
       "step": 142
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.459999999999998e-05,
-      "loss": 4.3956,
+      "loss": 4.3548,
       "step": 143
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.519999999999998e-05,
-      "loss": 4.3278,
+      "loss": 4.348,
       "step": 144
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 8.579999999999998e-05,
-      "loss": 4.6011,
+      "loss": 4.2612,
       "step": 145
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 8.639999999999999e-05,
-      "loss": 4.4849,
+      "loss": 4.3299,
       "step": 146
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 8.699999999999999e-05,
-      "loss": 4.5009,
+      "loss": 4.6279,
       "step": 147
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 8.759999999999999e-05,
-      "loss": 4.5048,
+      "loss": 4.6792,
       "step": 148
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 8.819999999999999e-05,
-      "loss": 4.6745,
+      "loss": 4.537,
       "step": 149
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.879999999999999e-05,
-      "loss": 4.407,
+      "loss": 4.5154,
       "step": 150
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.939999999999999e-05,
-      "loss": 4.4633,
+      "loss": 4.1244,
       "step": 151
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.999999999999999e-05,
-      "loss": 4.321,
+      "loss": 4.2241,
       "step": 152
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 9.059999999999999e-05,
-      "loss": 4.4206,
+      "loss": 4.2773,
       "step": 153
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 9.12e-05,
-      "loss": 4.1797,
+      "loss": 4.0853,
       "step": 154
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 9.18e-05,
-      "loss": 4.1876,
+      "loss": 4.2703,
       "step": 155
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 9.24e-05,
-      "loss": 4.2306,
+      "loss": 4.0987,
       "step": 156
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 9.3e-05,
-      "loss": 4.4635,
+      "loss": 4.1206,
       "step": 157
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 9.36e-05,
-      "loss": 4.4991,
+      "loss": 4.2341,
       "step": 158
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.419999999999999e-05,
-      "loss": 4.5696,
+      "loss": 4.2542,
       "step": 159
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.479999999999999e-05,
-      "loss": 4.2998,
+      "loss": 4.2671,
       "step": 160
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.539999999999999e-05,
-      "loss": 4.334,
+      "loss": 4.298,
       "step": 161
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.599999999999999e-05,
-      "loss": 4.2896,
+      "loss": 4.2375,
       "step": 162
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 9.659999999999999e-05,
-      "loss": 4.2937,
+      "loss": 4.2119,
       "step": 163
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 9.719999999999999e-05,
-      "loss": 4.4709,
+      "loss": 4.271,
       "step": 164
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 9.779999999999999e-05,
-      "loss": 4.3272,
+      "loss": 4.1551,
       "step": 165
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 9.839999999999999e-05,
-      "loss": 4.3714,
+      "loss": 4.1953,
       "step": 166
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 9.9e-05,
-      "loss": 4.3042,
+      "loss": 4.0992,
       "step": 167
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 9.96e-05,
-      "loss": 4.1345,
+      "loss": 4.1463,
       "step": 168
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 0.0001002,
-      "loss": 4.2959,
+      "loss": 4.367,
       "step": 169
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 0.0001008,
-      "loss": 4.2594,
+      "loss": 4.3162,
       "step": 170
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 0.0001014,
-      "loss": 4.4797,
+      "loss": 4.2243,
       "step": 171
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 0.000102,
-      "loss": 4.5183,
+      "loss": 4.3289,
       "step": 172
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 0.0001026,
-      "loss": 4.3362,
+      "loss": 4.1431,
       "step": 173
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 0.00010319999999999999,
-      "loss": 4.2346,
+      "loss": 4.3012,
       "step": 174
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 0.00010379999999999999,
-      "loss": 4.3054,
+      "loss": 4.3199,
       "step": 175
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 0.00010439999999999999,
-      "loss": 4.1806,
+      "loss": 4.1572,
       "step": 176
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010499999999999999,
-      "loss": 4.2475,
+      "loss": 4.1236,
       "step": 177
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010559999999999998,
-      "loss": 4.2608,
+      "loss": 4.2767,
       "step": 178
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010619999999999998,
-      "loss": 4.432,
+      "loss": 4.3445,
       "step": 179
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010679999999999998,
-      "loss": 4.3326,
+      "loss": 4.2821,
       "step": 180
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 0.00010739999999999998,
-      "loss": 4.3451,
+      "loss": 4.228,
       "step": 181
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 0.00010799999999999998,
-      "loss": 4.2187,
+      "loss": 4.2785,
       "step": 182
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 0.00010859999999999998,
-      "loss": 4.3603,
+      "loss": 4.3994,
       "step": 183
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 0.00010919999999999998,
-      "loss": 4.4496,
+      "loss": 4.2999,
       "step": 184
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 0.00010979999999999999,
-      "loss": 4.3975,
+      "loss": 4.2597,
       "step": 185
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00011039999999999999,
-      "loss": 4.2659,
+      "loss": 4.1352,
       "step": 186
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00011099999999999999,
-      "loss": 4.4568,
+      "loss": 4.1805,
       "step": 187
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00011159999999999999,
-      "loss": 4.151,
+      "loss": 4.127,
       "step": 188
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00011219999999999999,
-      "loss": 4.2923,
+      "loss": 4.31,
       "step": 189
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 0.00011279999999999999,
-      "loss": 4.4521,
+      "loss": 4.3199,
       "step": 190
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 0.00011339999999999999,
-      "loss": 4.4017,
+      "loss": 4.2928,
       "step": 191
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 0.00011399999999999999,
-      "loss": 4.4227,
+      "loss": 4.2784,
       "step": 192
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 0.0001146,
-      "loss": 4.6037,
+      "loss": 4.2223,
       "step": 193
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 0.0001152,
-      "loss": 4.2636,
+      "loss": 4.3253,
       "step": 194
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.0001158,
-      "loss": 4.7476,
+      "loss": 4.4979,
       "step": 195
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.0001164,
-      "loss": 4.7315,
+      "loss": 4.3866,
       "step": 196
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.000117,
-      "loss": 4.6674,
+      "loss": 4.5051,
       "step": 197
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.0001176,
-      "loss": 5.1666,
+      "loss": 4.5433,
       "step": 198
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 0.0001182,
-      "loss": 5.1316,
+      "loss": 4.6245,
       "step": 199
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 0.0001188,
-      "loss": 4.7765,
+      "loss": 4.5374,
       "step": 200
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 0.0001194,
-      "loss": 4.3566,
+      "loss": 4.3464,
       "step": 201
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 0.00011999999999999999,
-      "loss": 4.4518,
+      "loss": 4.3094,
       "step": 202
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00012059999999999999,
-      "loss": 4.4986,
+      "loss": 4.3137,
       "step": 203
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00012119999999999999,
-      "loss": 4.3431,
+      "loss": 4.2578,
       "step": 204
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00012179999999999999,
-      "loss": 4.238,
+      "loss": 4.3459,
       "step": 205
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.0001224,
-      "loss": 4.2508,
+      "loss": 4.3108,
       "step": 206
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00012299999999999998,
-      "loss": 4.3551,
+      "loss": 4.144,
       "step": 207
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 0.0001236,
-      "loss": 4.1455,
+      "loss": 4.3351,
       "step": 208
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 0.00012419999999999998,
-      "loss": 4.3884,
+      "loss": 4.3923,
       "step": 209
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 0.00012479999999999997,
-      "loss": 4.3141,
+      "loss": 4.367,
       "step": 210
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 0.00012539999999999999,
-      "loss": 4.4313,
+      "loss": 4.1492,
       "step": 211
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.00012599999999999997,
-      "loss": 4.4131,
+      "loss": 4.2222,
       "step": 212
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.0001266,
-      "loss": 4.3574,
+      "loss": 4.1713,
       "step": 213
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.00012719999999999997,
-      "loss": 4.5215,
+      "loss": 4.1952,
       "step": 214
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.0001278,
-      "loss": 4.1939,
+      "loss": 4.2625,
       "step": 215
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.00012839999999999998,
-      "loss": 4.3366,
+      "loss": 4.2425,
       "step": 216
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 0.000129,
-      "loss": 4.4412,
+      "loss": 4.2802,
       "step": 217
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 0.00012959999999999998,
-      "loss": 4.3818,
+      "loss": 4.2234,
       "step": 218
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 0.0001302,
-      "loss": 4.4043,
+      "loss": 4.303,
       "step": 219
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 0.00013079999999999998,
-      "loss": 4.4044,
+      "loss": 4.2587,
       "step": 220
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.0001314,
-      "loss": 4.208,
+      "loss": 4.2141,
       "step": 221
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.00013199999999999998,
-      "loss": 4.3869,
+      "loss": 4.2823,
       "step": 222
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.0001326,
-      "loss": 4.4836,
+      "loss": 4.3153,
       "step": 223
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.00013319999999999999,
-      "loss": 4.367,
+      "loss": 4.2467,
       "step": 224
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.0001338,
-      "loss": 4.63,
+      "loss": 4.3831,
       "step": 225
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 0.0001344,
-      "loss": 4.5293,
+      "loss": 4.2606,
       "step": 226
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 0.000135,
-      "loss": 4.487,
+      "loss": 4.4986,
       "step": 227
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 0.0001356,
-      "loss": 4.4011,
+      "loss": 4.3744,
       "step": 228
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 0.0001362,
-      "loss": 4.2859,
+      "loss": 4.245,
       "step": 229
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001368,
-      "loss": 4.4017,
+      "loss": 4.4023,
       "step": 230
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001374,
-      "loss": 4.6719,
+      "loss": 4.3228,
       "step": 231
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.000138,
-      "loss": 4.5177,
+      "loss": 4.1694,
       "step": 232
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001386,
-      "loss": 4.3213,
+      "loss": 4.4445,
       "step": 233
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001392,
-      "loss": 4.5602,
+      "loss": 4.3381,
       "step": 234
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 0.00013979999999999998,
-      "loss": 4.5167,
+      "loss": 4.4417,
       "step": 235
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 0.0001404,
-      "loss": 4.3494,
+      "loss": 4.2774,
       "step": 236
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 0.00014099999999999998,
-      "loss": 4.2911,
+      "loss": 4.3152,
       "step": 237
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 0.00014159999999999997,
-      "loss": 4.6164,
+      "loss": 4.2731,
       "step": 238
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.0001422,
-      "loss": 4.3868,
+      "loss": 4.2984,
       "step": 239
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.00014279999999999997,
-      "loss": 4.4622,
+      "loss": 4.4744,
       "step": 240
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.0001434,
-      "loss": 4.5795,
+      "loss": 4.2872,
       "step": 241
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.00014399999999999998,
-      "loss": 4.6158,
+      "loss": 4.3544,
       "step": 242
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.0001446,
-      "loss": 4.786,
+      "loss": 4.5594,
       "step": 243
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 0.00014519999999999998,
-      "loss": 5.3147,
+      "loss": 4.4725,
       "step": 244
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 0.0001458,
-      "loss": 4.7131,
+      "loss": 4.3135,
       "step": 245
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 0.00014639999999999998,
-      "loss": 4.5343,
+      "loss": 4.1827,
       "step": 246
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 0.000147,
-      "loss": 5.0141,
+      "loss": 4.0959,
       "step": 247
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.00014759999999999998,
-      "loss": 4.8573,
+      "loss": 4.9479,
       "step": 248
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.0001482,
-      "loss": 5.0538,
+      "loss": 4.9316,
       "step": 249
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.00014879999999999998,
-      "loss": 4.4255,
+      "loss": 4.9687,
       "step": 250
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.0001494,
-      "loss": 4.7096,
+      "loss": 4.4753,
       "step": 251
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 0.00015,
-      "loss": 4.4132,
+      "loss": 4.3512,
       "step": 252
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 0.00015059999999999997,
-      "loss": 4.5483,
+      "loss": 4.4217,
       "step": 253
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 0.0001512,
-      "loss": 4.5101,
+      "loss": 4.2391,
       "step": 254
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 0.00015179999999999998,
-      "loss": 4.4749,
+      "loss": 4.3475,
       "step": 255
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 0.0001524,
-      "loss": 4.4696,
+      "loss": 4.251,
       "step": 256
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.00015299999999999998,
-      "loss": 4.5023,
+      "loss": 4.2616,
       "step": 257
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.0001536,
-      "loss": 4.2994,
+      "loss": 4.3103,
       "step": 258
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.00015419999999999998,
-      "loss": 4.4836,
+      "loss": 4.3225,
       "step": 259
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.0001548,
-      "loss": 4.4787,
+      "loss": 4.573,
       "step": 260
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 0.00015539999999999998,
-      "loss": 4.5402,
+      "loss": 4.5106,
       "step": 261
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 0.000156,
-      "loss": 4.4271,
+      "loss": 4.28,
       "step": 262
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 0.00015659999999999998,
-      "loss": 4.3551,
+      "loss": 4.5133,
       "step": 263
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 0.0001572,
-      "loss": 4.356,
+      "loss": 4.3204,
       "step": 264
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 0.0001578,
-      "loss": 4.6543,
+      "loss": 4.4585,
       "step": 265
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.0001584,
-      "loss": 4.7402,
+      "loss": 4.3421,
       "step": 266
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.000159,
-      "loss": 4.3412,
+      "loss": 4.3581,
       "step": 267
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.0001596,
-      "loss": 4.7831,
+      "loss": 4.295,
       "step": 268
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.0001602,
-      "loss": 4.3223,
+      "loss": 4.5293,
       "step": 269
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 0.0001608,
-      "loss": 4.7522,
+      "loss": 4.445,
       "step": 270
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 0.0001614,
-      "loss": 4.4004,
+      "loss": 4.2543,
       "step": 271
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 0.000162,
-      "loss": 4.3634,
+      "loss": 4.3813,
       "step": 272
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 0.0001626,
-      "loss": 4.5498,
+      "loss": 4.3964,
       "step": 273
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 0.0001632,
-      "loss": 4.2597,
+      "loss": 4.4097,
       "step": 274
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.0001638,
-      "loss": 4.6339,
+      "loss": 4.4565,
       "step": 275
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.0001644,
-      "loss": 4.6287,
+      "loss": 4.2819,
       "step": 276
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.000165,
-      "loss": 4.4704,
+      "loss": 4.302,
       "step": 277
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.0001656,
-      "loss": 4.2664,
+      "loss": 4.3931,
       "step": 278
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 0.0001662,
-      "loss": 4.3407,
+      "loss": 4.3709,
       "step": 279
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 0.0001668,
-      "loss": 4.609,
+      "loss": 4.4844,
       "step": 280
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 0.0001674,
-      "loss": 4.3177,
+      "loss": 4.4591,
       "step": 281
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 0.000168,
-      "loss": 4.4474,
+      "loss": 4.3974,
       "step": 282
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 0.0001686,
-      "loss": 4.1915,
+      "loss": 4.2969,
       "step": 283
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.00016919999999999997,
-      "loss": 4.472,
+      "loss": 4.3746,
       "step": 284
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.00016979999999999998,
-      "loss": 4.5332,
+      "loss": 4.5457,
       "step": 285
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.00017039999999999997,
-      "loss": 4.4325,
+      "loss": 4.3635,
       "step": 286
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.00017099999999999998,
-      "loss": 4.6275,
+      "loss": 4.5762,
       "step": 287
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 0.00017159999999999997,
-      "loss": 4.6464,
+      "loss": 4.4033,
       "step": 288
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 0.00017219999999999998,
-      "loss": 4.6767,
+      "loss": 4.3375,
       "step": 289
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 0.00017279999999999997,
-      "loss": 4.5076,
+      "loss": 4.3955,
       "step": 290
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 0.00017339999999999996,
-      "loss": 4.5415,
+      "loss": 4.7462,
       "step": 291
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 0.00017399999999999997,
-      "loss": 4.346,
+      "loss": 4.4709,
       "step": 292
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017459999999999996,
-      "loss": 5.3879,
+      "loss": 4.5451,
       "step": 293
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017519999999999998,
-      "loss": 4.4703,
+      "loss": 4.6423,
       "step": 294
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017579999999999996,
-      "loss": 4.5646,
+      "loss": 4.4917,
       "step": 295
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017639999999999998,
-      "loss": 5.0213,
+      "loss": 4.6416,
       "step": 296
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 0.00017699999999999997,
-      "loss": 4.9791,
+      "loss": 4.6124,
       "step": 297
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 0.00017759999999999998,
-      "loss": 4.6661,
+      "loss": 5.0896,
       "step": 298
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 0.00017819999999999997,
-      "loss": 5.1077,
+      "loss": 4.4072,
       "step": 299
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 0.00017879999999999998,
-      "loss": 4.7887,
+      "loss": 5.0349,
       "step": 300
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 0.00017939999999999997,
-      "loss": 4.2862,
+      "loss": 4.4267,
       "step": 301
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00017999999999999998,
-      "loss": 4.3871,
+      "loss": 4.4898,
       "step": 302
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00018059999999999997,
-      "loss": 4.4105,
+      "loss": 4.6272,
       "step": 303
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00018119999999999999,
-      "loss": 4.4237,
+      "loss": 4.4936,
       "step": 304
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00018179999999999997,
-      "loss": 4.2553,
+      "loss": 4.387,
       "step": 305
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 0.0001824,
-      "loss": 4.4235,
+      "loss": 4.3621,
       "step": 306
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 0.00018299999999999998,
-      "loss": 4.297,
+      "loss": 4.5623,
       "step": 307
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 0.0001836,
-      "loss": 4.7458,
+      "loss": 4.4278,
       "step": 308
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 0.00018419999999999998,
-      "loss": 4.5352,
+      "loss": 4.3864,
       "step": 309
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.0001848,
-      "loss": 4.5423,
+      "loss": 4.505,
       "step": 310
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.00018539999999999998,
-      "loss": 4.663,
+      "loss": 4.2331,
       "step": 311
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.000186,
-      "loss": 4.4946,
+      "loss": 4.5312,
       "step": 312
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.00018659999999999998,
-      "loss": 4.437,
+      "loss": 4.6578,
       "step": 313
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.0001872,
-      "loss": 4.5147,
+      "loss": 4.3891,
       "step": 314
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 0.00018779999999999998,
-      "loss": 4.3544,
+      "loss": 4.3873,
       "step": 315
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 0.00018839999999999997,
-      "loss": 4.6494,
+      "loss": 4.2965,
       "step": 316
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 0.00018899999999999999,
-      "loss": 4.3586,
+      "loss": 4.3478,
       "step": 317
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 0.00018959999999999997,
-      "loss": 4.5048,
+      "loss": 4.3313,
       "step": 318
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.0001902,
-      "loss": 4.633,
+      "loss": 4.3946,
       "step": 319
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.00019079999999999998,
-      "loss": 4.3602,
+      "loss": 4.4888,
       "step": 320
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.0001914,
-      "loss": 4.6183,
+      "loss": 4.3891,
       "step": 321
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.00019199999999999998,
-      "loss": 4.3107,
+      "loss": 4.1888,
       "step": 322
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.0001926,
-      "loss": 4.573,
+      "loss": 4.4259,
       "step": 323
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 0.00019319999999999998,
-      "loss": 4.8246,
+      "loss": 4.2324,
       "step": 324
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 0.0001938,
-      "loss": 4.7945,
+      "loss": 4.5447,
       "step": 325
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 0.00019439999999999998,
-      "loss": 4.5823,
+      "loss": 4.5354,
       "step": 326
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 0.000195,
-      "loss": 4.2816,
+      "loss": 4.244,
       "step": 327
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.00019559999999999998,
-      "loss": 4.3848,
+      "loss": 4.1891,
       "step": 328
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.0001962,
-      "loss": 4.911,
+      "loss": 4.3217,
       "step": 329
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.00019679999999999999,
-      "loss": 4.8547,
+      "loss": 4.4298,
       "step": 330
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.0001974,
-      "loss": 4.4584,
+      "loss": 4.3572,
       "step": 331
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.000198,
-      "loss": 4.5381,
+      "loss": 4.3305,
       "step": 332
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 0.0001986,
-      "loss": 4.6458,
+      "loss": 4.4236,
       "step": 333
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 0.0001992,
-      "loss": 4.7019,
+      "loss": 4.5337,
       "step": 334
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 0.0001998,
-      "loss": 4.6578,
+      "loss": 4.6162,
       "step": 335
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 0.0002004,
-      "loss": 4.4093,
+      "loss": 4.3679,
       "step": 336
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.000201,
-      "loss": 4.3511,
+      "loss": 4.3996,
       "step": 337
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.0002016,
-      "loss": 4.5368,
+      "loss": 4.3851,
       "step": 338
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.0002022,
-      "loss": 4.4722,
+      "loss": 4.6605,
       "step": 339
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.0002028,
-      "loss": 4.5231,
+      "loss": 4.4989,
       "step": 340
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.00020339999999999998,
-      "loss": 5.0982,
+      "loss": 4.2929,
       "step": 341
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 0.000204,
-      "loss": 4.6772,
+      "loss": 4.7402,
       "step": 342
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 0.00020459999999999999,
-      "loss": 4.6195,
+      "loss": 4.5554,
       "step": 343
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 0.0002052,
-      "loss": 4.5202,
+      "loss": 4.5082,
       "step": 344
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 0.0002058,
-      "loss": 4.5568,
+      "loss": 4.4385,
       "step": 345
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020639999999999998,
-      "loss": 4.7313,
+      "loss": 4.399,
       "step": 346
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020699999999999996,
-      "loss": 5.9084,
+      "loss": 4.643,
       "step": 347
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020759999999999998,
-      "loss": 4.6661,
+      "loss": 4.6824,
       "step": 348
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020819999999999996,
-      "loss": 4.9368,
+      "loss": 4.7186,
       "step": 349
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020879999999999998,
-      "loss": 5.0985,
+      "loss": 4.4537,
       "step": 350
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 0.00020939999999999997,
-      "loss": 4.8769,
+      "loss": 4.5032,
       "step": 351
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 0.00020999999999999998,
-      "loss": 4.4071,
+      "loss": 4.4422,
       "step": 352
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 0.00021059999999999997,
-      "loss": 4.4793,
+      "loss": 4.3128,
       "step": 353
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 0.00021119999999999996,
-      "loss": 4.5593,
+      "loss": 4.4256,
       "step": 354
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021179999999999997,
-      "loss": 5.1268,
+      "loss": 4.4806,
       "step": 355
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021239999999999996,
-      "loss": 4.5701,
+      "loss": 4.6034,
       "step": 356
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021299999999999997,
-      "loss": 4.723,
+      "loss": 4.5852,
       "step": 357
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021359999999999996,
-      "loss": 4.9217,
+      "loss": 4.6319,
       "step": 358
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021419999999999998,
-      "loss": 4.5271,
+      "loss": 4.4082,
       "step": 359
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 0.00021479999999999996,
-      "loss": 4.6167,
+      "loss": 4.4631,
       "step": 360
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 0.00021539999999999998,
-      "loss": 4.6068,
+      "loss": 4.4657,
       "step": 361
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 0.00021599999999999996,
-      "loss": 4.3911,
+      "loss": 4.386,
       "step": 362
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 0.00021659999999999998,
-      "loss": 4.3834,
+      "loss": 4.2731,
       "step": 363
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021719999999999997,
-      "loss": 4.8017,
+      "loss": 4.5098,
       "step": 364
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021779999999999998,
-      "loss": 4.3505,
+      "loss": 4.5502,
       "step": 365
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021839999999999997,
-      "loss": 4.9693,
+      "loss": 4.3979,
       "step": 366
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021899999999999998,
-      "loss": 4.5851,
+      "loss": 4.4207,
       "step": 367
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 0.00021959999999999997,
-      "loss": 4.4264,
+      "loss": 4.3131,
       "step": 368
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 0.00022019999999999999,
-      "loss": 4.5813,
+      "loss": 4.3822,
       "step": 369
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 0.00022079999999999997,
-      "loss": 4.4873,
+      "loss": 4.4616,
       "step": 370
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 0.0002214,
-      "loss": 4.4055,
+      "loss": 4.3793,
       "step": 371
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 0.00022199999999999998,
-      "loss": 4.7036,
+      "loss": 4.4134,
       "step": 372
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.0002226,
-      "loss": 4.5203,
+      "loss": 4.4609,
       "step": 373
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.00022319999999999998,
-      "loss": 4.9871,
+      "loss": 4.1396,
       "step": 374
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.0002238,
-      "loss": 4.3301,
+      "loss": 4.3943,
       "step": 375
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.00022439999999999998,
-      "loss": 4.7857,
+      "loss": 4.3628,
       "step": 376
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 0.000225,
-      "loss": 4.4468,
+      "loss": 4.3679,
       "step": 377
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 0.00022559999999999998,
-      "loss": 4.5666,
+      "loss": 4.5214,
       "step": 378
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 0.00022619999999999997,
-      "loss": 4.4932,
+      "loss": 4.3838,
       "step": 379
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 0.00022679999999999998,
-      "loss": 4.5757,
+      "loss": 4.2615,
       "step": 380
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 0.00022739999999999997,
-      "loss": 4.5106,
+      "loss": 4.6558,
       "step": 381
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.00022799999999999999,
-      "loss": 4.54,
+      "loss": 4.5738,
       "step": 382
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.00022859999999999997,
-      "loss": 5.2259,
+      "loss": 4.4223,
       "step": 383
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.0002292,
-      "loss": 4.4351,
+      "loss": 4.2627,
       "step": 384
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.00022979999999999997,
-      "loss": 4.3878,
+      "loss": 4.4766,
       "step": 385
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 0.0002304,
-      "loss": 4.6613,
+      "loss": 4.5283,
       "step": 386
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 0.00023099999999999998,
-      "loss": 4.5201,
+      "loss": 4.6417,
       "step": 387
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 0.0002316,
-      "loss": 4.5499,
+      "loss": 4.4494,
       "step": 388
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 0.00023219999999999998,
-      "loss": 4.5244,
+      "loss": 4.4591,
       "step": 389
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 0.0002328,
-      "loss": 4.6239,
+      "loss": 4.3444,
       "step": 390
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.00023339999999999998,
-      "loss": 4.8577,
+      "loss": 4.5171,
       "step": 391
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.000234,
-      "loss": 4.775,
+      "loss": 4.3703,
       "step": 392
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.00023459999999999998,
-      "loss": 4.8676,
+      "loss": 4.24,
       "step": 393
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.0002352,
-      "loss": 4.7473,
+      "loss": 4.6065,
       "step": 394
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 0.00023579999999999999,
-      "loss": 5.1698,
+      "loss": 4.5494,
       "step": 395
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 0.0002364,
-      "loss": 5.0435,
+      "loss": 4.4742,
       "step": 396
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 0.000237,
-      "loss": 5.0145,
+      "loss": 4.7762,
       "step": 397
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 0.0002376,
-      "loss": 4.9902,
+      "loss": 4.5632,
       "step": 398
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 0.0002382,
-      "loss": 4.8242,
+      "loss": 5.0284,
       "step": 399
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.0002388,
-      "loss": 4.968,
+      "loss": 4.5484,
       "step": 400
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.0002394,
-      "loss": 4.7138,
+      "loss": 4.7688,
       "step": 401
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.00023999999999999998,
-      "loss": 4.6469,
+      "loss": 4.8056,
       "step": 402
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.0002406,
-      "loss": 5.0462,
+      "loss": 4.6578,
       "step": 403
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 0.00024119999999999998,
-      "loss": 4.5933,
+      "loss": 4.5998,
       "step": 404
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 0.0002418,
-      "loss": 4.6843,
+      "loss": 4.3381,
       "step": 405
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 0.00024239999999999998,
-      "loss": 4.6204,
+      "loss": 4.6957,
       "step": 406
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 0.000243,
-      "loss": 4.4525,
+      "loss": 4.3475,
       "step": 407
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 0.00024359999999999999,
-      "loss": 4.5578,
+      "loss": 4.4122,
       "step": 408
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.00024419999999999997,
-      "loss": 4.6749,
+      "loss": 4.2999,
       "step": 409
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.0002448,
-      "loss": 4.6923,
+      "loss": 4.382,
       "step": 410
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.00024539999999999995,
-      "loss": 4.5342,
+      "loss": 4.6021,
       "step": 411
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.00024599999999999996,
-      "loss": 4.5524,
+      "loss": 4.5202,
       "step": 412
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 0.0002466,
-      "loss": 4.9933,
+      "loss": 4.2693,
       "step": 413
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 0.0002472,
-      "loss": 4.2736,
+      "loss": 4.4864,
       "step": 414
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 0.00024779999999999995,
-      "loss": 4.549,
+      "loss": 4.6728,
       "step": 415
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 0.00024839999999999997,
-      "loss": 4.579,
+      "loss": 4.4965,
       "step": 416
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 0.000249,
-      "loss": 4.5159,
+      "loss": 4.5908,
       "step": 417
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.00024959999999999994,
-      "loss": 4.5826,
+      "loss": 4.323,
       "step": 418
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.00025019999999999996,
-      "loss": 4.4912,
+      "loss": 4.3817,
       "step": 419
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.00025079999999999997,
-      "loss": 4.472,
+      "loss": 4.5368,
       "step": 420
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.0002514,
-      "loss": 4.5307,
+      "loss": 4.6007,
       "step": 421
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 0.00025199999999999995,
-      "loss": 4.8471,
+      "loss": 4.3087,
       "step": 422
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 0.00025259999999999996,
-      "loss": 4.715,
+      "loss": 4.4084,
       "step": 423
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 0.0002532,
-      "loss": 4.7253,
+      "loss": 4.4871,
       "step": 424
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 0.0002538,
-      "loss": 4.6058,
+      "loss": 4.4925,
       "step": 425
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.00025439999999999995,
-      "loss": 4.5831,
+      "loss": 4.3477,
       "step": 426
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.00025499999999999996,
-      "loss": 4.6162,
+      "loss": 4.3819,
       "step": 427
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.0002556,
-      "loss": 4.8844,
+      "loss": 4.4199,
       "step": 428
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.0002562,
-      "loss": 4.6076,
+      "loss": 4.2349,
       "step": 429
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.00025679999999999995,
-      "loss": 4.5971,
+      "loss": 4.6076,
       "step": 430
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.97,
       "learning_rate": 0.00025739999999999997,
-      "loss": 4.8466,
+      "loss": 4.4529,
       "step": 431
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.97,
       "learning_rate": 0.000258,
-      "loss": 4.9865,
+      "loss": 4.6179,
       "step": 432
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.97,
       "learning_rate": 0.0002586,
-      "loss": 4.4494,
+      "loss": 4.3895,
       "step": 433
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.97,
       "learning_rate": 0.00025919999999999996,
-      "loss": 4.6771,
+      "loss": 4.389,
       "step": 434
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 0.00025979999999999997,
-      "loss": 4.5576,
+      "loss": 4.7689,
       "step": 435
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 0.0002604,
-      "loss": 4.8061,
+      "loss": 4.3116,
       "step": 436
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 0.000261,
-      "loss": 5.0413,
+      "loss": 4.5383,
       "step": 437
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 0.00026159999999999996,
-      "loss": 4.6549,
+      "loss": 4.8375,
       "step": 438
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 0.0002622,
-      "loss": 4.5003,
+      "loss": 4.6096,
       "step": 439
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.99,
       "learning_rate": 0.0002628,
-      "loss": 4.6389,
+      "loss": 4.5625,
       "step": 440
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.99,
       "learning_rate": 0.00026339999999999995,
-      "loss": 4.8199,
+      "loss": 4.3984,
       "step": 441
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.99,
       "learning_rate": 0.00026399999999999997,
-      "loss": 4.4152,
+      "loss": 4.8585,
       "step": 442
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.99,
       "learning_rate": 0.0002646,
-      "loss": 4.6895,
+      "loss": 5.0958,
       "step": 443
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 0.0002652,
-      "loss": 5.2644,
+      "loss": 5.1536,
       "step": 444
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 0.00026579999999999996,
-      "loss": 4.6618,
+      "loss": 5.0367,
       "step": 445
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 0.00026639999999999997,
-      "loss": 4.4472,
+      "loss": 4.5288,
       "step": 446
     },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000267,
-      "loss": 4.656,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0002676,
-      "loss": 6.3588,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.00026819999999999996,
-      "loss": 5.6505,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0002688,
-      "loss": 7.89,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002694,
-      "loss": 4.7483,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.00027,
-      "loss": 4.8522,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.00027059999999999996,
-      "loss": 4.8658,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002712,
-      "loss": 4.5986,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002718,
-      "loss": 4.8901,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002724,
-      "loss": 4.9632,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.00027299999999999997,
-      "loss": 4.6942,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002736,
-      "loss": 4.7098,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002742,
-      "loss": 4.726,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002748,
-      "loss": 4.8856,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.00027539999999999997,
-      "loss": 4.4117,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.000276,
-      "loss": 4.7871,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002766,
-      "loss": 4.755,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002772,
-      "loss": 4.4392,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002778,
-      "loss": 4.5389,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002784,
-      "loss": 4.839,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.000279,
-      "loss": 4.5853,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.00027959999999999997,
-      "loss": 4.7598,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002802,
-      "loss": 4.9769,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002808,
-      "loss": 4.5845,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00028139999999999996,
-      "loss": 4.3553,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 4.5508,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002826,
-      "loss": 5.4725,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00028319999999999994,
-      "loss": 4.7822,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00028379999999999996,
-      "loss": 4.7387,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002844,
-      "loss": 4.7934,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.000285,
-      "loss": 4.951,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028559999999999995,
-      "loss": 4.5306,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028619999999999996,
-      "loss": 4.6858,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0002868,
-      "loss": 4.6976,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028739999999999994,
-      "loss": 4.573,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028799999999999995,
-      "loss": 4.7996,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028859999999999997,
-      "loss": 4.5807,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0002892,
-      "loss": 4.788,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028979999999999994,
-      "loss": 4.8346,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00029039999999999996,
-      "loss": 4.467,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029099999999999997,
-      "loss": 4.5186,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0002916,
-      "loss": 4.4779,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029219999999999995,
-      "loss": 4.8221,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029279999999999996,
-      "loss": 4.7609,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0002934,
-      "loss": 4.3691,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000294,
-      "loss": 4.9645,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029459999999999995,
-      "loss": 5.023,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029519999999999997,
-      "loss": 4.7208,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0002958,
-      "loss": 4.9856,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0002964,
-      "loss": 4.6846,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00029699999999999996,
-      "loss": 4.314,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00029759999999999997,
-      "loss": 5.0479,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0002982,
-      "loss": 4.9042,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0002988,
-      "loss": 4.4011,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 4.9983367919921875,
-      "eval_runtime": 875.6736,
-      "eval_samples_per_second": 3.017,
-      "eval_steps_per_second": 0.378,
-      "eval_wer": 1.95178500595002,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00029939999999999996,
-      "loss": 5.1628,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0003,
-      "loss": 4.4468,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000299234693877551,
-      "loss": 4.9952,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029846938775510205,
-      "loss": 4.8411,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029770408163265304,
-      "loss": 4.7504,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002969387755102041,
-      "loss": 4.6484,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029617346938775506,
-      "loss": 4.9217,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002954081632653061,
-      "loss": 4.715,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002946428571428571,
-      "loss": 4.8663,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002938775510204081,
-      "loss": 4.9294,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029311224489795917,
-      "loss": 4.619,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002923469387755102,
-      "loss": 4.6691,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002915816326530612,
-      "loss": 5.4484,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.00029081632653061223,
-      "loss": 4.6958,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002900510204081632,
-      "loss": 4.7626,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.00028928571428571425,
-      "loss": 4.5196,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002885204081632653,
-      "loss": 4.7934,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002877551020408163,
-      "loss": 4.7005,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002869897959183673,
-      "loss": 5.0122,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.00028622448979591836,
-      "loss": 4.819,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002854591836734694,
-      "loss": 4.6823,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002846938775510204,
-      "loss": 4.5671,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00028392857142857137,
-      "loss": 4.7677,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002831632653061224,
-      "loss": 4.9105,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00028239795918367345,
-      "loss": 5.1979,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002816326530612245,
-      "loss": 4.5759,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00028086734693877547,
-      "loss": 4.7022,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002801020408163265,
-      "loss": 4.659,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00027933673469387755,
-      "loss": 4.7747,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00027857142857142854,
-      "loss": 4.9831,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002778061224489796,
-      "loss": 5.1574,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027704081632653056,
-      "loss": 4.5261,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002762755102040816,
-      "loss": 4.731,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027551020408163264,
-      "loss": 4.4291,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002747448979591837,
-      "loss": 4.7664,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027397959183673466,
-      "loss": 4.79,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002732142857142857,
-      "loss": 4.8222,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002724489795918367,
-      "loss": 4.8898,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027168367346938773,
-      "loss": 4.513,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002709183673469387,
-      "loss": 4.7453,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00027015306122448975,
-      "loss": 4.8368,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002693877551020408,
-      "loss": 4.7447,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00026862244897959183,
-      "loss": 4.8438,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00026785714285714287,
-      "loss": 5.2433,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00026709183673469386,
-      "loss": 5.3212,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002663265306122449,
-      "loss": 5.1616,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002655612244897959,
-      "loss": 5.7126,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002647959183673469,
-      "loss": 8.5064,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002640306122448979,
-      "loss": 4.518,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.00026326530612244894,
-      "loss": 5.9617,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002625,
-      "loss": 4.7309,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.000261734693877551,
-      "loss": 4.7899,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.000260969387755102,
-      "loss": 4.7302,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.00026020408163265305,
-      "loss": 4.5338,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.00025943877551020403,
-      "loss": 4.5405,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.00025867346938775507,
-      "loss": 4.6252,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002579081632653061,
-      "loss": 4.4507,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002571428571428571,
-      "loss": 4.6526,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.00025637755102040814,
-      "loss": 4.5054,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002556122448979592,
-      "loss": 4.7053,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002548469387755102,
-      "loss": 4.9174,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002540816326530612,
-      "loss": 4.6808,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002533163265306122,
-      "loss": 4.6278,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002525510204081632,
-      "loss": 4.6145,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.00025178571428571426,
-      "loss": 4.6322,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002510204081632653,
-      "loss": 4.6631,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002502551020408163,
-      "loss": 4.6991,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024948979591836733,
-      "loss": 4.4694,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024872448979591837,
-      "loss": 4.5414,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024795918367346935,
-      "loss": 4.6524,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002471938775510204,
-      "loss": 4.5379,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002464285714285714,
-      "loss": 4.5682,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002456632653061224,
-      "loss": 4.4738,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024489795918367346,
-      "loss": 4.5086,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024413265306122447,
-      "loss": 5.1461,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024336734693877548,
-      "loss": 4.4839,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0002426020408163265,
-      "loss": 4.6375,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024183673469387753,
-      "loss": 4.8622,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024107142857142857,
-      "loss": 4.1603,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024030612244897956,
-      "loss": 4.847,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00023954081632653057,
-      "loss": 4.7087,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0002387755102040816,
-      "loss": 4.492,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00023801020408163265,
-      "loss": 4.7726,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00023724489795918366,
-      "loss": 4.5406,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023647959183673467,
-      "loss": 4.8607,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023571428571428569,
-      "loss": 4.6273,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023494897959183673,
-      "loss": 4.7729,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023418367346938774,
-      "loss": 4.5205,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023341836734693875,
-      "loss": 4.484,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023265306122448976,
-      "loss": 4.6765,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0002318877551020408,
-      "loss": 4.5714,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023112244897959181,
-      "loss": 5.1497,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023035714285714285,
-      "loss": 5.0901,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022959183673469384,
-      "loss": 4.5892,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022882653061224488,
-      "loss": 4.7668,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0002280612244897959,
-      "loss": 4.556,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022729591836734693,
-      "loss": 4.5433,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022653061224489791,
-      "loss": 5.3044,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022576530612244895,
-      "loss": 5.2622,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.000225,
-      "loss": 4.5595,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.000224234693877551,
-      "loss": 4.678,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022346938775510205,
-      "loss": 4.5891,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022270408163265303,
-      "loss": 4.6629,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022193877551020407,
-      "loss": 4.582,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022117346938775508,
-      "loss": 4.5646,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022040816326530612,
-      "loss": 4.6629,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0002196428571428571,
-      "loss": 4.886,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00021887755102040815,
-      "loss": 4.5766,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00021811224489795916,
-      "loss": 4.5959,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0002173469387755102,
-      "loss": 4.6078,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0002165816326530612,
-      "loss": 4.481,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021581632653061222,
-      "loss": 4.7335,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021505102040816324,
-      "loss": 4.5534,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021428571428571427,
-      "loss": 4.9747,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002135204081632653,
-      "loss": 4.5713,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002127551020408163,
-      "loss": 4.5623,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002119897959183673,
-      "loss": 4.3925,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021122448979591835,
-      "loss": 4.3779,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002104591836734694,
-      "loss": 4.3789,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002096938775510204,
-      "loss": 4.5137,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002089285714285714,
-      "loss": 4.7276,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020816326530612243,
-      "loss": 4.49,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020739795918367347,
-      "loss": 4.6151,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020663265306122448,
-      "loss": 4.4768,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002058673469387755,
-      "loss": 4.8337,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002051020408163265,
-      "loss": 4.637,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020433673469387754,
-      "loss": 4.5035,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020357142857142856,
-      "loss": 4.3709,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020280612244897957,
-      "loss": 4.5253,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020204081632653058,
-      "loss": 5.0382,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020127551020408162,
-      "loss": 4.6007,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020051020408163263,
-      "loss": 4.7532,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00019974489795918367,
-      "loss": 4.5891,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00019897959183673466,
-      "loss": 4.4885,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0001982142857142857,
-      "loss": 4.636,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0001974489795918367,
-      "loss": 4.5268,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00019668367346938775,
-      "loss": 4.6191,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019591836734693873,
-      "loss": 4.3703,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019515306122448977,
-      "loss": 4.7706,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0001943877551020408,
-      "loss": 4.5403,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019362244897959182,
-      "loss": 4.4686,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019285714285714286,
-      "loss": 5.3291,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019209183673469385,
-      "loss": 5.1482,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0001913265306122449,
-      "loss": 4.4342,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0001905612244897959,
-      "loss": 4.2783,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00018979591836734694,
-      "loss": 4.6092,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018903061224489793,
-      "loss": 4.4284,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018826530612244896,
-      "loss": 4.4722,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018749999999999998,
-      "loss": 4.8462,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018673469387755102,
-      "loss": 4.7313,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018596938775510203,
-      "loss": 4.8217,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018520408163265304,
-      "loss": 4.7025,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018443877551020405,
-      "loss": 4.6611,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0001836734693877551,
-      "loss": 4.8046,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0001829081632653061,
-      "loss": 4.5228,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00018214285714285712,
-      "loss": 4.6526,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00018137755102040813,
-      "loss": 4.42,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00018061224489795917,
-      "loss": 4.6764,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0001798469387755102,
-      "loss": 4.4985,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00017908163265306122,
-      "loss": 4.6511,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0001783163265306122,
-      "loss": 4.3388,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00017755102040816325,
-      "loss": 4.3057,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00017678571428571428,
-      "loss": 4.981,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0001760204081632653,
-      "loss": 4.5279,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001752551020408163,
-      "loss": 4.316,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017448979591836732,
-      "loss": 4.615,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017372448979591836,
-      "loss": 4.3881,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017295918367346937,
-      "loss": 4.5504,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001721938775510204,
-      "loss": 4.6914,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001714285714285714,
-      "loss": 4.6686,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017066326530612244,
-      "loss": 4.3039,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00016989795918367345,
-      "loss": 4.4468,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001691326530612245,
-      "loss": 4.3383,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016836734693877547,
-      "loss": 4.4286,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016760204081632651,
-      "loss": 4.3745,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016683673469387753,
-      "loss": 4.9497,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016607142857142857,
-      "loss": 4.3681,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016530612244897955,
-      "loss": 4.21,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0001645408163265306,
-      "loss": 4.2797,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016377551020408163,
-      "loss": 4.4616,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016301020408163264,
-      "loss": 4.677,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016224489795918368,
-      "loss": 4.3418,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00016147959183673467,
-      "loss": 4.627,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0001607142857142857,
-      "loss": 4.6252,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015994897959183672,
-      "loss": 4.6252,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015918367346938776,
-      "loss": 4.5539,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015841836734693874,
-      "loss": 4.5075,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015765306122448978,
-      "loss": 4.5828,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0001568877551020408,
-      "loss": 4.5769,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015612244897959183,
-      "loss": 4.8288,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015535714285714285,
-      "loss": 4.3714,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015459183673469386,
-      "loss": 4.5635,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015382653061224487,
-      "loss": 4.3405,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0001530612244897959,
-      "loss": 5.1176,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015229591836734692,
-      "loss": 5.0306,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015153061224489794,
-      "loss": 4.527,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015076530612244895,
-      "loss": 4.9182,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015,
-      "loss": 4.534,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00014923469387755103,
-      "loss": 4.8521,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00014846938775510204,
-      "loss": 4.6326,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014770408163265305,
-      "loss": 4.2919,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014693877551020406,
-      "loss": 4.562,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0001461734693877551,
-      "loss": 4.346,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014540816326530611,
-      "loss": 4.3445,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014464285714285713,
-      "loss": 4.7065,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014387755102040814,
-      "loss": 4.6758,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014311224489795918,
-      "loss": 4.659,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0001423469387755102,
-      "loss": 4.6311,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0001415816326530612,
-      "loss": 4.5146,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00014081632653061224,
-      "loss": 4.5062,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00014005102040816326,
-      "loss": 4.5418,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013928571428571427,
-      "loss": 4.3461,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013852040816326528,
-      "loss": 4.3042,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013775510204081632,
-      "loss": 4.6424,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013698979591836733,
-      "loss": 4.7171,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013622448979591834,
-      "loss": 4.2692,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013545918367346936,
-      "loss": 4.8031,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0001346938775510204,
-      "loss": 4.359,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013392857142857144,
-      "loss": 4.1378,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013316326530612245,
-      "loss": 4.4067,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013239795918367346,
-      "loss": 4.6956,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013163265306122447,
-      "loss": 4.2525,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0001308673469387755,
-      "loss": 4.396,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013010204081632652,
-      "loss": 4.5159,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00012933673469387754,
-      "loss": 4.3099,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00012857142857142855,
-      "loss": 4.6627,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0001278061224489796,
-      "loss": 4.4086,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0001270408163265306,
-      "loss": 4.3537,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0001262755102040816,
-      "loss": 4.303,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012551020408163265,
-      "loss": 4.5849,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012474489795918366,
-      "loss": 4.9582,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012397959183673468,
-      "loss": 4.5712,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0001232142857142857,
-      "loss": 4.3395,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012244897959183673,
-      "loss": 4.5477,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012168367346938774,
-      "loss": 4.2483,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00012091836734693877,
-      "loss": 4.423,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00012015306122448978,
-      "loss": 4.4887,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0001193877551020408,
-      "loss": 4.4166,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011862244897959183,
-      "loss": 4.3083,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011785714285714284,
-      "loss": 4.5407,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011709183673469387,
-      "loss": 4.7349,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011632653061224488,
-      "loss": 4.4709,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011556122448979591,
-      "loss": 4.1676,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011479591836734692,
-      "loss": 4.4411,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011403061224489795,
-      "loss": 4.3631,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011326530612244896,
-      "loss": 4.8217,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0001125,
-      "loss": 4.4436,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011173469387755102,
-      "loss": 4.5294,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011096938775510204,
-      "loss": 4.8539,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011020408163265306,
-      "loss": 4.4265,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00010943877551020407,
-      "loss": 4.4376,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0001086734693877551,
-      "loss": 4.419,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00010790816326530611,
-      "loss": 4.4154,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010714285714285714,
-      "loss": 4.3623,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010637755102040815,
-      "loss": 4.5445,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010561224489795918,
-      "loss": 4.6167,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0001048469387755102,
-      "loss": 4.4716,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010408163265306121,
-      "loss": 4.4618,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010331632653061224,
-      "loss": 4.3187,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010255102040816325,
-      "loss": 4.3345,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010178571428571428,
-      "loss": 4.3305,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010102040816326529,
-      "loss": 4.3747,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00010025510204081632,
-      "loss": 4.3711,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.948979591836733e-05,
-      "loss": 4.2536,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.872448979591835e-05,
-      "loss": 4.6226,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.795918367346937e-05,
-      "loss": 4.3495,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.71938775510204e-05,
-      "loss": 4.3953,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.642857142857143e-05,
-      "loss": 4.4752,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.566326530612244e-05,
-      "loss": 4.4922,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.489795918367347e-05,
-      "loss": 4.4253,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.413265306122448e-05,
-      "loss": 4.2783,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.336734693877551e-05,
-      "loss": 4.5397,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.260204081632652e-05,
-      "loss": 4.6664,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.183673469387755e-05,
-      "loss": 4.1661,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.107142857142856e-05,
-      "loss": 4.1963,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.030612244897958e-05,
-      "loss": 4.8492,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.954081632653061e-05,
-      "loss": 4.1818,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.877551020408162e-05,
-      "loss": 4.5544,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.801020408163265e-05,
-      "loss": 4.3203,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.724489795918366e-05,
-      "loss": 4.5191,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.647959183673469e-05,
-      "loss": 4.4704,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.57142857142857e-05,
-      "loss": 4.4288,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.494897959183672e-05,
-      "loss": 4.5226,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.418367346938774e-05,
-      "loss": 4.2287,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.341836734693876e-05,
-      "loss": 4.3641,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.265306122448978e-05,
-      "loss": 4.2329,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.188775510204081e-05,
-      "loss": 4.471,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.112244897959184e-05,
-      "loss": 4.3837,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.035714285714285e-05,
-      "loss": 4.267,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.959183673469388e-05,
-      "loss": 4.4014,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.882653061224489e-05,
-      "loss": 4.3438,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.806122448979592e-05,
-      "loss": 4.542,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.729591836734693e-05,
-      "loss": 4.3979,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.653061224489796e-05,
-      "loss": 4.5004,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.576530612244897e-05,
-      "loss": 4.6188,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.5e-05,
-      "loss": 4.3081,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.423469387755102e-05,
-      "loss": 4.5427,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.346938775510203e-05,
-      "loss": 4.5724,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.270408163265306e-05,
-      "loss": 4.996,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.193877551020407e-05,
-      "loss": 4.6813,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.11734693877551e-05,
-      "loss": 4.4014,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.040816326530612e-05,
-      "loss": 4.3033,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.964285714285713e-05,
-      "loss": 4.4386,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.887755102040816e-05,
-      "loss": 4.4227,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.811224489795917e-05,
-      "loss": 4.1591,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.73469387755102e-05,
-      "loss": 4.4486,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.658163265306122e-05,
-      "loss": 4.3162,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.581632653061224e-05,
-      "loss": 4.2139,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.505102040816326e-05,
-      "loss": 4.5535,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.428571428571427e-05,
-      "loss": 4.1688,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.35204081632653e-05,
-      "loss": 4.519,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.275510204081633e-05,
-      "loss": 4.4361,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.198979591836734e-05,
-      "loss": 4.1804,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.122448979591836e-05,
-      "loss": 4.2692,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.045918367346938e-05,
-      "loss": 4.2193,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.96938775510204e-05,
-      "loss": 4.2585,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.892857142857142e-05,
-      "loss": 4.3902,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.816326530612244e-05,
-      "loss": 4.2958,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.739795918367346e-05,
-      "loss": 4.2298,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.663265306122448e-05,
-      "loss": 4.3006,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.586734693877551e-05,
-      "loss": 4.3207,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.510204081632653e-05,
-      "loss": 4.1856,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.433673469387755e-05,
-      "loss": 4.346,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.357142857142857e-05,
-      "loss": 4.3479,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.280612244897959e-05,
-      "loss": 4.6312,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.204081632653061e-05,
-      "loss": 4.4575,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.1275510204081626e-05,
-      "loss": 4.3191,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.0510204081632645e-05,
-      "loss": 4.1908,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.9744897959183664e-05,
-      "loss": 4.2245,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.897959183673468e-05,
-      "loss": 4.4717,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.8214285714285716e-05,
-      "loss": 4.1618,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.7448979591836735e-05,
-      "loss": 4.2323,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.6683673469387754e-05,
-      "loss": 4.4067,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.591836734693877e-05,
-      "loss": 4.2481,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.515306122448979e-05,
-      "loss": 4.2898,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.438775510204081e-05,
-      "loss": 4.3192,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.362244897959183e-05,
-      "loss": 4.1546,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.285714285714285e-05,
-      "loss": 4.6285,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.209183673469387e-05,
-      "loss": 4.1972,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.132653061224489e-05,
-      "loss": 4.541,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.056122448979592e-05,
-      "loss": 4.518,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.979591836734694e-05,
-      "loss": 4.3209,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.903061224489796e-05,
-      "loss": 4.5801,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.826530612244898e-05,
-      "loss": 4.4673,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.75e-05,
-      "loss": 4.6158,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.6734693877551016e-05,
-      "loss": 5.7137,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.5969387755102035e-05,
-      "loss": 4.3646,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.520408163265306e-05,
-      "loss": 4.9115,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.443877551020408e-05,
-      "loss": 4.6824,
-      "step": 849
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.36734693877551e-05,
-      "loss": 4.4915,
-      "step": 850
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.290816326530612e-05,
-      "loss": 4.3668,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.214285714285714e-05,
-      "loss": 4.4653,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.137755102040816e-05,
-      "loss": 4.2881,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.061224489795918e-05,
-      "loss": 4.2501,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.98469387755102e-05,
-      "loss": 4.2627,
-      "step": 855
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.908163265306122e-05,
-      "loss": 4.357,
-      "step": 856
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.831632653061224e-05,
-      "loss": 4.5015,
-      "step": 857
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.7551020408163265e-05,
-      "loss": 4.3649,
-      "step": 858
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.6785714285714284e-05,
-      "loss": 4.1966,
-      "step": 859
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.6020408163265303e-05,
-      "loss": 4.334,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.5255102040816323e-05,
-      "loss": 4.6841,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.448979591836734e-05,
-      "loss": 4.2356,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.3724489795918367e-05,
-      "loss": 4.4959,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.2959183673469387e-05,
-      "loss": 4.6351,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.2193877551020406e-05,
-      "loss": 4.449,
-      "step": 865
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.1428571428571425e-05,
-      "loss": 4.2843,
-      "step": 866
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.0663265306122444e-05,
-      "loss": 4.367,
-      "step": 867
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.989795918367347e-05,
-      "loss": 4.2656,
-      "step": 868
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.913265306122449e-05,
-      "loss": 4.273,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.8367346938775508e-05,
-      "loss": 4.4547,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.760204081632653e-05,
-      "loss": 4.3189,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.683673469387755e-05,
-      "loss": 4.3244,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.607142857142857e-05,
-      "loss": 4.381,
-      "step": 873
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.530612244897959e-05,
-      "loss": 4.3439,
-      "step": 874
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.454081632653061e-05,
-      "loss": 4.4997,
-      "step": 875
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.3775510204081633e-05,
-      "loss": 4.332,
-      "step": 876
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.3010204081632652e-05,
-      "loss": 4.2311,
-      "step": 877
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.224489795918367e-05,
-      "loss": 4.3291,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.1479591836734693e-05,
-      "loss": 4.3849,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.0714285714285712e-05,
-      "loss": 4.9003,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 9.948979591836735e-06,
-      "loss": 4.171,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 9.183673469387754e-06,
-      "loss": 4.3592,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 8.418367346938775e-06,
-      "loss": 4.2932,
-      "step": 883
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 7.653061224489796e-06,
-      "loss": 4.3174,
-      "step": 884
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 6.887755102040816e-06,
-      "loss": 4.3816,
-      "step": 885
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 6.122448979591835e-06,
-      "loss": 4.4197,
-      "step": 886
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.357142857142856e-06,
-      "loss": 4.2511,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 4.591836734693877e-06,
-      "loss": 4.4924,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 3.826530612244898e-06,
-      "loss": 4.4857,
-      "step": 889
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 3.0612244897959177e-06,
-      "loss": 4.3505,
-      "step": 890
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.2959183673469385e-06,
-      "loss": 4.5082,
-      "step": 891
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.5306122448979589e-06,
-      "loss": 4.0072,
-      "step": 892
-    },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 4.5466820565574375,
-      "train_runtime": 5895.7357,
-      "train_samples_per_second": 4.84,
-      "train_steps_per_second": 0.151
+      "train_loss": 4.388703477756859,
+      "train_runtime": 4866.733,
+      "train_samples_per_second": 5.864,
+      "train_steps_per_second": 0.092
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,