diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,7 +2,7 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,5375 +10,2690 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.0999,
+      "loss": 9.0593,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.7186,
+      "loss": 8.6798,
       "step": 2
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 8.6553,
+      "loss": 9.0015,
       "step": 3
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 6e-07,
-      "loss": 9.057,
+      "loss": 8.9647,
       "step": 4
     },
     {
       "epoch": 0.01,
       "learning_rate": 6e-07,
-      "loss": 8.9152,
+      "loss": 8.8841,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.2e-06,
-      "loss": 8.6189,
+      "loss": 8.9194,
       "step": 6
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.8e-06,
-      "loss": 8.9645,
+      "loss": 8.7058,
       "step": 7
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-06,
-      "loss": 8.6078,
+      "epoch": 0.02,
+      "learning_rate": 2.4e-06,
+      "loss": 8.4161,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.4e-06,
-      "loss": 8.5715,
+      "epoch": 0.02,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 8.2294,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.9999999999999997e-06,
-      "loss": 8.4233,
+      "epoch": 0.02,
+      "learning_rate": 3.6e-06,
+      "loss": 8.3752,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.6e-06,
-      "loss": 8.2472,
+      "epoch": 0.02,
+      "learning_rate": 4.2e-06,
+      "loss": 8.0861,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.2e-06,
-      "loss": 8.2332,
+      "epoch": 0.03,
+      "learning_rate": 4.8e-06,
+      "loss": 8.0248,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.8e-06,
-      "loss": 7.9932,
+      "epoch": 0.03,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 7.759,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.399999999999999e-06,
-      "loss": 8.0419,
+      "epoch": 0.03,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 7.8552,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.999999999999999e-06,
-      "loss": 7.9345,
+      "epoch": 0.03,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 7.6648,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.599999999999999e-06,
-      "loss": 7.7465,
+      "epoch": 0.04,
+      "learning_rate": 7.2e-06,
+      "loss": 7.6878,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.2e-06,
-      "loss": 7.6716,
+      "epoch": 0.04,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 7.5286,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.799999999999998e-06,
-      "loss": 7.6993,
+      "epoch": 0.04,
+      "learning_rate": 8.4e-06,
+      "loss": 7.5442,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.4e-06,
-      "loss": 7.5265,
+      "epoch": 0.04,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 7.3928,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.999999999999999e-06,
-      "loss": 7.5412,
+      "epoch": 0.04,
+      "learning_rate": 9.6e-06,
+      "loss": 7.4898,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 9.6e-06,
-      "loss": 7.5321,
+      "epoch": 0.05,
+      "learning_rate": 1.02e-05,
+      "loss": 7.4324,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.02e-05,
-      "loss": 7.4543,
+      "epoch": 0.05,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 7.4232,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.0799999999999998e-05,
-      "loss": 7.4675,
+      "epoch": 0.05,
+      "learning_rate": 1.14e-05,
+      "loss": 7.3906,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.14e-05,
-      "loss": 7.4239,
+      "epoch": 0.05,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 7.3289,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.1999999999999999e-05,
-      "loss": 7.3541,
+      "epoch": 0.06,
+      "learning_rate": 1.26e-05,
+      "loss": 7.2432,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.26e-05,
-      "loss": 7.424,
+      "epoch": 0.06,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 7.378,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.3199999999999997e-05,
-      "loss": 7.4128,
+      "epoch": 0.06,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 7.2846,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.3799999999999998e-05,
-      "loss": 7.1907,
+      "epoch": 0.06,
+      "learning_rate": 1.44e-05,
+      "loss": 7.1664,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.44e-05,
-      "loss": 7.3099,
+      "epoch": 0.07,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 7.2449,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.4999999999999999e-05,
-      "loss": 7.28,
+      "epoch": 0.07,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 7.2556,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.5599999999999996e-05,
-      "loss": 7.3751,
+      "epoch": 0.07,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 7.2762,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.6199999999999997e-05,
-      "loss": 7.2808,
+      "epoch": 0.07,
+      "learning_rate": 1.68e-05,
+      "loss": 7.2141,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.68e-05,
-      "loss": 7.2691,
+      "epoch": 0.07,
+      "learning_rate": 1.74e-05,
+      "loss": 7.2166,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.74e-05,
-      "loss": 7.173,
+      "epoch": 0.08,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 7.1265,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.7999999999999997e-05,
-      "loss": 7.2658,
+      "epoch": 0.08,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 7.1288,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.8599999999999998e-05,
-      "loss": 7.234,
+      "epoch": 0.08,
+      "learning_rate": 1.92e-05,
+      "loss": 7.1726,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.92e-05,
-      "loss": 7.2649,
+      "epoch": 0.08,
+      "learning_rate": 1.98e-05,
+      "loss": 7.2633,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.98e-05,
-      "loss": 7.2282,
+      "epoch": 0.09,
+      "learning_rate": 2.04e-05,
+      "loss": 7.1457,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.04e-05,
-      "loss": 7.363,
+      "epoch": 0.09,
+      "learning_rate": 2.1e-05,
+      "loss": 7.2118,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.1e-05,
-      "loss": 7.1886,
+      "epoch": 0.09,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 7.1554,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.1599999999999996e-05,
-      "loss": 7.1061,
+      "epoch": 0.09,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 7.2002,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.2199999999999998e-05,
-      "loss": 7.2658,
+      "epoch": 0.09,
+      "learning_rate": 2.28e-05,
+      "loss": 7.2066,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.28e-05,
-      "loss": 7.0366,
+      "epoch": 0.1,
+      "learning_rate": 2.34e-05,
+      "loss": 6.9742,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.34e-05,
-      "loss": 7.2335,
+      "epoch": 0.1,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 7.1299,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.3999999999999997e-05,
-      "loss": 7.2617,
+      "epoch": 0.1,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 7.1903,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.4599999999999998e-05,
-      "loss": 7.2279,
+      "epoch": 0.1,
+      "learning_rate": 2.52e-05,
+      "loss": 7.1265,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.52e-05,
-      "loss": 6.9998,
+      "epoch": 0.11,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 7.0296,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.5799999999999997e-05,
-      "loss": 7.0601,
+      "epoch": 0.11,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 7.0528,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.6399999999999995e-05,
-      "loss": 6.891,
+      "epoch": 0.11,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 6.7974,
       "step": 49
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.6999999999999996e-05,
-      "loss": 6.7733,
+      "epoch": 0.11,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 6.5233,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.7599999999999997e-05,
-      "loss": 7.2612,
+      "epoch": 0.11,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 7.3589,
       "step": 51
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 7.2054,
+      "epoch": 0.12,
+      "learning_rate": 2.88e-05,
+      "loss": 7.2445,
       "step": 52
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.88e-05,
-      "loss": 7.0902,
+      "epoch": 0.12,
+      "learning_rate": 2.94e-05,
+      "loss": 7.2458,
       "step": 53
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.94e-05,
-      "loss": 7.2092,
+      "epoch": 0.12,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 7.1335,
       "step": 54
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.9999999999999997e-05,
-      "loss": 7.0408,
+      "epoch": 0.12,
+      "learning_rate": 3.06e-05,
+      "loss": 6.976,
       "step": 55
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.06e-05,
-      "loss": 7.1568,
+      "epoch": 0.13,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 7.0973,
       "step": 56
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.119999999999999e-05,
-      "loss": 7.1719,
+      "epoch": 0.13,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 7.1603,
       "step": 57
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.1799999999999994e-05,
-      "loss": 7.2245,
+      "epoch": 0.13,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 7.228,
       "step": 58
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.2399999999999995e-05,
-      "loss": 7.0008,
+      "epoch": 0.13,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 7.081,
       "step": 59
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.2999999999999996e-05,
-      "loss": 7.1629,
+      "epoch": 0.13,
+      "learning_rate": 3.36e-05,
+      "loss": 7.1042,
       "step": 60
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.36e-05,
-      "loss": 7.154,
+      "epoch": 0.14,
+      "learning_rate": 3.42e-05,
+      "loss": 7.0489,
       "step": 61
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.42e-05,
-      "loss": 7.103,
+      "epoch": 0.14,
+      "learning_rate": 3.48e-05,
+      "loss": 7.0797,
       "step": 62
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.48e-05,
-      "loss": 7.0558,
+      "epoch": 0.14,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 7.0067,
       "step": 63
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.539999999999999e-05,
-      "loss": 7.1201,
+      "epoch": 0.14,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 7.1171,
       "step": 64
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.5999999999999994e-05,
-      "loss": 6.9715,
+      "epoch": 0.15,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 7.0183,
       "step": 65
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.6599999999999995e-05,
-      "loss": 7.2314,
+      "epoch": 0.15,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 6.987,
       "step": 66
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.7199999999999996e-05,
-      "loss": 7.0264,
+      "epoch": 0.15,
+      "learning_rate": 3.78e-05,
+      "loss": 7.0242,
       "step": 67
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.78e-05,
-      "loss": 6.9733,
+      "epoch": 0.15,
+      "learning_rate": 3.84e-05,
+      "loss": 7.0413,
       "step": 68
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.84e-05,
-      "loss": 7.0547,
+      "epoch": 0.15,
+      "learning_rate": 3.9e-05,
+      "loss": 7.0896,
       "step": 69
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.9e-05,
-      "loss": 7.0174,
+      "epoch": 0.16,
+      "learning_rate": 3.96e-05,
+      "loss": 6.9828,
       "step": 70
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.96e-05,
-      "loss": 7.1657,
+      "epoch": 0.16,
+      "learning_rate": 4.02e-05,
+      "loss": 7.1095,
       "step": 71
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.02e-05,
-      "loss": 7.2093,
+      "epoch": 0.16,
+      "learning_rate": 4.08e-05,
+      "loss": 7.0894,
       "step": 72
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.08e-05,
-      "loss": 6.9803,
+      "epoch": 0.16,
+      "learning_rate": 4.14e-05,
+      "loss": 7.0473,
       "step": 73
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.14e-05,
-      "loss": 7.1377,
+      "epoch": 0.17,
+      "learning_rate": 4.2e-05,
+      "loss": 7.0871,
       "step": 74
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.2e-05,
-      "loss": 6.9877,
+      "epoch": 0.17,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 7.0546,
       "step": 75
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.259999999999999e-05,
-      "loss": 7.0808,
+      "epoch": 0.17,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 7.081,
       "step": 76
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.319999999999999e-05,
-      "loss": 7.0058,
+      "epoch": 0.17,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 7.0875,
       "step": 77
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.3799999999999994e-05,
-      "loss": 7.0018,
+      "epoch": 0.17,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 7.0912,
       "step": 78
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.4399999999999995e-05,
-      "loss": 6.953,
+      "epoch": 0.18,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 6.971,
       "step": 79
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.4999999999999996e-05,
-      "loss": 7.0265,
+      "epoch": 0.18,
+      "learning_rate": 4.56e-05,
+      "loss": 7.0758,
       "step": 80
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.56e-05,
-      "loss": 7.0557,
+      "epoch": 0.18,
+      "learning_rate": 4.62e-05,
+      "loss": 7.0612,
       "step": 81
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.62e-05,
-      "loss": 6.9811,
+      "epoch": 0.18,
+      "learning_rate": 4.68e-05,
+      "loss": 7.0599,
       "step": 82
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.68e-05,
-      "loss": 7.1628,
+      "epoch": 0.19,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 7.053,
       "step": 83
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.7399999999999993e-05,
-      "loss": 6.9113,
+      "epoch": 0.19,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 7.0249,
       "step": 84
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.7999999999999994e-05,
-      "loss": 6.9547,
+      "epoch": 0.19,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 7.0136,
       "step": 85
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.8599999999999995e-05,
-      "loss": 7.0466,
+      "epoch": 0.19,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 7.0577,
       "step": 86
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.9199999999999997e-05,
-      "loss": 6.9777,
+      "epoch": 0.2,
+      "learning_rate": 4.98e-05,
+      "loss": 7.0064,
       "step": 87
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.98e-05,
-      "loss": 6.9685,
+      "epoch": 0.2,
+      "learning_rate": 5.04e-05,
+      "loss": 6.9519,
       "step": 88
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.04e-05,
-      "loss": 7.0273,
+      "epoch": 0.2,
+      "learning_rate": 5.1e-05,
+      "loss": 6.9031,
       "step": 89
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.1e-05,
-      "loss": 7.0866,
+      "epoch": 0.2,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 7.039,
       "step": 90
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.1599999999999994e-05,
-      "loss": 7.1824,
+      "epoch": 0.2,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 6.9379,
       "step": 91
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.2199999999999995e-05,
-      "loss": 7.0954,
+      "epoch": 0.21,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 7.1254,
       "step": 92
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.279999999999999e-05,
-      "loss": 6.8591,
+      "epoch": 0.21,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 7.0077,
       "step": 93
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.339999999999999e-05,
-      "loss": 6.9837,
+      "epoch": 0.21,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 6.8346,
       "step": 94
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.399999999999999e-05,
-      "loss": 7.1024,
+      "epoch": 0.21,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 6.7611,
       "step": 95
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.459999999999999e-05,
-      "loss": 7.1105,
+      "epoch": 0.22,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 6.982,
       "step": 96
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.519999999999999e-05,
-      "loss": 6.9283,
+      "epoch": 0.22,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 6.7893,
       "step": 97
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.5799999999999994e-05,
-      "loss": 6.9465,
+      "epoch": 0.22,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 6.5833,
       "step": 98
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 6.6928,
+      "epoch": 0.22,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 6.6579,
       "step": 99
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 6.3529,
+      "epoch": 0.22,
+      "learning_rate": 5.76e-05,
+      "loss": 6.2381,
       "step": 100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.76e-05,
-      "loss": 7.2781,
+      "epoch": 0.23,
+      "learning_rate": 5.82e-05,
+      "loss": 7.1487,
       "step": 101
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.82e-05,
-      "loss": 7.2954,
+      "epoch": 0.23,
+      "learning_rate": 5.88e-05,
+      "loss": 7.1538,
       "step": 102
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 5.88e-05,
-      "loss": 7.1813,
+      "epoch": 0.23,
+      "learning_rate": 5.94e-05,
+      "loss": 6.9778,
       "step": 103
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 5.94e-05,
-      "loss": 7.0713,
+      "epoch": 0.23,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 6.9727,
       "step": 104
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 6.9738,
+      "epoch": 0.24,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 6.9908,
       "step": 105
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.0599999999999996e-05,
-      "loss": 7.0337,
+      "epoch": 0.24,
+      "learning_rate": 6.12e-05,
+      "loss": 6.9922,
       "step": 106
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.12e-05,
-      "loss": 7.1243,
+      "epoch": 0.24,
+      "learning_rate": 6.18e-05,
+      "loss": 6.9424,
       "step": 107
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.18e-05,
-      "loss": 7.1842,
+      "epoch": 0.24,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 6.9689,
       "step": 108
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.239999999999999e-05,
-      "loss": 6.8733,
+      "epoch": 0.24,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 6.8709,
       "step": 109
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.299999999999999e-05,
-      "loss": 7.0055,
+      "epoch": 0.25,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 6.9555,
       "step": 110
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.359999999999999e-05,
-      "loss": 7.0452,
+      "epoch": 0.25,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 6.8452,
       "step": 111
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.419999999999999e-05,
-      "loss": 7.167,
+      "epoch": 0.25,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 6.8586,
       "step": 112
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.479999999999999e-05,
-      "loss": 7.0355,
+      "epoch": 0.25,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 6.8972,
       "step": 113
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.539999999999999e-05,
-      "loss": 7.0055,
+      "epoch": 0.26,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 6.9495,
       "step": 114
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.599999999999999e-05,
-      "loss": 7.0149,
+      "epoch": 0.26,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 6.8688,
       "step": 115
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.659999999999999e-05,
-      "loss": 6.8778,
+      "epoch": 0.26,
+      "learning_rate": 6.72e-05,
+      "loss": 7.0212,
       "step": 116
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.72e-05,
-      "loss": 7.0386,
+      "epoch": 0.26,
+      "learning_rate": 6.78e-05,
+      "loss": 6.9658,
       "step": 117
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.78e-05,
-      "loss": 6.9789,
+      "epoch": 0.26,
+      "learning_rate": 6.84e-05,
+      "loss": 6.9449,
       "step": 118
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.84e-05,
-      "loss": 7.0005,
+      "epoch": 0.27,
+      "learning_rate": 6.9e-05,
+      "loss": 6.8678,
       "step": 119
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 6.9e-05,
-      "loss": 7.0509,
+      "epoch": 0.27,
+      "learning_rate": 6.96e-05,
+      "loss": 6.9486,
       "step": 120
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 6.96e-05,
-      "loss": 6.9427,
+      "epoch": 0.27,
+      "learning_rate": 7.02e-05,
+      "loss": 6.8725,
       "step": 121
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.02e-05,
-      "loss": 7.098,
+      "epoch": 0.27,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 6.8252,
       "step": 122
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.079999999999999e-05,
-      "loss": 6.8362,
+      "epoch": 0.28,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 6.8825,
       "step": 123
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.139999999999999e-05,
-      "loss": 6.9948,
+      "epoch": 0.28,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 6.8312,
       "step": 124
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.199999999999999e-05,
-      "loss": 7.0379,
+      "epoch": 0.28,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 6.9445,
       "step": 125
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.259999999999999e-05,
-      "loss": 7.1541,
+      "epoch": 0.28,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 6.9267,
       "step": 126
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.319999999999999e-05,
-      "loss": 7.1344,
+      "epoch": 0.28,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 6.91,
       "step": 127
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.379999999999999e-05,
-      "loss": 7.0745,
+      "epoch": 0.29,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 6.8933,
       "step": 128
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 7.439999999999999e-05,
-      "loss": 6.8535,
+      "epoch": 0.29,
+      "learning_rate": 7.5e-05,
+      "loss": 6.9331,
       "step": 129
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 7.5e-05,
-      "loss": 7.0349,
-      "step": 130
-    },
-    {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 7.56e-05,
-      "loss": 6.9984,
-      "step": 131
+      "loss": 6.998,
+      "step": 130
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 7.62e-05,
-      "loss": 7.0613,
-      "step": 132
+      "loss": 6.7967,
+      "step": 131
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.68e-05,
-      "loss": 6.9975,
-      "step": 133
+      "loss": 6.9952,
+      "step": 132
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.74e-05,
-      "loss": 7.0731,
-      "step": 134
+      "loss": 6.844,
+      "step": 133
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.8e-05,
-      "loss": 6.8142,
-      "step": 135
+      "loss": 6.8629,
+      "step": 134
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.86e-05,
-      "loss": 6.9361,
-      "step": 136
+      "loss": 6.9994,
+      "step": 135
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 7.92e-05,
-      "loss": 7.1242,
-      "step": 137
+      "loss": 6.7906,
+      "step": 136
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 7.98e-05,
-      "loss": 6.9682,
-      "step": 138
+      "loss": 6.9598,
+      "step": 137
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 8.04e-05,
-      "loss": 6.8986,
-      "step": 139
+      "loss": 6.8899,
+      "step": 138
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 8.1e-05,
-      "loss": 6.8913,
-      "step": 140
+      "loss": 6.8555,
+      "step": 139
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 8.16e-05,
-      "loss": 7.1061,
-      "step": 141
+      "loss": 6.9729,
+      "step": 140
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.22e-05,
-      "loss": 6.9993,
-      "step": 142
+      "loss": 6.9761,
+      "step": 141
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.28e-05,
-      "loss": 6.9815,
-      "step": 143
+      "loss": 6.9451,
+      "step": 142
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.34e-05,
-      "loss": 6.7689,
-      "step": 144
+      "loss": 6.7803,
+      "step": 143
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.4e-05,
-      "loss": 6.8309,
-      "step": 145
+      "loss": 6.9795,
+      "step": 144
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 8.459999999999998e-05,
-      "loss": 6.7216,
-      "step": 146
+      "loss": 6.7497,
+      "step": 145
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 8.519999999999998e-05,
-      "loss": 6.9865,
-      "step": 147
+      "loss": 6.7688,
+      "step": 146
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 8.579999999999998e-05,
-      "loss": 6.5033,
-      "step": 148
+      "loss": 6.8215,
+      "step": 147
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 8.639999999999999e-05,
-      "loss": 6.7463,
-      "step": 149
+      "loss": 6.8295,
+      "step": 148
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 8.699999999999999e-05,
-      "loss": 6.2402,
-      "step": 150
+      "loss": 6.5939,
+      "step": 149
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.759999999999999e-05,
-      "loss": 7.0729,
-      "step": 151
+      "loss": 6.4406,
+      "step": 150
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.819999999999999e-05,
-      "loss": 7.0306,
-      "step": 152
+      "loss": 7.0335,
+      "step": 151
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.879999999999999e-05,
-      "loss": 7.0497,
-      "step": 153
+      "loss": 7.0794,
+      "step": 152
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.939999999999999e-05,
-      "loss": 6.8533,
-      "step": 154
+      "loss": 6.962,
+      "step": 153
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 8.999999999999999e-05,
-      "loss": 6.8796,
-      "step": 155
+      "loss": 6.8484,
+      "step": 154
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 9.059999999999999e-05,
-      "loss": 6.9443,
-      "step": 156
+      "loss": 6.8841,
+      "step": 155
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 9.12e-05,
-      "loss": 7.1109,
-      "step": 157
+      "loss": 6.8999,
+      "step": 156
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 9.18e-05,
-      "loss": 7.0729,
-      "step": 158
+      "loss": 6.9885,
+      "step": 157
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 9.24e-05,
-      "loss": 7.0457,
-      "step": 159
+      "loss": 6.9219,
+      "step": 158
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.3e-05,
-      "loss": 7.0171,
-      "step": 160
+      "loss": 6.91,
+      "step": 159
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.36e-05,
-      "loss": 6.8649,
-      "step": 161
+      "loss": 6.9633,
+      "step": 160
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.419999999999999e-05,
-      "loss": 6.9131,
-      "step": 162
+      "loss": 6.9086,
+      "step": 161
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 9.479999999999999e-05,
-      "loss": 6.949,
-      "step": 163
+      "loss": 6.967,
+      "step": 162
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 9.539999999999999e-05,
-      "loss": 6.9085,
-      "step": 164
+      "loss": 6.9233,
+      "step": 163
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 9.599999999999999e-05,
-      "loss": 6.957,
-      "step": 165
+      "loss": 6.9295,
+      "step": 164
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 9.659999999999999e-05,
-      "loss": 6.9318,
-      "step": 166
+      "loss": 6.8712,
+      "step": 165
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 9.719999999999999e-05,
-      "loss": 6.847,
-      "step": 167
+      "loss": 6.9343,
+      "step": 166
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 9.779999999999999e-05,
-      "loss": 6.9671,
-      "step": 168
+      "loss": 6.8514,
+      "step": 167
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 9.839999999999999e-05,
-      "loss": 6.9981,
-      "step": 169
+      "loss": 6.7933,
+      "step": 168
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 9.9e-05,
-      "loss": 6.9109,
-      "step": 170
+      "loss": 6.8052,
+      "step": 169
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 9.96e-05,
-      "loss": 7.1042,
-      "step": 171
+      "loss": 6.9495,
+      "step": 170
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 0.0001002,
-      "loss": 7.0595,
-      "step": 172
+      "loss": 6.8916,
+      "step": 171
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 0.0001008,
-      "loss": 7.0398,
-      "step": 173
+      "loss": 6.9415,
+      "step": 172
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 0.0001014,
-      "loss": 6.934,
-      "step": 174
+      "loss": 6.8972,
+      "step": 173
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 0.000102,
-      "loss": 6.7943,
-      "step": 175
+      "loss": 6.9499,
+      "step": 174
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 0.0001026,
-      "loss": 6.9293,
-      "step": 176
+      "loss": 6.9356,
+      "step": 175
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 0.00010319999999999999,
-      "loss": 6.9352,
-      "step": 177
+      "loss": 6.8336,
+      "step": 176
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010379999999999999,
-      "loss": 6.82,
-      "step": 178
+      "loss": 6.8248,
+      "step": 177
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010439999999999999,
-      "loss": 7.03,
-      "step": 179
+      "loss": 6.8928,
+      "step": 178
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010499999999999999,
-      "loss": 7.015,
-      "step": 180
+      "loss": 6.9024,
+      "step": 179
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 0.00010559999999999998,
-      "loss": 7.0113,
-      "step": 181
+      "loss": 6.9626,
+      "step": 180
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 0.00010619999999999998,
-      "loss": 6.8457,
-      "step": 182
+      "loss": 6.9259,
+      "step": 181
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 0.00010679999999999998,
-      "loss": 6.8977,
-      "step": 183
+      "loss": 6.8326,
+      "step": 182
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 0.00010739999999999998,
-      "loss": 7.0413,
-      "step": 184
+      "loss": 6.8831,
+      "step": 183
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 0.00010799999999999998,
-      "loss": 6.9299,
-      "step": 185
+      "loss": 6.97,
+      "step": 184
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 0.00010859999999999998,
-      "loss": 7.0626,
-      "step": 186
+      "loss": 6.8943,
+      "step": 185
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00010919999999999998,
-      "loss": 7.0246,
-      "step": 187
+      "loss": 6.7704,
+      "step": 186
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00010979999999999999,
-      "loss": 6.8238,
-      "step": 188
+      "loss": 6.8563,
+      "step": 187
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00011039999999999999,
-      "loss": 6.8894,
-      "step": 189
+      "loss": 6.8277,
+      "step": 188
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 0.00011099999999999999,
-      "loss": 6.8789,
-      "step": 190
+      "loss": 6.919,
+      "step": 189
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 0.00011159999999999999,
-      "loss": 6.8802,
-      "step": 191
+      "loss": 6.9639,
+      "step": 190
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 0.00011219999999999999,
-      "loss": 7.0403,
-      "step": 192
+      "loss": 6.9247,
+      "step": 191
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 0.00011279999999999999,
-      "loss": 7.0831,
-      "step": 193
+      "loss": 6.8216,
+      "step": 192
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 0.00011339999999999999,
-      "loss": 6.8587,
-      "step": 194
+      "loss": 6.8592,
+      "step": 193
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 0.00011399999999999999,
-      "loss": 6.6345,
-      "step": 195
+      "loss": 6.7915,
+      "step": 194
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.0001146,
-      "loss": 7.0325,
-      "step": 196
+      "loss": 6.7807,
+      "step": 195
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.0001152,
-      "loss": 6.6957,
-      "step": 197
+      "loss": 6.6464,
+      "step": 196
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.0001158,
-      "loss": 6.626,
-      "step": 198
+      "loss": 6.62,
+      "step": 197
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 0.0001164,
-      "loss": 6.5982,
-      "step": 199
+      "loss": 6.6885,
+      "step": 198
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 0.000117,
-      "loss": 6.2445,
-      "step": 200
+      "loss": 6.5504,
+      "step": 199
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 0.0001176,
-      "loss": 6.9832,
-      "step": 201
+      "loss": 6.3884,
+      "step": 200
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 0.0001182,
-      "loss": 7.0513,
-      "step": 202
+      "loss": 7.2501,
+      "step": 201
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 0.0001188,
-      "loss": 6.9627,
-      "step": 203
+      "loss": 7.1882,
+      "step": 202
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.0001194,
-      "loss": 6.9495,
-      "step": 204
+      "loss": 7.0557,
+      "step": 203
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00011999999999999999,
-      "loss": 6.9463,
-      "step": 205
+      "loss": 6.9325,
+      "step": 204
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00012059999999999999,
-      "loss": 6.8831,
-      "step": 206
+      "loss": 7.0032,
+      "step": 205
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00012119999999999999,
-      "loss": 6.9067,
-      "step": 207
+      "loss": 6.9799,
+      "step": 206
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 0.00012179999999999999,
-      "loss": 6.8835,
-      "step": 208
+      "loss": 6.8919,
+      "step": 207
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 0.0001224,
-      "loss": 6.8963,
-      "step": 209
+      "loss": 6.9957,
+      "step": 208
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 0.00012299999999999998,
-      "loss": 6.8928,
-      "step": 210
+      "loss": 6.9433,
+      "step": 209
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 0.0001236,
-      "loss": 6.7872,
-      "step": 211
+      "loss": 6.9953,
+      "step": 210
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 0.00012419999999999998,
-      "loss": 6.8213,
-      "step": 212
+      "loss": 6.855,
+      "step": 211
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.00012479999999999997,
-      "loss": 6.8724,
-      "step": 213
+      "loss": 6.9195,
+      "step": 212
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.00012539999999999999,
-      "loss": 6.9498,
-      "step": 214
+      "loss": 6.794,
+      "step": 213
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.00012599999999999997,
-      "loss": 6.8644,
-      "step": 215
+      "loss": 6.8751,
+      "step": 214
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.0001266,
-      "loss": 6.9756,
-      "step": 216
+      "loss": 6.8524,
+      "step": 215
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 0.00012719999999999997,
-      "loss": 6.894,
-      "step": 217
+      "loss": 6.8992,
+      "step": 216
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 0.0001278,
-      "loss": 6.9161,
-      "step": 218
+      "loss": 6.8588,
+      "step": 217
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 0.00012839999999999998,
-      "loss": 6.8155,
-      "step": 219
+      "loss": 6.881,
+      "step": 218
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 0.000129,
-      "loss": 6.9999,
-      "step": 220
+      "loss": 6.9031,
+      "step": 219
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 0.00012959999999999998,
-      "loss": 6.8851,
-      "step": 221
+      "loss": 6.9355,
+      "step": 220
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.0001302,
-      "loss": 6.8,
-      "step": 222
+      "loss": 6.9106,
+      "step": 221
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.00013079999999999998,
-      "loss": 6.9343,
-      "step": 223
+      "loss": 6.9313,
+      "step": 222
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.0001314,
-      "loss": 6.8666,
-      "step": 224
+      "loss": 6.8701,
+      "step": 223
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.00013199999999999998,
-      "loss": 6.898,
-      "step": 225
+      "loss": 6.8505,
+      "step": 224
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 0.0001326,
-      "loss": 7.059,
-      "step": 226
+      "loss": 6.9826,
+      "step": 225
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 0.00013319999999999999,
-      "loss": 6.8504,
-      "step": 227
+      "loss": 6.8822,
+      "step": 226
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 0.0001338,
-      "loss": 6.8917,
-      "step": 228
+      "loss": 6.992,
+      "step": 227
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 0.0001344,
-      "loss": 6.9666,
-      "step": 229
+      "loss": 6.9271,
+      "step": 228
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 0.000135,
-      "loss": 7.0132,
-      "step": 230
+      "loss": 6.8334,
+      "step": 229
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001356,
-      "loss": 6.8798,
-      "step": 231
+      "loss": 6.9154,
+      "step": 230
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001362,
-      "loss": 7.0676,
-      "step": 232
+      "loss": 6.8884,
+      "step": 231
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001368,
-      "loss": 6.9102,
-      "step": 233
+      "loss": 6.7658,
+      "step": 232
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.0001374,
-      "loss": 6.9132,
-      "step": 234
+      "loss": 6.8042,
+      "step": 233
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 0.000138,
-      "loss": 7.0219,
-      "step": 235
+      "loss": 6.8418,
+      "step": 234
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 0.0001386,
-      "loss": 6.8001,
-      "step": 236
+      "loss": 6.9673,
+      "step": 235
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 0.0001392,
-      "loss": 6.8554,
-      "step": 237
+      "loss": 6.8433,
+      "step": 236
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 0.00013979999999999998,
-      "loss": 7.0862,
-      "step": 238
+      "loss": 6.9966,
+      "step": 237
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 0.0001404,
-      "loss": 6.7867,
-      "step": 239
+      "loss": 6.8619,
+      "step": 238
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.00014099999999999998,
-      "loss": 6.8978,
-      "step": 240
+      "loss": 6.7954,
+      "step": 239
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.00014159999999999997,
-      "loss": 6.8487,
-      "step": 241
+      "loss": 7.0012,
+      "step": 240
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.0001422,
-      "loss": 6.986,
-      "step": 242
+      "loss": 6.8357,
+      "step": 241
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.00014279999999999997,
-      "loss": 6.8579,
-      "step": 243
+      "loss": 6.8615,
+      "step": 242
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 0.0001434,
-      "loss": 6.736,
-      "step": 244
+      "loss": 7.4713,
+      "step": 243
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 0.00014399999999999998,
-      "loss": 6.8802,
-      "step": 245
+      "loss": 6.7664,
+      "step": 244
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 0.0001446,
-      "loss": 6.8379,
-      "step": 246
+      "loss": 6.7984,
+      "step": 245
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 0.00014519999999999998,
-      "loss": 6.9125,
-      "step": 247
+      "loss": 6.7477,
+      "step": 246
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 0.0001458,
-      "loss": 6.9465,
-      "step": 248
+      "loss": 6.5329,
+      "step": 247
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.00014639999999999998,
-      "loss": 6.7558,
-      "step": 249
+      "loss": 6.754,
+      "step": 248
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.000147,
-      "loss": 6.5203,
-      "step": 250
+      "loss": 6.6694,
+      "step": 249
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.00014759999999999998,
-      "loss": 7.1518,
-      "step": 251
+      "loss": 6.333,
+      "step": 250
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 0.0001482,
-      "loss": 7.0714,
-      "step": 252
+      "loss": 7.1766,
+      "step": 251
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 0.00014879999999999998,
-      "loss": 6.9628,
-      "step": 253
+      "loss": 7.1344,
+      "step": 252
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 0.0001494,
-      "loss": 7.0631,
-      "step": 254
+      "loss": 7.0821,
+      "step": 253
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 0.00015,
-      "loss": 6.8613,
-      "step": 255
+      "loss": 7.0323,
+      "step": 254
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 0.00015059999999999997,
-      "loss": 7.0024,
-      "step": 256
+      "loss": 7.0408,
+      "step": 255
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 0.0001512,
-      "loss": 6.9627,
-      "step": 257
+      "loss": 6.9713,
+      "step": 256
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.00015179999999999998,
-      "loss": 6.9489,
-      "step": 258
+      "loss": 6.9548,
+      "step": 257
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.0001524,
-      "loss": 6.8986,
-      "step": 259
+      "loss": 6.9471,
+      "step": 258
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.00015299999999999998,
-      "loss": 7.0233,
-      "step": 260
+      "loss": 6.8788,
+      "step": 259
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 0.0001536,
-      "loss": 6.9037,
-      "step": 261
+      "loss": 6.9004,
+      "step": 260
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 0.00015419999999999998,
-      "loss": 6.8283,
-      "step": 262
+      "loss": 6.9344,
+      "step": 261
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 0.0001548,
-      "loss": 6.9515,
-      "step": 263
+      "loss": 6.8965,
+      "step": 262
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 0.00015539999999999998,
-      "loss": 6.9285,
-      "step": 264
+      "loss": 6.9719,
+      "step": 263
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 0.000156,
-      "loss": 6.8905,
-      "step": 265
+      "loss": 6.9434,
+      "step": 264
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 0.00015659999999999998,
-      "loss": 7.0416,
-      "step": 266
+      "loss": 6.9478,
+      "step": 265
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.0001572,
-      "loss": 7.0372,
-      "step": 267
+      "loss": 6.761,
+      "step": 266
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.0001578,
-      "loss": 6.9921,
-      "step": 268
+      "loss": 6.9736,
+      "step": 267
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.0001584,
-      "loss": 6.888,
-      "step": 269
+      "loss": 6.7987,
+      "step": 268
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 0.000159,
-      "loss": 6.9847,
-      "step": 270
+      "loss": 6.9504,
+      "step": 269
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 0.0001596,
-      "loss": 6.8239,
-      "step": 271
+      "loss": 6.9853,
+      "step": 270
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 0.0001602,
-      "loss": 6.8453,
-      "step": 272
+      "loss": 6.8711,
+      "step": 271
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 0.0001608,
-      "loss": 6.8599,
-      "step": 273
+      "loss": 6.864,
+      "step": 272
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 0.0001614,
-      "loss": 6.7597,
-      "step": 274
+      "loss": 6.779,
+      "step": 273
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 0.000162,
-      "loss": 7.0201,
-      "step": 275
+      "loss": 6.7297,
+      "step": 274
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.0001626,
-      "loss": 6.873,
-      "step": 276
+      "loss": 6.968,
+      "step": 275
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.0001632,
-      "loss": 6.9579,
-      "step": 277
+      "loss": 6.8534,
+      "step": 276
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.0001638,
-      "loss": 6.8859,
-      "step": 278
+      "loss": 6.8285,
+      "step": 277
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 0.0001644,
-      "loss": 6.9457,
-      "step": 279
+      "loss": 6.9105,
+      "step": 278
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 0.000165,
-      "loss": 7.0055,
-      "step": 280
+      "loss": 6.9451,
+      "step": 279
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 0.0001656,
-      "loss": 6.7921,
-      "step": 281
+      "loss": 6.9857,
+      "step": 280
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 0.0001662,
-      "loss": 6.9926,
-      "step": 282
+      "loss": 6.9182,
+      "step": 281
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 0.0001668,
-      "loss": 6.6923,
-      "step": 283
+      "loss": 6.852,
+      "step": 282
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 0.0001674,
-      "loss": 7.0047,
-      "step": 284
+      "loss": 6.8052,
+      "step": 283
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.000168,
-      "loss": 6.9741,
-      "step": 285
+      "loss": 6.856,
+      "step": 284
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.0001686,
-      "loss": 6.7653,
-      "step": 286
+      "loss": 6.871,
+      "step": 285
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.00016919999999999997,
-      "loss": 6.9668,
-      "step": 287
+      "loss": 6.95,
+      "step": 286
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 0.00016979999999999998,
-      "loss": 6.8915,
-      "step": 288
+      "loss": 6.907,
+      "step": 287
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 0.00017039999999999997,
-      "loss": 7.0164,
-      "step": 289
+      "loss": 6.8627,
+      "step": 288
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 0.00017099999999999998,
-      "loss": 7.0853,
-      "step": 290
+      "loss": 6.8602,
+      "step": 289
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 0.00017159999999999997,
-      "loss": 7.002,
-      "step": 291
+      "loss": 6.9221,
+      "step": 290
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 0.00017219999999999998,
-      "loss": 6.922,
-      "step": 292
+      "loss": 6.9727,
+      "step": 291
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 0.00017279999999999997,
-      "loss": 6.8444,
-      "step": 293
+      "loss": 6.9088,
+      "step": 292
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017339999999999996,
-      "loss": 6.7917,
-      "step": 294
+      "loss": 6.7969,
+      "step": 293
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017399999999999997,
-      "loss": 6.6621,
-      "step": 295
+      "loss": 6.7438,
+      "step": 294
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017459999999999996,
-      "loss": 6.7716,
-      "step": 296
+      "loss": 6.7324,
+      "step": 295
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 0.00017519999999999998,
-      "loss": 6.7518,
-      "step": 297
+      "loss": 6.732,
+      "step": 296
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 0.00017579999999999996,
-      "loss": 6.5265,
-      "step": 298
+      "loss": 6.7565,
+      "step": 297
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 0.00017639999999999998,
-      "loss": 6.5202,
-      "step": 299
+      "loss": 6.7671,
+      "step": 298
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 0.00017699999999999997,
-      "loss": 6.3254,
-      "step": 300
+      "loss": 6.3662,
+      "step": 299
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 0.00017759999999999998,
-      "loss": 7.0382,
-      "step": 301
+      "loss": 6.4025,
+      "step": 300
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 0.00017819999999999997,
-      "loss": 7.019,
-      "step": 302
+      "loss": 7.0981,
+      "step": 301
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00017879999999999998,
-      "loss": 6.9302,
-      "step": 303
+      "loss": 7.0077,
+      "step": 302
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00017939999999999997,
-      "loss": 7.0206,
-      "step": 304
+      "loss": 6.9916,
+      "step": 303
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00017999999999999998,
-      "loss": 6.7929,
-      "step": 305
+      "loss": 6.8943,
+      "step": 304
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 0.00018059999999999997,
-      "loss": 6.8614,
-      "step": 306
+      "loss": 6.9923,
+      "step": 305
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 0.00018119999999999999,
-      "loss": 6.9282,
-      "step": 307
+      "loss": 6.8215,
+      "step": 306
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 0.00018179999999999997,
-      "loss": 6.9358,
-      "step": 308
+      "loss": 6.8283,
+      "step": 307
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 0.0001824,
-      "loss": 6.8621,
-      "step": 309
+      "loss": 6.9746,
+      "step": 308
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 0.00018299999999999998,
-      "loss": 6.9782,
-      "step": 310
+      "loss": 6.8177,
+      "step": 309
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.0001836,
-      "loss": 7.0606,
-      "step": 311
+      "loss": 6.9226,
+      "step": 310
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.00018419999999999998,
-      "loss": 6.9457,
-      "step": 312
+      "loss": 6.9195,
+      "step": 311
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.0001848,
-      "loss": 6.9702,
-      "step": 313
+      "loss": 6.9117,
+      "step": 312
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.00018539999999999998,
-      "loss": 6.9048,
-      "step": 314
+      "loss": 6.8309,
+      "step": 313
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 0.000186,
-      "loss": 6.796,
-      "step": 315
+      "loss": 6.8095,
+      "step": 314
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 0.00018659999999999998,
-      "loss": 7.0186,
-      "step": 316
+      "loss": 6.942,
+      "step": 315
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 0.0001872,
-      "loss": 6.9154,
-      "step": 317
+      "loss": 6.8752,
+      "step": 316
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 0.00018779999999999998,
-      "loss": 6.8215,
-      "step": 318
+      "loss": 6.8158,
+      "step": 317
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 0.00018839999999999997,
-      "loss": 6.7335,
-      "step": 319
+      "loss": 6.7456,
+      "step": 318
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.00018899999999999999,
-      "loss": 6.9722,
-      "step": 320
+      "loss": 6.8388,
+      "step": 319
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.00018959999999999997,
-      "loss": 6.922,
-      "step": 321
+      "loss": 6.8635,
+      "step": 320
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.0001902,
-      "loss": 6.7933,
-      "step": 322
+      "loss": 6.8421,
+      "step": 321
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.00019079999999999998,
-      "loss": 6.9454,
-      "step": 323
+      "loss": 6.8038,
+      "step": 322
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 0.0001914,
-      "loss": 6.9225,
-      "step": 324
+      "loss": 6.9022,
+      "step": 323
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 0.00019199999999999998,
-      "loss": 6.9729,
-      "step": 325
+      "loss": 6.8039,
+      "step": 324
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 0.0001926,
-      "loss": 6.9193,
-      "step": 326
+      "loss": 6.88,
+      "step": 325
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 0.00019319999999999998,
-      "loss": 6.8074,
-      "step": 327
+      "loss": 6.9172,
+      "step": 326
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 0.0001938,
-      "loss": 6.9116,
-      "step": 328
+      "loss": 6.8517,
+      "step": 327
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.00019439999999999998,
-      "loss": 6.9647,
-      "step": 329
+      "loss": 6.8168,
+      "step": 328
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.000195,
-      "loss": 6.9612,
-      "step": 330
+      "loss": 6.8759,
+      "step": 329
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.00019559999999999998,
-      "loss": 6.9816,
-      "step": 331
+      "loss": 6.9028,
+      "step": 330
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.0001962,
-      "loss": 6.7078,
-      "step": 332
+      "loss": 6.8595,
+      "step": 331
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 0.00019679999999999999,
-      "loss": 6.8303,
-      "step": 333
+      "loss": 6.7604,
+      "step": 332
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 0.0001974,
-      "loss": 6.9353,
-      "step": 334
+      "loss": 6.9378,
+      "step": 333
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 0.000198,
-      "loss": 6.9853,
-      "step": 335
+      "loss": 6.8531,
+      "step": 334
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 0.0001986,
-      "loss": 6.7819,
-      "step": 336
+      "loss": 6.9461,
+      "step": 335
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 0.0001992,
-      "loss": 6.7857,
-      "step": 337
+      "loss": 6.8454,
+      "step": 336
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.0001998,
-      "loss": 6.9085,
-      "step": 338
+      "loss": 6.7238,
+      "step": 337
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.0002004,
-      "loss": 6.9153,
-      "step": 339
+      "loss": 6.871,
+      "step": 338
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.000201,
-      "loss": 6.8557,
-      "step": 340
+      "loss": 6.8716,
+      "step": 339
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.0002016,
-      "loss": 6.9274,
-      "step": 341
+      "loss": 6.8667,
+      "step": 340
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 0.0002022,
-      "loss": 6.8212,
-      "step": 342
+      "loss": 6.7651,
+      "step": 341
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 0.0002028,
-      "loss": 6.7922,
-      "step": 343
+      "loss": 6.8711,
+      "step": 342
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 0.00020339999999999998,
-      "loss": 6.7738,
-      "step": 344
+      "loss": 6.7457,
+      "step": 343
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 0.000204,
-      "loss": 6.6782,
-      "step": 345
+      "loss": 6.712,
+      "step": 344
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 0.00020459999999999999,
-      "loss": 6.765,
-      "step": 346
+      "loss": 6.7178,
+      "step": 345
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.0002052,
-      "loss": 6.7327,
-      "step": 347
+      "loss": 6.7846,
+      "step": 346
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.0002058,
-      "loss": 6.7604,
-      "step": 348
+      "loss": 6.6471,
+      "step": 347
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020639999999999998,
-      "loss": 6.6491,
-      "step": 349
+      "loss": 6.4892,
+      "step": 348
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020699999999999996,
-      "loss": 6.5297,
-      "step": 350
+      "loss": 6.5734,
+      "step": 349
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 0.00020759999999999998,
-      "loss": 6.9381,
-      "step": 351
+      "loss": 6.2744,
+      "step": 350
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 0.00020819999999999996,
-      "loss": 6.9526,
-      "step": 352
+      "loss": 6.9985,
+      "step": 351
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 0.00020879999999999998,
-      "loss": 6.9151,
-      "step": 353
+      "loss": 7.0057,
+      "step": 352
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 0.00020939999999999997,
-      "loss": 6.8661,
-      "step": 354
+      "loss": 6.8872,
+      "step": 353
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 0.00020999999999999998,
-      "loss": 6.9779,
-      "step": 355
+      "loss": 6.8159,
+      "step": 354
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021059999999999997,
-      "loss": 6.8233,
-      "step": 356
+      "loss": 6.8683,
+      "step": 355
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021119999999999996,
-      "loss": 7.0543,
-      "step": 357
+      "loss": 6.9613,
+      "step": 356
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021179999999999997,
-      "loss": 6.8327,
-      "step": 358
+      "loss": 6.8534,
+      "step": 357
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021239999999999996,
-      "loss": 6.937,
-      "step": 359
+      "loss": 6.9346,
+      "step": 358
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.00021299999999999997,
-      "loss": 6.7641,
-      "step": 360
+      "loss": 6.8493,
+      "step": 359
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 0.00021359999999999996,
-      "loss": 6.8524,
-      "step": 361
+      "loss": 6.8302,
+      "step": 360
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 0.00021419999999999998,
-      "loss": 6.9425,
-      "step": 362
+      "loss": 6.8248,
+      "step": 361
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 0.00021479999999999996,
-      "loss": 6.8399,
-      "step": 363
+      "loss": 6.7857,
+      "step": 362
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 0.00021539999999999998,
-      "loss": 6.8941,
-      "step": 364
+      "loss": 6.7263,
+      "step": 363
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021599999999999996,
-      "loss": 6.8468,
-      "step": 365
+      "loss": 6.7746,
+      "step": 364
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021659999999999998,
-      "loss": 6.7709,
-      "step": 366
+      "loss": 6.8748,
+      "step": 365
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021719999999999997,
-      "loss": 6.8377,
-      "step": 367
+      "loss": 6.8155,
+      "step": 366
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 0.00021779999999999998,
-      "loss": 6.8153,
-      "step": 368
+      "loss": 6.9323,
+      "step": 367
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 0.00021839999999999997,
-      "loss": 6.7531,
-      "step": 369
+      "loss": 6.8714,
+      "step": 368
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 0.00021899999999999998,
-      "loss": 6.9445,
-      "step": 370
+      "loss": 6.7399,
+      "step": 369
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 0.00021959999999999997,
-      "loss": 6.8509,
-      "step": 371
+      "loss": 6.8,
+      "step": 370
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 0.00022019999999999999,
-      "loss": 7.0474,
-      "step": 372
+      "loss": 6.8585,
+      "step": 371
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 0.00022079999999999997,
-      "loss": 6.8976,
-      "step": 373
+      "loss": 6.8913,
+      "step": 372
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.0002214,
-      "loss": 6.956,
-      "step": 374
+      "loss": 6.7891,
+      "step": 373
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.00022199999999999998,
-      "loss": 6.7783,
-      "step": 375
+      "loss": 6.6911,
+      "step": 374
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.0002226,
-      "loss": 6.8574,
-      "step": 376
+      "loss": 6.8265,
+      "step": 375
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 0.00022319999999999998,
-      "loss": 6.7116,
-      "step": 377
+      "loss": 6.7551,
+      "step": 376
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 0.0002238,
-      "loss": 6.8273,
-      "step": 378
+      "loss": 6.8387,
+      "step": 377
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 0.00022439999999999998,
-      "loss": 6.8252,
-      "step": 379
+      "loss": 6.7973,
+      "step": 378
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 0.000225,
-      "loss": 6.9099,
-      "step": 380
+      "loss": 6.795,
+      "step": 379
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 0.00022559999999999998,
-      "loss": 6.9029,
-      "step": 381
+      "loss": 6.8265,
+      "step": 380
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 0.00022619999999999997,
-      "loss": 6.7356,
-      "step": 382
+      "loss": 6.9408,
+      "step": 381
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.00022679999999999998,
-      "loss": 6.935,
-      "step": 383
+      "loss": 6.8602,
+      "step": 382
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.00022739999999999997,
-      "loss": 6.9279,
-      "step": 384
+      "loss": 6.848,
+      "step": 383
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.00022799999999999999,
-      "loss": 6.7585,
-      "step": 385
+      "loss": 6.8098,
+      "step": 384
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 0.00022859999999999997,
-      "loss": 6.8771,
-      "step": 386
+      "loss": 6.8624,
+      "step": 385
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 0.0002292,
-      "loss": 6.8192,
-      "step": 387
+      "loss": 6.8417,
+      "step": 386
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 0.00022979999999999997,
-      "loss": 6.8009,
-      "step": 388
+      "loss": 6.7887,
+      "step": 387
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 0.0002304,
-      "loss": 6.8385,
-      "step": 389
+      "loss": 6.8669,
+      "step": 388
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 0.00023099999999999998,
-      "loss": 7.0575,
-      "step": 390
+      "loss": 6.8086,
+      "step": 389
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 0.0002316,
-      "loss": 6.8882,
-      "step": 391
+      "loss": 6.7165,
+      "step": 390
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.00023219999999999998,
-      "loss": 6.8121,
-      "step": 392
+      "loss": 6.7425,
+      "step": 391
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.0002328,
-      "loss": 6.8311,
-      "step": 393
+      "loss": 6.7985,
+      "step": 392
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.00023339999999999998,
-      "loss": 7.0132,
-      "step": 394
+      "loss": 6.71,
+      "step": 393
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 0.000234,
-      "loss": 6.7881,
-      "step": 395
+      "loss": 6.716,
+      "step": 394
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 0.00023459999999999998,
-      "loss": 6.6595,
-      "step": 396
+      "loss": 6.7783,
+      "step": 395
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 0.0002352,
-      "loss": 6.5624,
-      "step": 397
+      "loss": 6.7704,
+      "step": 396
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 0.00023579999999999999,
-      "loss": 6.6603,
-      "step": 398
+      "loss": 6.7056,
+      "step": 397
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 0.0002364,
-      "loss": 6.6189,
-      "step": 399
+      "loss": 6.5548,
+      "step": 398
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 0.000237,
-      "loss": 6.4334,
-      "step": 400
+      "loss": 6.6221,
+      "step": 399
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.0002376,
-      "loss": 7.0859,
-      "step": 401
+      "loss": 6.276,
+      "step": 400
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.0002382,
-      "loss": 7.0288,
-      "step": 402
+      "loss": 6.9953,
+      "step": 401
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.0002388,
-      "loss": 6.998,
-      "step": 403
+      "loss": 6.8796,
+      "step": 402
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 0.0002394,
-      "loss": 6.8879,
-      "step": 404
+      "loss": 7.0043,
+      "step": 403
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 0.00023999999999999998,
-      "loss": 7.0649,
-      "step": 405
+      "loss": 6.8759,
+      "step": 404
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 0.0002406,
-      "loss": 6.8729,
-      "step": 406
+      "loss": 6.7594,
+      "step": 405
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 0.00024119999999999998,
-      "loss": 6.9037,
-      "step": 407
+      "loss": 6.886,
+      "step": 406
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 0.0002418,
-      "loss": 6.8431,
-      "step": 408
+      "loss": 6.8996,
+      "step": 407
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 0.00024239999999999998,
-      "loss": 7.0446,
-      "step": 409
+      "loss": 6.8223,
+      "step": 408
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.000243,
-      "loss": 6.9666,
-      "step": 410
+      "loss": 6.7938,
+      "step": 409
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.00024359999999999999,
-      "loss": 6.8929,
-      "step": 411
+      "loss": 6.8277,
+      "step": 410
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.00024419999999999997,
-      "loss": 6.8846,
-      "step": 412
+      "loss": 6.9263,
+      "step": 411
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 0.0002448,
-      "loss": 6.8704,
-      "step": 413
+      "loss": 6.8408,
+      "step": 412
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 0.00024539999999999995,
-      "loss": 6.7452,
-      "step": 414
+      "loss": 6.8565,
+      "step": 413
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 0.00024599999999999996,
-      "loss": 6.7961,
-      "step": 415
+      "loss": 6.8711,
+      "step": 414
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 0.0002466,
-      "loss": 6.9163,
-      "step": 416
+      "loss": 6.9096,
+      "step": 415
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 0.0002472,
-      "loss": 6.8202,
-      "step": 417
+      "loss": 6.852,
+      "step": 416
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 0.00024779999999999995,
-      "loss": 6.8721,
-      "step": 418
+      "loss": 6.8442,
+      "step": 417
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.00024839999999999997,
-      "loss": 6.8826,
-      "step": 419
+      "loss": 6.7525,
+      "step": 418
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.000249,
-      "loss": 6.884,
-      "step": 420
+      "loss": 6.7989,
+      "step": 419
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.00024959999999999994,
-      "loss": 6.8824,
-      "step": 421
+      "loss": 6.8196,
+      "step": 420
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 0.00025019999999999996,
-      "loss": 6.9743,
-      "step": 422
+      "loss": 6.8717,
+      "step": 421
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 0.00025079999999999997,
-      "loss": 6.8894,
-      "step": 423
+      "loss": 6.7613,
+      "step": 422
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 0.0002514,
-      "loss": 6.8829,
-      "step": 424
+      "loss": 6.8156,
+      "step": 423
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 0.00025199999999999995,
-      "loss": 6.9305,
-      "step": 425
+      "loss": 6.9202,
+      "step": 424
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 0.00025259999999999996,
-      "loss": 7.0412,
-      "step": 426
+      "loss": 6.7485,
+      "step": 425
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.0002532,
-      "loss": 6.788,
-      "step": 427
+      "loss": 6.8243,
+      "step": 426
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.0002538,
-      "loss": 6.9715,
-      "step": 428
+      "loss": 6.7921,
+      "step": 427
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.00025439999999999995,
-      "loss": 6.8377,
-      "step": 429
+      "loss": 6.7203,
+      "step": 428
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.00025499999999999996,
-      "loss": 6.8427,
-      "step": 430
+      "loss": 6.6956,
+      "step": 429
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 0.0002556,
-      "loss": 6.996,
-      "step": 431
+      "loss": 6.7347,
+      "step": 430
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.97,
       "learning_rate": 0.0002562,
-      "loss": 6.9316,
-      "step": 432
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00025679999999999995,
-      "loss": 6.5887,
-      "step": 433
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00025739999999999997,
-      "loss": 6.8635,
-      "step": 434
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000258,
-      "loss": 6.767,
-      "step": 435
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0002586,
-      "loss": 6.9778,
-      "step": 436
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00025919999999999996,
-      "loss": 6.8633,
-      "step": 437
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00025979999999999997,
-      "loss": 6.9399,
-      "step": 438
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0002604,
-      "loss": 6.7401,
-      "step": 439
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000261,
-      "loss": 6.8552,
-      "step": 440
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00026159999999999996,
-      "loss": 6.9278,
-      "step": 441
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0002622,
-      "loss": 6.8031,
-      "step": 442
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0002628,
-      "loss": 6.7692,
-      "step": 443
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.00026339999999999995,
-      "loss": 6.9128,
-      "step": 444
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.00026399999999999997,
-      "loss": 6.6816,
-      "step": 445
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0002646,
-      "loss": 6.6995,
-      "step": 446
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0002652,
-      "loss": 6.6771,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.00026579999999999996,
-      "loss": 6.6629,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.00026639999999999997,
-      "loss": 6.3252,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000267,
-      "loss": 6.3625,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002676,
-      "loss": 6.9431,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.00026819999999999996,
-      "loss": 7.0696,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002688,
-      "loss": 7.0835,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002694,
-      "loss": 6.88,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.00027,
-      "loss": 6.98,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.00027059999999999996,
-      "loss": 7.04,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002712,
-      "loss": 6.8282,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002718,
-      "loss": 6.9422,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0002724,
-      "loss": 6.9104,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.00027299999999999997,
-      "loss": 7.0045,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002736,
-      "loss": 6.8041,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002742,
-      "loss": 6.9422,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002748,
-      "loss": 6.8194,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.00027539999999999997,
-      "loss": 6.7778,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.000276,
-      "loss": 6.9162,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002766,
-      "loss": 6.8971,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002772,
-      "loss": 6.8153,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0002778,
-      "loss": 6.9018,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002784,
-      "loss": 6.8423,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.000279,
-      "loss": 6.9256,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00027959999999999997,
-      "loss": 6.8025,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002802,
-      "loss": 6.7987,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002808,
-      "loss": 6.8809,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00028139999999999996,
-      "loss": 6.7699,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 6.9145,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0002826,
-      "loss": 6.9334,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00028319999999999994,
-      "loss": 7.0333,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028379999999999996,
-      "loss": 6.8036,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0002844,
-      "loss": 6.7853,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000285,
-      "loss": 6.8205,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028559999999999995,
-      "loss": 6.7719,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028619999999999996,
-      "loss": 6.8131,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0002868,
-      "loss": 6.827,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028739999999999994,
-      "loss": 6.8999,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028799999999999995,
-      "loss": 6.8928,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00028859999999999997,
-      "loss": 6.8204,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0002892,
-      "loss": 6.8927,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00028979999999999994,
-      "loss": 6.7876,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029039999999999996,
-      "loss": 7.011,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029099999999999997,
-      "loss": 6.9904,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0002916,
-      "loss": 6.7925,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029219999999999995,
-      "loss": 6.8486,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00029279999999999996,
-      "loss": 6.8364,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0002934,
-      "loss": 6.7825,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000294,
-      "loss": 6.9096,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00029459999999999995,
-      "loss": 6.6241,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00029519999999999997,
-      "loss": 6.3106,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0002958,
-      "loss": 7.0315,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0002964,
-      "loss": 6.7356,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00029699999999999996,
-      "loss": 6.2901,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 6.832920551300049,
-      "eval_runtime": 362.0157,
-      "eval_samples_per_second": 7.298,
-      "eval_steps_per_second": 0.914,
-      "eval_wer": 1.0,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00029759999999999997,
-      "loss": 7.2547,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0002982,
-      "loss": 6.8181,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0002988,
-      "loss": 6.9446,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029939999999999996,
-      "loss": 7.0057,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0003,
-      "loss": 6.976,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.000299234693877551,
-      "loss": 6.9817,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029846938775510205,
-      "loss": 6.9377,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029770408163265304,
-      "loss": 6.8353,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002969387755102041,
-      "loss": 6.8738,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00029617346938775506,
-      "loss": 6.9288,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002954081632653061,
-      "loss": 6.8304,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0002946428571428571,
-      "loss": 6.9171,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002938775510204081,
-      "loss": 7.022,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.00029311224489795917,
-      "loss": 6.9531,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002923469387755102,
-      "loss": 6.8854,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002915816326530612,
-      "loss": 6.7042,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.00029081632653061223,
-      "loss": 6.9831,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002900510204081632,
-      "loss": 6.7991,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.00028928571428571425,
-      "loss": 6.9969,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002885204081632653,
-      "loss": 6.9657,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0002877551020408163,
-      "loss": 6.7878,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002869897959183673,
-      "loss": 6.8026,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00028622448979591836,
-      "loss": 6.7209,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002854591836734694,
-      "loss": 6.8029,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002846938775510204,
-      "loss": 6.9162,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00028392857142857137,
-      "loss": 6.7592,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002831632653061224,
-      "loss": 6.8772,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00028239795918367345,
-      "loss": 6.8832,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0002816326530612245,
-      "loss": 6.9216,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.00028086734693877547,
-      "loss": 6.9836,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002801020408163265,
-      "loss": 6.8133,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027933673469387755,
-      "loss": 6.781,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027857142857142854,
-      "loss": 6.7889,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002778061224489796,
-      "loss": 6.7747,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027704081632653056,
-      "loss": 6.8035,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002762755102040816,
-      "loss": 6.9291,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027551020408163264,
-      "loss": 6.8163,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0002747448979591837,
-      "loss": 6.9158,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.00027397959183673466,
-      "loss": 6.9038,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002732142857142857,
-      "loss": 6.8366,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002724489795918367,
-      "loss": 6.9376,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00027168367346938773,
-      "loss": 6.8482,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002709183673469387,
-      "loss": 6.7671,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00027015306122448975,
-      "loss": 6.6458,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0002693877551020408,
-      "loss": 6.7696,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00026862244897959183,
-      "loss": 6.6372,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00026785714285714287,
-      "loss": 6.9219,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.00026709183673469386,
-      "loss": 6.6973,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002663265306122449,
-      "loss": 6.3204,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002655612244897959,
-      "loss": 6.5614,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002647959183673469,
-      "loss": 6.9267,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002640306122448979,
-      "loss": 7.09,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.00026326530612244894,
-      "loss": 6.9054,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0002625,
-      "loss": 6.8814,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.000261734693877551,
-      "loss": 6.8698,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.000260969387755102,
-      "loss": 6.8921,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.00026020408163265305,
-      "loss": 6.8406,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.00025943877551020403,
-      "loss": 6.9311,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.00025867346938775507,
-      "loss": 6.8586,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002579081632653061,
-      "loss": 6.7891,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002571428571428571,
-      "loss": 6.8937,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.00025637755102040814,
-      "loss": 6.7656,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002556122448979592,
-      "loss": 6.8204,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002548469387755102,
-      "loss": 6.8283,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002540816326530612,
-      "loss": 7.0003,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0002533163265306122,
-      "loss": 6.6763,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002525510204081632,
-      "loss": 6.845,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00025178571428571426,
-      "loss": 6.7038,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002510204081632653,
-      "loss": 6.786,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002502551020408163,
-      "loss": 6.8934,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024948979591836733,
-      "loss": 6.768,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024872448979591837,
-      "loss": 6.8869,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.00024795918367346935,
-      "loss": 6.7311,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002471938775510204,
-      "loss": 6.6567,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0002464285714285714,
-      "loss": 6.8348,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0002456632653061224,
-      "loss": 6.6822,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024489795918367346,
-      "loss": 6.8552,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024413265306122447,
-      "loss": 6.9091,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024336734693877548,
-      "loss": 6.7568,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0002426020408163265,
-      "loss": 6.9483,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024183673469387753,
-      "loss": 6.9424,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024107142857142857,
-      "loss": 6.7909,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00024030612244897956,
-      "loss": 6.7228,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.00023954081632653057,
-      "loss": 6.8291,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0002387755102040816,
-      "loss": 6.8442,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023801020408163265,
-      "loss": 6.8812,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023724489795918366,
-      "loss": 6.8635,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023647959183673467,
-      "loss": 6.8035,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023571428571428569,
-      "loss": 6.7253,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023494897959183673,
-      "loss": 6.7687,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023418367346938774,
-      "loss": 6.8081,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023341836734693875,
-      "loss": 6.8181,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.00023265306122448976,
-      "loss": 6.9183,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0002318877551020408,
-      "loss": 6.605,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00023112244897959181,
-      "loss": 6.7087,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00023035714285714285,
-      "loss": 6.5905,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022959183673469384,
-      "loss": 6.5806,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022882653061224488,
-      "loss": 6.497,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0002280612244897959,
-      "loss": 6.3569,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022729591836734693,
-      "loss": 6.036,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022653061224489791,
-      "loss": 6.993,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00022576530612244895,
-      "loss": 6.7804,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.000225,
-      "loss": 6.9282,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.000224234693877551,
-      "loss": 6.8609,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022346938775510205,
-      "loss": 6.9853,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022270408163265303,
-      "loss": 6.6673,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022193877551020407,
-      "loss": 6.8504,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022117346938775508,
-      "loss": 6.8609,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00022040816326530612,
-      "loss": 6.8363,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0002196428571428571,
-      "loss": 6.8206,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.00021887755102040815,
-      "loss": 6.8077,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021811224489795916,
-      "loss": 6.8301,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002173469387755102,
-      "loss": 6.7382,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002165816326530612,
-      "loss": 6.839,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021581632653061222,
-      "loss": 6.9974,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021505102040816324,
-      "loss": 6.6923,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.00021428571428571427,
-      "loss": 6.8107,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002135204081632653,
-      "loss": 6.658,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0002127551020408163,
-      "loss": 6.7308,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002119897959183673,
-      "loss": 6.8192,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00021122448979591835,
-      "loss": 6.8473,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002104591836734694,
-      "loss": 6.7264,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002096938775510204,
-      "loss": 6.9049,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002089285714285714,
-      "loss": 6.7086,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020816326530612243,
-      "loss": 6.8953,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020739795918367347,
-      "loss": 6.7709,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.00020663265306122448,
-      "loss": 6.8614,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0002058673469387755,
-      "loss": 6.8522,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0002051020408163265,
-      "loss": 6.8211,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020433673469387754,
-      "loss": 6.9599,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020357142857142856,
-      "loss": 6.7826,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020280612244897957,
-      "loss": 6.6684,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020204081632653058,
-      "loss": 6.8832,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020127551020408162,
-      "loss": 6.7759,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00020051020408163263,
-      "loss": 6.8656,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00019974489795918367,
-      "loss": 6.8239,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.00019897959183673466,
-      "loss": 6.8342,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0001982142857142857,
-      "loss": 6.6465,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0001974489795918367,
-      "loss": 6.9261,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019668367346938775,
-      "loss": 6.671,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019591836734693873,
-      "loss": 6.7317,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019515306122448977,
-      "loss": 6.922,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0001943877551020408,
-      "loss": 6.8189,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019362244897959182,
-      "loss": 6.5989,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019285714285714286,
-      "loss": 6.5903,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.00019209183673469385,
-      "loss": 6.7557,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0001913265306122449,
-      "loss": 6.5452,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0001905612244897959,
-      "loss": 6.4467,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018979591836734694,
-      "loss": 6.6444,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018903061224489793,
-      "loss": 6.3974,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018826530612244896,
-      "loss": 7.0178,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018749999999999998,
-      "loss": 6.9998,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018673469387755102,
-      "loss": 6.8985,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018596938775510203,
-      "loss": 6.8224,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00018520408163265304,
-      "loss": 6.8712,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00018443877551020405,
-      "loss": 6.8948,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0001836734693877551,
-      "loss": 6.7172,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0001829081632653061,
-      "loss": 6.9833,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00018214285714285712,
-      "loss": 6.7223,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00018137755102040813,
-      "loss": 6.722,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00018061224489795917,
-      "loss": 6.7625,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0001798469387755102,
-      "loss": 6.8163,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00017908163265306122,
-      "loss": 6.9104,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0001783163265306122,
-      "loss": 6.6764,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017755102040816325,
-      "loss": 6.7642,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017678571428571428,
-      "loss": 6.8771,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001760204081632653,
-      "loss": 6.7731,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001752551020408163,
-      "loss": 6.7133,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017448979591836732,
-      "loss": 6.8526,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017372448979591836,
-      "loss": 6.7391,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00017295918367346937,
-      "loss": 6.6748,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001721938775510204,
-      "loss": 6.8101,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0001714285714285714,
-      "loss": 6.7354,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00017066326530612244,
-      "loss": 6.7228,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016989795918367345,
-      "loss": 6.8262,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0001691326530612245,
-      "loss": 6.8924,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016836734693877547,
-      "loss": 6.765,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016760204081632651,
-      "loss": 6.5957,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016683673469387753,
-      "loss": 6.7606,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016607142857142857,
-      "loss": 6.7172,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00016530612244897955,
-      "loss": 6.8478,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0001645408163265306,
-      "loss": 6.7654,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00016377551020408163,
-      "loss": 6.7287,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00016301020408163264,
-      "loss": 6.8379,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00016224489795918368,
-      "loss": 6.8713,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00016147959183673467,
-      "loss": 6.6862,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0001607142857142857,
-      "loss": 6.6621,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015994897959183672,
-      "loss": 6.9119,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015918367346938776,
-      "loss": 6.7827,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015841836734693874,
-      "loss": 6.8166,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00015765306122448978,
-      "loss": 6.6425,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0001568877551020408,
-      "loss": 6.7403,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015612244897959183,
-      "loss": 6.5451,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015535714285714285,
-      "loss": 6.7156,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015459183673469386,
-      "loss": 6.7341,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015382653061224487,
-      "loss": 6.7433,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0001530612244897959,
-      "loss": 6.5209,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015229591836734692,
-      "loss": 6.3372,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015153061224489794,
-      "loss": 6.4115,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00015076530612244895,
-      "loss": 6.0466,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00015,
-      "loss": 6.8879,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014923469387755103,
-      "loss": 6.8209,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014846938775510204,
-      "loss": 6.767,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014770408163265305,
-      "loss": 6.6467,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014693877551020406,
-      "loss": 6.771,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0001461734693877551,
-      "loss": 6.8762,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014540816326530611,
-      "loss": 6.8515,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014464285714285713,
-      "loss": 6.9306,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00014387755102040814,
-      "loss": 6.8259,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00014311224489795918,
-      "loss": 6.7757,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0001423469387755102,
-      "loss": 6.753,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0001415816326530612,
-      "loss": 6.6252,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00014081632653061224,
-      "loss": 6.6745,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00014005102040816326,
-      "loss": 6.7449,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013928571428571427,
-      "loss": 6.8163,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013852040816326528,
-      "loss": 6.8405,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013775510204081632,
-      "loss": 7.0215,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00013698979591836733,
-      "loss": 6.7862,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013622448979591834,
-      "loss": 6.6331,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013545918367346936,
-      "loss": 6.717,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0001346938775510204,
-      "loss": 6.8046,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013392857142857144,
-      "loss": 6.685,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013316326530612245,
-      "loss": 6.6772,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013239795918367346,
-      "loss": 6.8847,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00013163265306122447,
-      "loss": 6.7749,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0001308673469387755,
-      "loss": 6.8872,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00013010204081632652,
-      "loss": 6.8119,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012933673469387754,
-      "loss": 6.676,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012857142857142855,
-      "loss": 6.7485,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0001278061224489796,
-      "loss": 6.8757,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0001270408163265306,
-      "loss": 7.0093,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0001262755102040816,
-      "loss": 6.8234,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012551020408163265,
-      "loss": 6.7583,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012474489795918366,
-      "loss": 6.9749,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00012397959183673468,
-      "loss": 6.7102,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0001232142857142857,
-      "loss": 6.7685,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00012244897959183673,
-      "loss": 6.7942,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00012168367346938774,
-      "loss": 6.8079,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00012091836734693877,
-      "loss": 6.695,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00012015306122448978,
-      "loss": 6.8162,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0001193877551020408,
-      "loss": 6.7592,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011862244897959183,
-      "loss": 6.79,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011785714285714284,
-      "loss": 6.6227,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00011709183673469387,
-      "loss": 6.7437,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011632653061224488,
-      "loss": 6.6668,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011556122448979591,
-      "loss": 6.8181,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011479591836734692,
-      "loss": 6.4942,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011403061224489795,
-      "loss": 6.4618,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011326530612244896,
-      "loss": 6.4448,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0001125,
-      "loss": 6.3197,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011173469387755102,
-      "loss": 6.8133,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011096938775510204,
-      "loss": 6.8236,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00011020408163265306,
-      "loss": 6.8616,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010943877551020407,
-      "loss": 6.7366,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0001086734693877551,
-      "loss": 6.7812,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010790816326530611,
-      "loss": 6.8787,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010714285714285714,
-      "loss": 6.7598,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010637755102040815,
-      "loss": 6.6895,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010561224489795918,
-      "loss": 6.7956,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0001048469387755102,
-      "loss": 6.724,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010408163265306121,
-      "loss": 6.7332,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00010331632653061224,
-      "loss": 6.7386,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00010255102040816325,
-      "loss": 6.5965,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00010178571428571428,
-      "loss": 6.5961,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00010102040816326529,
-      "loss": 6.7318,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00010025510204081632,
-      "loss": 6.6829,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.948979591836733e-05,
-      "loss": 6.6803,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.872448979591835e-05,
-      "loss": 6.7533,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.795918367346937e-05,
-      "loss": 6.8555,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.71938775510204e-05,
-      "loss": 6.652,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.642857142857143e-05,
-      "loss": 6.8121,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.566326530612244e-05,
-      "loss": 6.8903,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.489795918367347e-05,
-      "loss": 6.8009,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.413265306122448e-05,
-      "loss": 6.6108,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.336734693877551e-05,
-      "loss": 6.5482,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.260204081632652e-05,
-      "loss": 6.6848,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.183673469387755e-05,
-      "loss": 6.6021,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.107142857142856e-05,
-      "loss": 6.7729,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.030612244897958e-05,
-      "loss": 6.761,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.954081632653061e-05,
-      "loss": 6.704,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.877551020408162e-05,
-      "loss": 6.7131,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.801020408163265e-05,
-      "loss": 6.6933,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.724489795918366e-05,
-      "loss": 6.7786,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.647959183673469e-05,
-      "loss": 6.6551,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.57142857142857e-05,
-      "loss": 6.7676,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.494897959183672e-05,
-      "loss": 6.7696,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.418367346938774e-05,
-      "loss": 6.7007,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.341836734693876e-05,
-      "loss": 6.7635,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.265306122448978e-05,
-      "loss": 6.714,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 8.188775510204081e-05,
-      "loss": 6.7076,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 8.112244897959184e-05,
-      "loss": 6.5998,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 8.035714285714285e-05,
-      "loss": 6.5586,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.959183673469388e-05,
-      "loss": 6.649,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.882653061224489e-05,
-      "loss": 6.5968,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.806122448979592e-05,
-      "loss": 6.7391,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.729591836734693e-05,
-      "loss": 6.5806,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.653061224489796e-05,
-      "loss": 6.6103,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.576530612244897e-05,
-      "loss": 6.5708,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.5e-05,
-      "loss": 6.6839,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.423469387755102e-05,
-      "loss": 6.2081,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.346938775510203e-05,
-      "loss": 6.8204,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.270408163265306e-05,
-      "loss": 6.7398,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.193877551020407e-05,
-      "loss": 6.8149,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.11734693877551e-05,
-      "loss": 6.821,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 7.040816326530612e-05,
-      "loss": 6.5948,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.964285714285713e-05,
-      "loss": 6.8221,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.887755102040816e-05,
-      "loss": 6.8279,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.811224489795917e-05,
-      "loss": 6.6116,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.73469387755102e-05,
-      "loss": 6.7841,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.658163265306122e-05,
-      "loss": 6.6926,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.581632653061224e-05,
-      "loss": 6.8074,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.505102040816326e-05,
-      "loss": 6.8235,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.428571428571427e-05,
-      "loss": 6.6634,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.35204081632653e-05,
-      "loss": 6.7408,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.275510204081633e-05,
-      "loss": 6.7874,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 6.198979591836734e-05,
-      "loss": 6.6599,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 6.122448979591836e-05,
-      "loss": 6.8251,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 6.045918367346938e-05,
-      "loss": 6.7198,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.96938775510204e-05,
-      "loss": 6.649,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.892857142857142e-05,
-      "loss": 6.8378,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.816326530612244e-05,
-      "loss": 6.8143,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.739795918367346e-05,
-      "loss": 6.6955,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.663265306122448e-05,
-      "loss": 6.8389,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.586734693877551e-05,
-      "loss": 6.6973,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.510204081632653e-05,
-      "loss": 6.9625,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.433673469387755e-05,
-      "loss": 6.8445,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.357142857142857e-05,
-      "loss": 6.7165,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.280612244897959e-05,
-      "loss": 6.632,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.204081632653061e-05,
-      "loss": 6.6587,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.1275510204081626e-05,
-      "loss": 6.6842,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.0510204081632645e-05,
-      "loss": 6.6688,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.9744897959183664e-05,
-      "loss": 6.699,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.897959183673468e-05,
-      "loss": 6.7232,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.8214285714285716e-05,
-      "loss": 6.7536,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.7448979591836735e-05,
-      "loss": 6.6919,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.6683673469387754e-05,
-      "loss": 6.8364,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.591836734693877e-05,
-      "loss": 6.7028,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.515306122448979e-05,
-      "loss": 6.8399,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.438775510204081e-05,
-      "loss": 6.6795,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.362244897959183e-05,
-      "loss": 6.6221,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.285714285714285e-05,
-      "loss": 6.7963,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.209183673469387e-05,
-      "loss": 6.6097,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 4.132653061224489e-05,
-      "loss": 6.8284,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 4.056122448979592e-05,
-      "loss": 6.6447,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.979591836734694e-05,
-      "loss": 6.6868,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.903061224489796e-05,
-      "loss": 6.5554,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.826530612244898e-05,
-      "loss": 6.413,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.75e-05,
-      "loss": 6.6788,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.6734693877551016e-05,
-      "loss": 6.4647,
-      "step": 849
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.5969387755102035e-05,
-      "loss": 6.2421,
-      "step": 850
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.520408163265306e-05,
-      "loss": 6.8049,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.443877551020408e-05,
-      "loss": 6.6672,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.36734693877551e-05,
-      "loss": 6.9027,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.290816326530612e-05,
-      "loss": 6.8127,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.214285714285714e-05,
-      "loss": 6.7657,
-      "step": 855
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.137755102040816e-05,
-      "loss": 6.8388,
-      "step": 856
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 3.061224489795918e-05,
-      "loss": 6.8636,
-      "step": 857
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.98469387755102e-05,
-      "loss": 6.7972,
-      "step": 858
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.908163265306122e-05,
-      "loss": 6.6306,
-      "step": 859
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.831632653061224e-05,
-      "loss": 6.7981,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.7551020408163265e-05,
-      "loss": 6.8927,
-      "step": 861
+      "loss": 6.7942,
+      "step": 431
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.6785714285714284e-05,
-      "loss": 6.7818,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.6020408163265303e-05,
-      "loss": 6.7381,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.5255102040816323e-05,
-      "loss": 6.9252,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.448979591836734e-05,
-      "loss": 6.8262,
-      "step": 865
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.3724489795918367e-05,
-      "loss": 6.657,
-      "step": 866
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.2959183673469387e-05,
-      "loss": 6.8904,
-      "step": 867
+      "learning_rate": 0.00025679999999999995,
+      "loss": 6.9015,
+      "step": 432
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.2193877551020406e-05,
-      "loss": 6.6568,
-      "step": 868
+      "learning_rate": 0.00025739999999999997,
+      "loss": 6.7733,
+      "step": 433
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.1428571428571425e-05,
-      "loss": 6.7322,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 2.0663265306122444e-05,
-      "loss": 6.7062,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.989795918367347e-05,
-      "loss": 6.6488,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.913265306122449e-05,
-      "loss": 6.7146,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.8367346938775508e-05,
-      "loss": 6.7157,
-      "step": 873
+      "learning_rate": 0.000258,
+      "loss": 6.7356,
+      "step": 434
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.760204081632653e-05,
-      "loss": 6.7805,
-      "step": 874
+      "learning_rate": 0.0002586,
+      "loss": 6.8718,
+      "step": 435
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.683673469387755e-05,
-      "loss": 6.8813,
-      "step": 875
+      "learning_rate": 0.00025919999999999996,
+      "loss": 6.758,
+      "step": 436
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.607142857142857e-05,
-      "loss": 6.6575,
-      "step": 876
+      "learning_rate": 0.00025979999999999997,
+      "loss": 6.7061,
+      "step": 437
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.530612244897959e-05,
-      "loss": 6.5684,
-      "step": 877
+      "learning_rate": 0.0002604,
+      "loss": 6.8063,
+      "step": 438
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.454081632653061e-05,
-      "loss": 6.6932,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.3775510204081633e-05,
-      "loss": 6.7408,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.3010204081632652e-05,
-      "loss": 6.8176,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.224489795918367e-05,
-      "loss": 6.5252,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.1479591836734693e-05,
-      "loss": 6.7663,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.0714285714285712e-05,
-      "loss": 6.6353,
-      "step": 883
+      "learning_rate": 0.000261,
+      "loss": 6.787,
+      "step": 439
     },
     {
       "epoch": 0.99,
-      "learning_rate": 9.948979591836735e-06,
-      "loss": 6.6783,
-      "step": 884
+      "learning_rate": 0.00026159999999999996,
+      "loss": 6.7439,
+      "step": 440
     },
     {
       "epoch": 0.99,
-      "learning_rate": 9.183673469387754e-06,
-      "loss": 6.6815,
-      "step": 885
+      "learning_rate": 0.0002622,
+      "loss": 6.5944,
+      "step": 441
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.418367346938775e-06,
-      "loss": 6.7086,
-      "step": 886
+      "learning_rate": 0.0002628,
+      "loss": 6.7605,
+      "step": 442
     },
     {
       "epoch": 0.99,
-      "learning_rate": 7.653061224489796e-06,
-      "loss": 6.4744,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 6.887755102040816e-06,
-      "loss": 6.642,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 6.122448979591835e-06,
-      "loss": 6.4769,
-      "step": 889
+      "learning_rate": 0.00026339999999999995,
+      "loss": 6.4567,
+      "step": 443
     },
     {
       "epoch": 1.0,
-      "learning_rate": 5.357142857142856e-06,
-      "loss": 6.4305,
-      "step": 890
+      "learning_rate": 0.00026399999999999997,
+      "loss": 6.6392,
+      "step": 444
     },
     {
       "epoch": 1.0,
-      "learning_rate": 4.591836734693877e-06,
-      "loss": 6.3352,
-      "step": 891
+      "learning_rate": 0.0002646,
+      "loss": 6.4273,
+      "step": 445
     },
     {
       "epoch": 1.0,
-      "learning_rate": 3.826530612244898e-06,
-      "loss": 6.0518,
-      "step": 892
+      "learning_rate": 0.0002652,
+      "loss": 6.3367,
+      "step": 446
     },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 6.883699467363914,
-      "train_runtime": 5554.433,
-      "train_samples_per_second": 5.138,
-      "train_steps_per_second": 0.161
+      "train_loss": 6.953876193863394,
+      "train_runtime": 4987.98,
+      "train_samples_per_second": 5.721,
+      "train_steps_per_second": 0.089
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,