diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,5383 +2,2698 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 4.8157,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.747,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.0773,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.9006,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 4.9291,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.8744,
       "step": 3
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.9203,
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.8574,
       "step": 4
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.7409,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.7861,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.8712,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.8432,
       "step": 6
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.8337,
+      "epoch": 0.02,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.8179,
       "step": 7
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.7474,
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.8564,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.901,
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.8332,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.7999,
+      "epoch": 0.02,
+      "learning_rate": 1.8e-06,
+      "loss": 4.7164,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-06,
-      "loss": 4.8376,
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.8208,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.8031,
+      "epoch": 0.03,
+      "learning_rate": 2.2e-06,
+      "loss": 4.7747,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.2e-06,
-      "loss": 4.7704,
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.7053,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.7026,
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 4.7023,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.6e-06,
-      "loss": 4.6396,
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.5744,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.7621,
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 4.7191,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3e-06,
-      "loss": 4.7556,
+      "epoch": 0.04,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.6175,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.8365,
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.6957,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.5531,
+      "epoch": 0.04,
+      "learning_rate": 3.6e-06,
+      "loss": 4.4988,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-06,
-      "loss": 4.7096,
+      "epoch": 0.04,
+      "learning_rate": 3.8e-06,
+      "loss": 4.6422,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.8e-06,
-      "loss": 4.6264,
+      "epoch": 0.05,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.6299,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.5999,
+      "epoch": 0.05,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.5586,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.6059,
+      "epoch": 0.05,
+      "learning_rate": 4.4e-06,
+      "loss": 4.6489,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.4e-06,
-      "loss": 4.7259,
+      "epoch": 0.05,
+      "learning_rate": 4.6e-06,
+      "loss": 4.614,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.6e-06,
-      "loss": 4.5828,
+      "epoch": 0.06,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.4035,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.751,
+      "epoch": 0.06,
+      "learning_rate": 5e-06,
+      "loss": 4.6675,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5e-06,
-      "loss": 4.4654,
+      "epoch": 0.06,
+      "learning_rate": 5.2e-06,
+      "loss": 4.5006,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.2e-06,
-      "loss": 4.3872,
+      "epoch": 0.06,
+      "learning_rate": 5.4e-06,
+      "loss": 4.3643,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-06,
-      "loss": 4.3899,
+      "epoch": 0.07,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.3192,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.5933,
+      "epoch": 0.07,
+      "learning_rate": 5.8e-06,
+      "loss": 4.5311,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.8e-06,
-      "loss": 4.5498,
+      "epoch": 0.07,
+      "learning_rate": 6e-06,
+      "loss": 4.5042,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6e-06,
-      "loss": 4.3205,
+      "epoch": 0.07,
+      "learning_rate": 6.2e-06,
+      "loss": 4.3526,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.2e-06,
-      "loss": 4.4264,
+      "epoch": 0.07,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.4255,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.2957,
+      "epoch": 0.08,
+      "learning_rate": 6.6e-06,
+      "loss": 4.3723,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.6e-06,
-      "loss": 4.4812,
+      "epoch": 0.08,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.4424,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.4468,
+      "epoch": 0.08,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 4.3992,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 4.4754,
+      "epoch": 0.08,
+      "learning_rate": 7.2e-06,
+      "loss": 4.4579,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.2e-06,
-      "loss": 4.4744,
+      "epoch": 0.09,
+      "learning_rate": 7.4e-06,
+      "loss": 4.385,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.4e-06,
-      "loss": 4.6401,
+      "epoch": 0.09,
+      "learning_rate": 7.6e-06,
+      "loss": 4.5033,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.6e-06,
-      "loss": 4.4406,
+      "epoch": 0.09,
+      "learning_rate": 7.8e-06,
+      "loss": 4.3482,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.8e-06,
-      "loss": 4.3062,
+      "epoch": 0.09,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.3762,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.461,
+      "epoch": 0.09,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 4.4784,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 4.2756,
+      "epoch": 0.1,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.2417,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.4212,
+      "epoch": 0.1,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 4.4203,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 4.4871,
+      "epoch": 0.1,
+      "learning_rate": 8.8e-06,
+      "loss": 4.5079,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.8e-06,
-      "loss": 4.6143,
+      "epoch": 0.1,
+      "learning_rate": 9e-06,
+      "loss": 4.5547,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9e-06,
-      "loss": 4.4742,
+      "epoch": 0.11,
+      "learning_rate": 9.2e-06,
+      "loss": 4.4638,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.2e-06,
-      "loss": 4.8497,
+      "epoch": 0.11,
+      "learning_rate": 9.4e-06,
+      "loss": 4.699,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.2e-06,
-      "loss": 5.0414,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.8242,
       "step": 49
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.4e-06,
-      "loss": 4.6475,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.4994,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2063,
+      "epoch": 0.11,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.3463,
       "step": 51
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.3556,
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 4.2201,
       "step": 52
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1e-05,
-      "loss": 4.2277,
+      "epoch": 0.12,
+      "learning_rate": 1.02e-05,
+      "loss": 4.3763,
       "step": 53
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.02e-05,
-      "loss": 4.2997,
+      "epoch": 0.12,
+      "learning_rate": 1.04e-05,
+      "loss": 4.2946,
       "step": 54
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.04e-05,
-      "loss": 4.2926,
+      "epoch": 0.12,
+      "learning_rate": 1.06e-05,
+      "loss": 4.1038,
       "step": 55
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.06e-05,
-      "loss": 4.3759,
+      "epoch": 0.13,
+      "learning_rate": 1.08e-05,
+      "loss": 4.3279,
       "step": 56
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.08e-05,
-      "loss": 4.3415,
+      "epoch": 0.13,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 4.3403,
       "step": 57
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 4.5762,
+      "epoch": 0.13,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 4.4244,
       "step": 58
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 4.3118,
+      "epoch": 0.13,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.351,
       "step": 59
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.1467,
+      "epoch": 0.13,
+      "learning_rate": 1.16e-05,
+      "loss": 4.2072,
       "step": 60
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.16e-05,
-      "loss": 4.4006,
+      "epoch": 0.14,
+      "learning_rate": 1.18e-05,
+      "loss": 4.2895,
       "step": 61
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.18e-05,
-      "loss": 4.3534,
+      "epoch": 0.14,
+      "learning_rate": 1.2e-05,
+      "loss": 4.2445,
       "step": 62
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2e-05,
-      "loss": 4.1833,
+      "epoch": 0.14,
+      "learning_rate": 1.22e-05,
+      "loss": 4.2112,
       "step": 63
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.22e-05,
-      "loss": 4.3445,
+      "epoch": 0.14,
+      "learning_rate": 1.24e-05,
+      "loss": 4.3095,
       "step": 64
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.24e-05,
-      "loss": 4.138,
+      "epoch": 0.15,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 4.2884,
       "step": 65
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 4.4047,
+      "epoch": 0.15,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 4.154,
       "step": 66
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 4.1752,
+      "epoch": 0.15,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 4.2997,
       "step": 67
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 4.2174,
+      "epoch": 0.15,
+      "learning_rate": 1.32e-05,
+      "loss": 4.184,
       "step": 68
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.32e-05,
-      "loss": 4.1656,
+      "epoch": 0.15,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 4.2295,
       "step": 69
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 4.2814,
+      "epoch": 0.16,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 4.1964,
       "step": 70
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 4.4579,
+      "epoch": 0.16,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.3184,
       "step": 71
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.3513,
+      "epoch": 0.16,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 4.3462,
       "step": 72
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 4.2773,
+      "epoch": 0.16,
+      "learning_rate": 1.42e-05,
+      "loss": 4.1737,
       "step": 73
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.42e-05,
-      "loss": 4.3272,
+      "epoch": 0.17,
+      "learning_rate": 1.44e-05,
+      "loss": 4.2753,
       "step": 74
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.44e-05,
-      "loss": 4.129,
+      "epoch": 0.17,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 4.2565,
       "step": 75
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 4.4107,
+      "epoch": 0.17,
+      "learning_rate": 1.48e-05,
+      "loss": 4.2009,
       "step": 76
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.48e-05,
-      "loss": 4.3035,
+      "epoch": 0.17,
+      "learning_rate": 1.5e-05,
+      "loss": 4.2576,
       "step": 77
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5e-05,
-      "loss": 4.2116,
+      "epoch": 0.17,
+      "learning_rate": 1.52e-05,
+      "loss": 4.2042,
       "step": 78
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.52e-05,
-      "loss": 4.0825,
+      "epoch": 0.18,
+      "learning_rate": 1.54e-05,
+      "loss": 4.1836,
       "step": 79
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.54e-05,
-      "loss": 4.3311,
+      "epoch": 0.18,
+      "learning_rate": 1.56e-05,
+      "loss": 4.3454,
       "step": 80
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.56e-05,
-      "loss": 4.3758,
+      "epoch": 0.18,
+      "learning_rate": 1.58e-05,
+      "loss": 4.2717,
       "step": 81
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.58e-05,
-      "loss": 4.1791,
+      "epoch": 0.18,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 4.2216,
       "step": 82
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 4.4066,
+      "epoch": 0.19,
+      "learning_rate": 1.62e-05,
+      "loss": 4.3399,
       "step": 83
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.62e-05,
-      "loss": 4.3315,
+      "epoch": 0.19,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 4.2078,
       "step": 84
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 4.2736,
+      "epoch": 0.19,
+      "learning_rate": 1.66e-05,
+      "loss": 4.2838,
       "step": 85
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.66e-05,
-      "loss": 4.269,
+      "epoch": 0.19,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.2483,
       "step": 86
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.2719,
+      "epoch": 0.2,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 4.328,
       "step": 87
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 4.2614,
+      "epoch": 0.2,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 4.2141,
       "step": 88
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 4.3118,
+      "epoch": 0.2,
+      "learning_rate": 1.74e-05,
+      "loss": 4.3494,
       "step": 89
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.74e-05,
-      "loss": 4.2336,
+      "epoch": 0.2,
+      "learning_rate": 1.76e-05,
+      "loss": 4.335,
       "step": 90
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.76e-05,
-      "loss": 4.468,
+      "epoch": 0.2,
+      "learning_rate": 1.78e-05,
+      "loss": 4.2173,
       "step": 91
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.78e-05,
-      "loss": 4.3776,
+      "epoch": 0.21,
+      "learning_rate": 1.8e-05,
+      "loss": 4.4714,
       "step": 92
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.8e-05,
-      "loss": 4.099,
+      "epoch": 0.21,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 4.3728,
       "step": 93
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 4.3081,
+      "epoch": 0.21,
+      "learning_rate": 1.84e-05,
+      "loss": 4.3668,
       "step": 94
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.84e-05,
-      "loss": 4.5957,
+      "epoch": 0.21,
+      "learning_rate": 1.86e-05,
+      "loss": 4.2791,
       "step": 95
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.86e-05,
-      "loss": 4.5014,
+      "epoch": 0.22,
+      "learning_rate": 1.88e-05,
+      "loss": 4.5144,
       "step": 96
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.88e-05,
-      "loss": 4.2892,
+      "epoch": 0.22,
+      "learning_rate": 1.9e-05,
+      "loss": 4.4262,
       "step": 97
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9e-05,
-      "loss": 4.5866,
+      "epoch": 0.22,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.2953,
       "step": 98
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.3744,
+      "epoch": 0.22,
+      "learning_rate": 1.94e-05,
+      "loss": 4.6353,
       "step": 99
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.94e-05,
-      "loss": 4.3226,
+      "epoch": 0.22,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 4.4004,
       "step": 100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 4.1744,
+      "epoch": 0.23,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 4.2493,
       "step": 101
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 4.2069,
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 4.1879,
       "step": 102
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2e-05,
-      "loss": 4.3266,
+      "epoch": 0.23,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 4.0631,
       "step": 103
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 4.2493,
+      "epoch": 0.23,
+      "learning_rate": 2.04e-05,
+      "loss": 4.1411,
       "step": 104
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.04e-05,
-      "loss": 4.0665,
+      "epoch": 0.24,
+      "learning_rate": 2.06e-05,
+      "loss": 4.0814,
       "step": 105
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.06e-05,
-      "loss": 4.2994,
+      "epoch": 0.24,
+      "learning_rate": 2.08e-05,
+      "loss": 4.1362,
       "step": 106
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.08e-05,
-      "loss": 4.2977,
+      "epoch": 0.24,
+      "learning_rate": 2.1e-05,
+      "loss": 4.2139,
       "step": 107
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1e-05,
-      "loss": 4.4379,
+      "epoch": 0.24,
+      "learning_rate": 2.12e-05,
+      "loss": 4.1156,
       "step": 108
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.12e-05,
-      "loss": 4.0162,
+      "epoch": 0.24,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 4.1333,
       "step": 109
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 4.1472,
+      "epoch": 0.25,
+      "learning_rate": 2.16e-05,
+      "loss": 4.1727,
       "step": 110
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.16e-05,
-      "loss": 4.331,
+      "epoch": 0.25,
+      "learning_rate": 2.18e-05,
+      "loss": 4.2007,
       "step": 111
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.18e-05,
-      "loss": 4.193,
+      "epoch": 0.25,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.1927,
       "step": 112
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 4.186,
+      "epoch": 0.25,
+      "learning_rate": 2.22e-05,
+      "loss": 4.1808,
       "step": 113
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.22e-05,
-      "loss": 4.1242,
+      "epoch": 0.26,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 4.1332,
       "step": 114
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 4.1849,
+      "epoch": 0.26,
+      "learning_rate": 2.26e-05,
+      "loss": 4.1873,
       "step": 115
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.26e-05,
-      "loss": 3.9608,
+      "epoch": 0.26,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.2898,
       "step": 116
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.2707,
+      "epoch": 0.26,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.1585,
       "step": 117
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 4.1471,
+      "epoch": 0.26,
+      "learning_rate": 2.32e-05,
+      "loss": 4.3155,
       "step": 118
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.32e-05,
-      "loss": 4.2742,
+      "epoch": 0.27,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.216,
       "step": 119
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.3286,
+      "epoch": 0.27,
+      "learning_rate": 2.36e-05,
+      "loss": 4.239,
       "step": 120
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.36e-05,
-      "loss": 4.2384,
+      "epoch": 0.27,
+      "learning_rate": 2.38e-05,
+      "loss": 4.0644,
       "step": 121
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.38e-05,
-      "loss": 4.2224,
+      "epoch": 0.27,
+      "learning_rate": 2.4e-05,
+      "loss": 4.1797,
       "step": 122
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4e-05,
-      "loss": 4.0854,
+      "epoch": 0.28,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 4.1803,
       "step": 123
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 4.3342,
+      "epoch": 0.28,
+      "learning_rate": 2.44e-05,
+      "loss": 4.1742,
       "step": 124
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.44e-05,
-      "loss": 4.252,
+      "epoch": 0.28,
+      "learning_rate": 2.46e-05,
+      "loss": 4.2787,
       "step": 125
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.46e-05,
-      "loss": 4.298,
-      "step": 126
-    },
-    {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 2.48e-05,
-      "loss": 4.3579,
-      "step": 127
+      "loss": 4.2894,
+      "step": 126
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 2.5e-05,
-      "loss": 4.1777,
-      "step": 128
+      "loss": 4.2008,
+      "step": 127
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.29,
       "learning_rate": 2.5200000000000003e-05,
-      "loss": 4.1587,
-      "step": 129
+      "loss": 4.1734,
+      "step": 128
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 2.54e-05,
-      "loss": 4.3313,
-      "step": 130
+      "loss": 4.0823,
+      "step": 129
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 2.5600000000000002e-05,
-      "loss": 4.2219,
-      "step": 131
+      "loss": 4.2369,
+      "step": 130
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.29,
       "learning_rate": 2.58e-05,
-      "loss": 4.3092,
-      "step": 132
+      "loss": 4.1488,
+      "step": 131
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 4.2794,
-      "step": 133
+      "loss": 4.2138,
+      "step": 132
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.6200000000000003e-05,
-      "loss": 4.281,
-      "step": 134
+      "loss": 4.1217,
+      "step": 133
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.64e-05,
-      "loss": 4.148,
-      "step": 135
+      "loss": 4.1846,
+      "step": 134
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.6600000000000003e-05,
-      "loss": 4.2269,
-      "step": 136
+      "loss": 4.3267,
+      "step": 135
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 2.6800000000000004e-05,
-      "loss": 4.5121,
-      "step": 137
+      "loss": 4.184,
+      "step": 136
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 4.2656,
-      "step": 138
+      "loss": 4.263,
+      "step": 137
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 2.7200000000000004e-05,
-      "loss": 4.321,
-      "step": 139
+      "loss": 4.3025,
+      "step": 138
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 2.7400000000000002e-05,
-      "loss": 4.105,
-      "step": 140
+      "loss": 4.307,
+      "step": 139
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 2.7600000000000003e-05,
-      "loss": 4.4071,
-      "step": 141
+      "loss": 4.2831,
+      "step": 140
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.7800000000000005e-05,
-      "loss": 4.4735,
-      "step": 142
+      "loss": 4.3902,
+      "step": 141
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 4.3833,
-      "step": 143
+      "loss": 4.3271,
+      "step": 142
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.2836,
-      "step": 144
+      "loss": 4.2895,
+      "step": 143
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 2.84e-05,
-      "loss": 4.4196,
-      "step": 145
+      "loss": 4.3228,
+      "step": 144
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 2.86e-05,
-      "loss": 4.3708,
-      "step": 146
+      "loss": 4.2486,
+      "step": 145
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 2.88e-05,
-      "loss": 4.464,
-      "step": 147
+      "loss": 4.3242,
+      "step": 146
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 2.9e-05,
-      "loss": 4.2478,
-      "step": 148
+      "loss": 4.595,
+      "step": 147
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 2.9199999999999998e-05,
-      "loss": 4.5954,
-      "step": 149
+      "loss": 4.6988,
+      "step": 148
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 2.94e-05,
-      "loss": 4.2222,
-      "step": 150
+      "loss": 4.4544,
+      "step": 149
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 2.96e-05,
-      "loss": 4.3196,
-      "step": 151
+      "loss": 4.5323,
+      "step": 150
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 2.98e-05,
-      "loss": 4.2153,
-      "step": 152
+      "loss": 4.1013,
+      "step": 151
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 3e-05,
-      "loss": 4.2844,
-      "step": 153
+      "loss": 4.1979,
+      "step": 152
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 3.02e-05,
-      "loss": 4.1266,
-      "step": 154
+      "loss": 4.2243,
+      "step": 153
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 3.04e-05,
-      "loss": 4.0775,
-      "step": 155
+      "loss": 4.0781,
+      "step": 154
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 3.06e-05,
-      "loss": 4.1736,
-      "step": 156
+      "loss": 4.213,
+      "step": 155
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 3.08e-05,
-      "loss": 4.3044,
-      "step": 157
+      "loss": 4.0605,
+      "step": 156
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 3.1e-05,
-      "loss": 4.4134,
-      "step": 158
+      "loss": 4.0824,
+      "step": 157
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 3.12e-05,
-      "loss": 4.4487,
-      "step": 159
+      "loss": 4.1807,
+      "step": 158
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.1400000000000004e-05,
-      "loss": 4.177,
-      "step": 160
+      "loss": 4.1912,
+      "step": 159
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.16e-05,
-      "loss": 4.1921,
-      "step": 161
+      "loss": 4.2013,
+      "step": 160
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.18e-05,
-      "loss": 4.2225,
-      "step": 162
+      "loss": 4.2653,
+      "step": 161
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 4.2383,
-      "step": 163
+      "loss": 4.2088,
+      "step": 162
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 3.2200000000000003e-05,
-      "loss": 4.3612,
-      "step": 164
+      "loss": 4.1604,
+      "step": 163
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 3.24e-05,
-      "loss": 4.2632,
-      "step": 165
+      "loss": 4.218,
+      "step": 164
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 3.26e-05,
-      "loss": 4.2825,
-      "step": 166
+      "loss": 4.0929,
+      "step": 165
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 3.2800000000000004e-05,
-      "loss": 4.2461,
-      "step": 167
+      "loss": 4.1595,
+      "step": 166
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 3.3e-05,
-      "loss": 4.035,
-      "step": 168
+      "loss": 4.0662,
+      "step": 167
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.32e-05,
-      "loss": 4.0397,
-      "step": 169
+      "loss": 4.1203,
+      "step": 168
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.3400000000000005e-05,
-      "loss": 4.1816,
-      "step": 170
+      "loss": 4.2015,
+      "step": 169
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.3600000000000004e-05,
-      "loss": 4.4042,
-      "step": 171
+      "loss": 4.2155,
+      "step": 170
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 3.38e-05,
-      "loss": 4.3567,
-      "step": 172
+      "loss": 4.2056,
+      "step": 171
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.2443,
-      "step": 173
+      "loss": 4.2715,
+      "step": 172
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 3.4200000000000005e-05,
-      "loss": 4.0836,
-      "step": 174
+      "loss": 4.0736,
+      "step": 173
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 3.4399999999999996e-05,
-      "loss": 4.2463,
-      "step": 175
+      "loss": 4.2239,
+      "step": 174
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 3.46e-05,
-      "loss": 4.0456,
-      "step": 176
+      "loss": 4.2676,
+      "step": 175
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 3.48e-05,
-      "loss": 4.1016,
-      "step": 177
+      "loss": 4.1391,
+      "step": 176
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.5e-05,
-      "loss": 4.174,
-      "step": 178
+      "loss": 4.0946,
+      "step": 177
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.52e-05,
-      "loss": 4.2908,
-      "step": 179
+      "loss": 4.2066,
+      "step": 178
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.54e-05,
-      "loss": 4.2543,
-      "step": 180
+      "loss": 4.2802,
+      "step": 179
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 3.56e-05,
-      "loss": 4.2862,
-      "step": 181
+      "loss": 4.2446,
+      "step": 180
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 3.58e-05,
-      "loss": 4.0973,
-      "step": 182
+      "loss": 4.1873,
+      "step": 181
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 3.6e-05,
-      "loss": 4.3109,
-      "step": 183
+      "loss": 4.2523,
+      "step": 182
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 3.62e-05,
-      "loss": 4.2882,
-      "step": 184
+      "loss": 4.2105,
+      "step": 183
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 3.6400000000000004e-05,
-      "loss": 4.29,
-      "step": 185
+      "loss": 4.2264,
+      "step": 184
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 3.66e-05,
-      "loss": 4.1491,
-      "step": 186
+      "loss": 4.208,
+      "step": 185
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.68e-05,
-      "loss": 4.3615,
-      "step": 187
+      "loss": 4.0727,
+      "step": 186
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.7e-05,
-      "loss": 4.0704,
-      "step": 188
+      "loss": 4.0995,
+      "step": 187
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.72e-05,
-      "loss": 4.2079,
-      "step": 189
+      "loss": 4.0693,
+      "step": 188
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 3.74e-05,
-      "loss": 4.3052,
-      "step": 190
+      "loss": 4.1961,
+      "step": 189
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 3.76e-05,
-      "loss": 4.325,
-      "step": 191
+      "loss": 4.3002,
+      "step": 190
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 3.7800000000000004e-05,
-      "loss": 4.2566,
-      "step": 192
+      "loss": 4.2077,
+      "step": 191
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 3.8e-05,
-      "loss": 4.4982,
-      "step": 193
+      "loss": 4.1656,
+      "step": 192
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 3.82e-05,
-      "loss": 4.1911,
-      "step": 194
+      "loss": 4.1497,
+      "step": 193
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 3.8400000000000005e-05,
-      "loss": 4.2439,
-      "step": 195
+      "loss": 4.28,
+      "step": 194
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.86e-05,
-      "loss": 4.4769,
-      "step": 196
+      "loss": 4.4085,
+      "step": 195
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.88e-05,
-      "loss": 4.4989,
-      "step": 197
+      "loss": 4.3136,
+      "step": 196
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 4.384,
-      "step": 198
+      "loss": 4.4431,
+      "step": 197
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 3.9200000000000004e-05,
-      "loss": 4.6191,
-      "step": 199
+      "loss": 4.5063,
+      "step": 198
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 3.94e-05,
-      "loss": 4.658,
-      "step": 200
+      "loss": 4.4454,
+      "step": 199
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 3.960000000000001e-05,
-      "loss": 4.0765,
-      "step": 201
+      "loss": 4.436,
+      "step": 200
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 3.9800000000000005e-05,
-      "loss": 4.2211,
-      "step": 202
+      "loss": 4.2278,
+      "step": 201
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 4e-05,
-      "loss": 4.1234,
-      "step": 203
+      "loss": 4.1701,
+      "step": 202
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 4.02e-05,
-      "loss": 4.1443,
-      "step": 204
+      "loss": 4.1987,
+      "step": 203
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 4.0400000000000006e-05,
-      "loss": 4.0152,
-      "step": 205
+      "loss": 4.1717,
+      "step": 204
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 4.0600000000000004e-05,
-      "loss": 4.0674,
-      "step": 206
+      "loss": 4.2467,
+      "step": 205
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 4.08e-05,
-      "loss": 4.1268,
-      "step": 207
+      "loss": 4.2045,
+      "step": 206
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 4.1e-05,
-      "loss": 4.0033,
-      "step": 208
+      "loss": 4.0783,
+      "step": 207
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 4.12e-05,
-      "loss": 4.2234,
-      "step": 209
+      "loss": 4.1963,
+      "step": 208
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 4.14e-05,
-      "loss": 4.164,
-      "step": 210
+      "loss": 4.2476,
+      "step": 209
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 4.16e-05,
-      "loss": 4.1,
-      "step": 211
+      "loss": 4.197,
+      "step": 210
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 4.18e-05,
-      "loss": 4.2328,
-      "step": 212
+      "loss": 4.0622,
+      "step": 211
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.2e-05,
-      "loss": 4.1819,
-      "step": 213
+      "loss": 4.1235,
+      "step": 212
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.22e-05,
-      "loss": 4.2359,
-      "step": 214
+      "loss": 4.0617,
+      "step": 213
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.24e-05,
-      "loss": 4.0974,
-      "step": 215
+      "loss": 4.119,
+      "step": 214
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.26e-05,
-      "loss": 4.2581,
-      "step": 216
+      "loss": 4.1679,
+      "step": 215
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 4.2800000000000004e-05,
-      "loss": 4.1981,
-      "step": 217
+      "loss": 4.1364,
+      "step": 216
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 4.3e-05,
-      "loss": 4.1429,
-      "step": 218
+      "loss": 4.1803,
+      "step": 217
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 4.32e-05,
-      "loss": 4.241,
-      "step": 219
+      "loss": 4.132,
+      "step": 218
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 4.3400000000000005e-05,
-      "loss": 4.1933,
-      "step": 220
+      "loss": 4.0901,
+      "step": 219
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 4.36e-05,
-      "loss": 4.1245,
-      "step": 221
+      "loss": 4.1789,
+      "step": 220
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.38e-05,
-      "loss": 4.2515,
-      "step": 222
+      "loss": 4.1393,
+      "step": 221
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 4.2634,
-      "step": 223
+      "loss": 4.1346,
+      "step": 222
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.4200000000000004e-05,
-      "loss": 4.2013,
-      "step": 224
+      "loss": 4.1677,
+      "step": 223
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.44e-05,
-      "loss": 4.3057,
-      "step": 225
+      "loss": 4.1156,
+      "step": 224
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 4.46e-05,
-      "loss": 4.4526,
-      "step": 226
+      "loss": 4.27,
+      "step": 225
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 4.4800000000000005e-05,
-      "loss": 4.2218,
-      "step": 227
+      "loss": 4.159,
+      "step": 226
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 4.5e-05,
-      "loss": 4.1629,
-      "step": 228
+      "loss": 4.2819,
+      "step": 227
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 4.52e-05,
-      "loss": 4.1873,
-      "step": 229
+      "loss": 4.2656,
+      "step": 228
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 4.5400000000000006e-05,
-      "loss": 4.2066,
-      "step": 230
+      "loss": 4.1013,
+      "step": 229
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.5600000000000004e-05,
-      "loss": 4.1974,
-      "step": 231
+      "loss": 4.2321,
+      "step": 230
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.58e-05,
-      "loss": 4.3257,
-      "step": 232
+      "loss": 4.2513,
+      "step": 231
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 4.1752,
-      "step": 233
+      "loss": 4.0487,
+      "step": 232
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.6200000000000005e-05,
-      "loss": 4.3201,
-      "step": 234
+      "loss": 4.1196,
+      "step": 233
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 4.64e-05,
-      "loss": 4.3536,
-      "step": 235
+      "loss": 4.1884,
+      "step": 234
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 4.660000000000001e-05,
-      "loss": 4.197,
-      "step": 236
+      "loss": 4.2405,
+      "step": 235
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 4.6800000000000006e-05,
-      "loss": 4.1763,
-      "step": 237
+      "loss": 4.1805,
+      "step": 236
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 4.7e-05,
-      "loss": 4.4393,
-      "step": 238
+      "loss": 4.1694,
+      "step": 237
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 4.72e-05,
-      "loss": 4.2264,
-      "step": 239
+      "loss": 4.1361,
+      "step": 238
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.74e-05,
-      "loss": 4.3602,
-      "step": 240
+      "loss": 4.0898,
+      "step": 239
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.76e-05,
-      "loss": 4.316,
-      "step": 241
+      "loss": 4.1756,
+      "step": 240
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.78e-05,
-      "loss": 4.4743,
-      "step": 242
+      "loss": 4.1759,
+      "step": 241
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.8e-05,
-      "loss": 4.3131,
-      "step": 243
+      "loss": 4.2335,
+      "step": 242
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 4.82e-05,
-      "loss": 4.2784,
-      "step": 244
+      "loss": 4.4165,
+      "step": 243
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 4.8400000000000004e-05,
-      "loss": 4.4655,
-      "step": 245
+      "loss": 4.2451,
+      "step": 244
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 4.86e-05,
-      "loss": 4.3575,
-      "step": 246
+      "loss": 4.2268,
+      "step": 245
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 4.88e-05,
-      "loss": 4.907,
-      "step": 247
+      "loss": 4.0973,
+      "step": 246
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 4.9e-05,
-      "loss": 4.7885,
-      "step": 248
+      "loss": 4.0274,
+      "step": 247
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.92e-05,
-      "loss": 4.7936,
-      "step": 249
+      "loss": 4.5398,
+      "step": 248
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.94e-05,
-      "loss": 4.1855,
-      "step": 250
+      "loss": 4.4893,
+      "step": 249
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.96e-05,
-      "loss": 4.499,
-      "step": 251
+      "loss": 4.4634,
+      "step": 250
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 4.9800000000000004e-05,
-      "loss": 4.151,
-      "step": 252
+      "loss": 4.2059,
+      "step": 251
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 5e-05,
-      "loss": 4.1646,
-      "step": 253
+      "loss": 4.2231,
+      "step": 252
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 5.02e-05,
-      "loss": 4.2497,
-      "step": 254
+      "loss": 4.2464,
+      "step": 253
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 5.0400000000000005e-05,
-      "loss": 4.1886,
-      "step": 255
+      "loss": 4.0991,
+      "step": 254
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 5.0600000000000003e-05,
-      "loss": 4.1245,
-      "step": 256
+      "loss": 4.1931,
+      "step": 255
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.57,
       "learning_rate": 5.08e-05,
-      "loss": 4.2806,
-      "step": 257
+      "loss": 4.1384,
+      "step": 256
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.1000000000000006e-05,
-      "loss": 4.1536,
-      "step": 258
+      "loss": 4.1315,
+      "step": 257
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.1200000000000004e-05,
-      "loss": 4.1426,
-      "step": 259
+      "loss": 4.1089,
+      "step": 258
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.14e-05,
-      "loss": 4.2571,
-      "step": 260
+      "loss": 4.1466,
+      "step": 259
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 5.16e-05,
-      "loss": 4.2818,
-      "step": 261
+      "loss": 4.1282,
+      "step": 260
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 5.1800000000000005e-05,
-      "loss": 4.1184,
-      "step": 262
+      "loss": 4.213,
+      "step": 261
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 5.2000000000000004e-05,
-      "loss": 4.1301,
-      "step": 263
+      "loss": 4.1021,
+      "step": 262
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 5.22e-05,
-      "loss": 4.1637,
-      "step": 264
+      "loss": 4.2691,
+      "step": 263
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 5.2400000000000007e-05,
-      "loss": 4.2182,
-      "step": 265
+      "loss": 4.1385,
+      "step": 264
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 5.2600000000000005e-05,
-      "loss": 4.3134,
-      "step": 266
+      "loss": 4.1956,
+      "step": 265
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.28e-05,
-      "loss": 4.1519,
-      "step": 267
+      "loss": 4.1191,
+      "step": 266
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.300000000000001e-05,
-      "loss": 4.4625,
-      "step": 268
+      "loss": 4.1614,
+      "step": 267
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.3200000000000006e-05,
-      "loss": 4.2054,
-      "step": 269
+      "loss": 4.0642,
+      "step": 268
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 5.3400000000000004e-05,
-      "loss": 4.3867,
-      "step": 270
+      "loss": 4.2205,
+      "step": 269
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 5.360000000000001e-05,
-      "loss": 3.988,
-      "step": 271
+      "loss": 4.0858,
+      "step": 270
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.61,
       "learning_rate": 5.380000000000001e-05,
-      "loss": 4.1873,
-      "step": 272
+      "loss": 4.0458,
+      "step": 271
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 5.4000000000000005e-05,
-      "loss": 4.1982,
-      "step": 273
+      "loss": 4.2021,
+      "step": 272
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 5.420000000000001e-05,
-      "loss": 4.068,
-      "step": 274
+      "loss": 4.0632,
+      "step": 273
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 5.440000000000001e-05,
-      "loss": 4.3203,
-      "step": 275
+      "loss": 4.1333,
+      "step": 274
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.4600000000000006e-05,
-      "loss": 4.2613,
-      "step": 276
+      "loss": 4.2086,
+      "step": 275
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.4800000000000004e-05,
-      "loss": 4.1814,
-      "step": 277
+      "loss": 4.1195,
+      "step": 276
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 4.1196,
-      "step": 278
+      "loss": 4.08,
+      "step": 277
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 5.520000000000001e-05,
-      "loss": 4.139,
-      "step": 279
+      "loss": 4.1521,
+      "step": 278
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 5.5400000000000005e-05,
-      "loss": 4.3619,
-      "step": 280
+      "loss": 4.1239,
+      "step": 279
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 5.560000000000001e-05,
-      "loss": 4.0855,
-      "step": 281
+      "loss": 4.2814,
+      "step": 280
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 5.580000000000001e-05,
-      "loss": 4.2313,
-      "step": 282
+      "loss": 4.1766,
+      "step": 281
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 5.6000000000000006e-05,
-      "loss": 3.9799,
-      "step": 283
+      "loss": 4.0358,
+      "step": 282
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 5.620000000000001e-05,
-      "loss": 4.2634,
-      "step": 284
+      "loss": 4.0526,
+      "step": 283
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.2745,
-      "step": 285
+      "loss": 4.1761,
+      "step": 284
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.66e-05,
-      "loss": 4.2498,
-      "step": 286
+      "loss": 4.2562,
+      "step": 285
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.68e-05,
-      "loss": 4.3986,
-      "step": 287
+      "loss": 4.1342,
+      "step": 286
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.264,
-      "step": 288
+      "loss": 4.2481,
+      "step": 287
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 5.72e-05,
-      "loss": 4.3247,
-      "step": 289
+      "loss": 4.1786,
+      "step": 288
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 5.74e-05,
-      "loss": 4.3363,
-      "step": 290
+      "loss": 4.0954,
+      "step": 289
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 5.76e-05,
-      "loss": 4.3671,
-      "step": 291
+      "loss": 4.1567,
+      "step": 290
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 5.7799999999999995e-05,
-      "loss": 4.2068,
-      "step": 292
+      "loss": 4.44,
+      "step": 291
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 5.8e-05,
-      "loss": 4.5264,
-      "step": 293
+      "loss": 4.2732,
+      "step": 292
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.82e-05,
-      "loss": 4.3081,
-      "step": 294
+      "loss": 4.3449,
+      "step": 293
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.8399999999999997e-05,
-      "loss": 4.2125,
-      "step": 295
+      "loss": 4.3422,
+      "step": 294
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.86e-05,
-      "loss": 4.4924,
-      "step": 296
+      "loss": 4.2886,
+      "step": 295
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 5.88e-05,
-      "loss": 4.6536,
-      "step": 297
+      "loss": 4.3468,
+      "step": 296
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 5.9e-05,
-      "loss": 4.3406,
-      "step": 298
+      "loss": 4.1907,
+      "step": 297
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 5.92e-05,
-      "loss": 4.8184,
-      "step": 299
+      "loss": 4.5633,
+      "step": 298
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 5.94e-05,
-      "loss": 4.7172,
-      "step": 300
+      "loss": 4.2836,
+      "step": 299
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 5.96e-05,
-      "loss": 4.1678,
-      "step": 301
+      "loss": 4.6411,
+      "step": 300
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 5.9800000000000003e-05,
-      "loss": 4.2504,
-      "step": 302
+      "loss": 4.2357,
+      "step": 301
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 6e-05,
-      "loss": 4.1695,
-      "step": 303
+      "loss": 4.2123,
+      "step": 302
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 6.02e-05,
-      "loss": 4.2651,
-      "step": 304
+      "loss": 4.1736,
+      "step": 303
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 6.04e-05,
-      "loss": 4.1434,
-      "step": 305
+      "loss": 4.199,
+      "step": 304
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 6.06e-05,
-      "loss": 4.1557,
-      "step": 306
+      "loss": 4.1657,
+      "step": 305
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 6.08e-05,
-      "loss": 4.2189,
-      "step": 307
+      "loss": 4.1201,
+      "step": 306
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 6.1e-05,
-      "loss": 4.3707,
-      "step": 308
+      "loss": 4.1695,
+      "step": 307
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 6.12e-05,
-      "loss": 4.2428,
-      "step": 309
+      "loss": 4.1435,
+      "step": 308
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 6.14e-05,
-      "loss": 4.3603,
-      "step": 310
+      "loss": 4.1498,
+      "step": 309
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.16e-05,
-      "loss": 4.4532,
-      "step": 311
+      "loss": 4.2333,
+      "step": 310
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.18e-05,
-      "loss": 4.2964,
-      "step": 312
+      "loss": 4.0113,
+      "step": 311
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.2e-05,
-      "loss": 4.2266,
-      "step": 313
+      "loss": 4.2104,
+      "step": 312
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.220000000000001e-05,
-      "loss": 4.3037,
-      "step": 314
+      "loss": 4.3056,
+      "step": 313
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.24e-05,
-      "loss": 4.1271,
-      "step": 315
+      "loss": 4.1282,
+      "step": 314
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 6.26e-05,
-      "loss": 4.3572,
-      "step": 316
+      "loss": 4.1415,
+      "step": 315
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 6.280000000000001e-05,
-      "loss": 4.1578,
-      "step": 317
+      "loss": 4.1358,
+      "step": 316
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 6.3e-05,
-      "loss": 4.1251,
-      "step": 318
+      "loss": 4.1366,
+      "step": 317
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 6.32e-05,
-      "loss": 4.2534,
-      "step": 319
+      "loss": 4.1256,
+      "step": 318
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.340000000000001e-05,
-      "loss": 4.1787,
-      "step": 320
+      "loss": 4.1378,
+      "step": 319
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.36e-05,
-      "loss": 4.3733,
-      "step": 321
+      "loss": 4.2278,
+      "step": 320
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.38e-05,
-      "loss": 4.1954,
-      "step": 322
+      "loss": 4.1062,
+      "step": 321
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.400000000000001e-05,
-      "loss": 4.2272,
-      "step": 323
+      "loss": 3.9692,
+      "step": 322
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.42e-05,
-      "loss": 4.3435,
-      "step": 324
+      "loss": 4.1761,
+      "step": 323
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 6.440000000000001e-05,
-      "loss": 4.3948,
-      "step": 325
+      "loss": 4.0297,
+      "step": 324
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 6.460000000000001e-05,
-      "loss": 4.3071,
-      "step": 326
+      "loss": 4.196,
+      "step": 325
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 6.48e-05,
-      "loss": 4.1361,
-      "step": 327
+      "loss": 4.1788,
+      "step": 326
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 4.1913,
-      "step": 328
+      "loss": 4.0838,
+      "step": 327
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.52e-05,
-      "loss": 4.518,
-      "step": 329
+      "loss": 4.0498,
+      "step": 328
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.54e-05,
-      "loss": 4.5838,
-      "step": 330
+      "loss": 4.0674,
+      "step": 329
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.560000000000001e-05,
-      "loss": 4.1964,
-      "step": 331
+      "loss": 4.2265,
+      "step": 330
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.58e-05,
-      "loss": 4.2456,
-      "step": 332
+      "loss": 4.2266,
+      "step": 331
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.6e-05,
-      "loss": 4.2544,
-      "step": 333
+      "loss": 4.0707,
+      "step": 332
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 6.620000000000001e-05,
-      "loss": 4.2523,
-      "step": 334
+      "loss": 4.2441,
+      "step": 333
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 6.64e-05,
-      "loss": 4.4533,
-      "step": 335
+      "loss": 4.2306,
+      "step": 334
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 6.66e-05,
-      "loss": 4.1777,
-      "step": 336
+      "loss": 4.3024,
+      "step": 335
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 6.680000000000001e-05,
-      "loss": 4.0961,
-      "step": 337
+      "loss": 4.1858,
+      "step": 336
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.7e-05,
-      "loss": 4.3289,
-      "step": 338
+      "loss": 4.1005,
+      "step": 337
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.720000000000001e-05,
-      "loss": 4.1632,
-      "step": 339
+      "loss": 4.2044,
+      "step": 338
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.740000000000001e-05,
-      "loss": 4.2615,
-      "step": 340
+      "loss": 4.2649,
+      "step": 339
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.76e-05,
-      "loss": 4.6161,
-      "step": 341
+      "loss": 4.267,
+      "step": 340
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.780000000000001e-05,
-      "loss": 4.4731,
-      "step": 342
+      "loss": 4.1318,
+      "step": 341
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 6.800000000000001e-05,
-      "loss": 4.3543,
-      "step": 343
+      "loss": 4.3481,
+      "step": 342
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 6.82e-05,
-      "loss": 4.3963,
-      "step": 344
+      "loss": 4.2143,
+      "step": 343
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 6.840000000000001e-05,
-      "loss": 4.4064,
-      "step": 345
+      "loss": 4.336,
+      "step": 344
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 6.860000000000001e-05,
-      "loss": 4.6316,
-      "step": 346
+      "loss": 4.0903,
+      "step": 345
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.879999999999999e-05,
-      "loss": 4.5705,
-      "step": 347
+      "loss": 4.2886,
+      "step": 346
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.9e-05,
-      "loss": 4.4214,
-      "step": 348
+      "loss": 4.3189,
+      "step": 347
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.92e-05,
-      "loss": 4.5657,
-      "step": 349
+      "loss": 4.2777,
+      "step": 348
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.939999999999999e-05,
-      "loss": 4.4383,
-      "step": 350
+      "loss": 4.3815,
+      "step": 349
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 6.96e-05,
-      "loss": 4.5283,
-      "step": 351
+      "loss": 4.2971,
+      "step": 350
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 6.98e-05,
-      "loss": 4.2436,
-      "step": 352
+      "loss": 4.2663,
+      "step": 351
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 7e-05,
-      "loss": 4.2445,
-      "step": 353
+      "loss": 4.1588,
+      "step": 352
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 7.02e-05,
-      "loss": 4.3244,
-      "step": 354
+      "loss": 4.0897,
+      "step": 353
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 7.04e-05,
-      "loss": 4.9162,
-      "step": 355
+      "loss": 4.0429,
+      "step": 354
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.06e-05,
-      "loss": 4.3424,
-      "step": 356
+      "loss": 4.1905,
+      "step": 355
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.08e-05,
-      "loss": 4.2044,
-      "step": 357
+      "loss": 4.2756,
+      "step": 356
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.1e-05,
-      "loss": 4.4176,
-      "step": 358
+      "loss": 4.2279,
+      "step": 357
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.12e-05,
-      "loss": 4.2067,
-      "step": 359
+      "loss": 4.2192,
+      "step": 358
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 7.14e-05,
-      "loss": 4.3215,
-      "step": 360
+      "loss": 4.0772,
+      "step": 359
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 7.16e-05,
-      "loss": 4.3642,
-      "step": 361
+      "loss": 4.162,
+      "step": 360
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 7.18e-05,
-      "loss": 4.2402,
-      "step": 362
+      "loss": 4.0911,
+      "step": 361
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 7.2e-05,
-      "loss": 4.1775,
-      "step": 363
+      "loss": 4.1129,
+      "step": 362
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 7.22e-05,
-      "loss": 4.4372,
-      "step": 364
+      "loss": 4.0596,
+      "step": 363
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.24e-05,
-      "loss": 4.117,
-      "step": 365
+      "loss": 4.1996,
+      "step": 364
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.26e-05,
-      "loss": 4.4435,
-      "step": 366
+      "loss": 4.2779,
+      "step": 365
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.280000000000001e-05,
-      "loss": 4.1932,
-      "step": 367
+      "loss": 4.1437,
+      "step": 366
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 7.3e-05,
-      "loss": 4.2863,
-      "step": 368
+      "loss": 4.2111,
+      "step": 367
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 7.32e-05,
-      "loss": 4.3986,
-      "step": 369
+      "loss": 4.1603,
+      "step": 368
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 7.340000000000001e-05,
-      "loss": 4.3169,
-      "step": 370
+      "loss": 4.0618,
+      "step": 369
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 7.36e-05,
-      "loss": 4.1458,
-      "step": 371
+      "loss": 4.084,
+      "step": 370
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 7.38e-05,
-      "loss": 4.492,
-      "step": 372
+      "loss": 4.0958,
+      "step": 371
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 7.4e-05,
-      "loss": 4.36,
-      "step": 373
+      "loss": 4.1815,
+      "step": 372
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.42e-05,
-      "loss": 4.4395,
-      "step": 374
+      "loss": 4.161,
+      "step": 373
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.44e-05,
-      "loss": 4.1518,
-      "step": 375
+      "loss": 3.9419,
+      "step": 374
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.46e-05,
-      "loss": 4.2856,
-      "step": 376
+      "loss": 4.2128,
+      "step": 375
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.48e-05,
-      "loss": 4.2022,
-      "step": 377
+      "loss": 4.0403,
+      "step": 376
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 4.4587,
-      "step": 378
+      "loss": 4.1399,
+      "step": 377
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 7.52e-05,
-      "loss": 4.2571,
-      "step": 379
+      "loss": 4.2712,
+      "step": 378
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.54e-05,
-      "loss": 4.4047,
-      "step": 380
+      "loss": 4.141,
+      "step": 379
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.560000000000001e-05,
-      "loss": 4.3593,
-      "step": 381
+      "loss": 4.0592,
+      "step": 380
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.58e-05,
-      "loss": 4.2541,
-      "step": 382
+      "loss": 4.1622,
+      "step": 381
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.6e-05,
-      "loss": 4.4126,
-      "step": 383
+      "loss": 4.221,
+      "step": 382
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.620000000000001e-05,
-      "loss": 4.2624,
-      "step": 384
+      "loss": 4.1675,
+      "step": 383
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.64e-05,
-      "loss": 4.2589,
-      "step": 385
+      "loss": 4.0643,
+      "step": 384
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.66e-05,
-      "loss": 4.447,
-      "step": 386
+      "loss": 4.2672,
+      "step": 385
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 7.680000000000001e-05,
-      "loss": 4.3178,
-      "step": 387
+      "loss": 4.0361,
+      "step": 386
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 7.7e-05,
-      "loss": 4.3143,
-      "step": 388
+      "loss": 4.1967,
+      "step": 387
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.72e-05,
-      "loss": 4.3427,
-      "step": 389
+      "loss": 4.2709,
+      "step": 388
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.740000000000001e-05,
-      "loss": 4.3991,
-      "step": 390
+      "loss": 4.1456,
+      "step": 389
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.76e-05,
-      "loss": 4.5132,
-      "step": 391
+      "loss": 4.1303,
+      "step": 390
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.780000000000001e-05,
-      "loss": 4.3846,
-      "step": 392
+      "loss": 4.3029,
+      "step": 391
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.800000000000001e-05,
-      "loss": 4.3825,
-      "step": 393
+      "loss": 4.103,
+      "step": 392
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.82e-05,
-      "loss": 4.4269,
-      "step": 394
+      "loss": 4.0453,
+      "step": 393
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.840000000000001e-05,
-      "loss": 4.7255,
-      "step": 395
+      "loss": 4.2824,
+      "step": 394
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 7.860000000000001e-05,
-      "loss": 4.4568,
-      "step": 396
+      "loss": 4.376,
+      "step": 395
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.88e-05,
-      "loss": 4.3189,
-      "step": 397
+      "loss": 4.2418,
+      "step": 396
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.900000000000001e-05,
-      "loss": 4.5938,
-      "step": 398
+      "loss": 4.5705,
+      "step": 397
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.920000000000001e-05,
-      "loss": 4.4502,
-      "step": 399
+      "loss": 4.3234,
+      "step": 398
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.94e-05,
-      "loss": 5.0286,
-      "step": 400
+      "loss": 4.7077,
+      "step": 399
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.960000000000001e-05,
-      "loss": 4.5632,
-      "step": 401
+      "loss": 4.3861,
+      "step": 400
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.98e-05,
-      "loss": 4.2655,
-      "step": 402
+      "loss": 4.2475,
+      "step": 401
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 8e-05,
-      "loss": 4.7352,
-      "step": 403
+      "loss": 4.1823,
+      "step": 402
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 8.020000000000001e-05,
-      "loss": 4.4216,
-      "step": 404
+      "loss": 4.2527,
+      "step": 403
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 8.04e-05,
-      "loss": 4.4327,
-      "step": 405
+      "loss": 4.2234,
+      "step": 404
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.060000000000001e-05,
-      "loss": 4.4426,
-      "step": 406
+      "loss": 4.0678,
+      "step": 405
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.080000000000001e-05,
-      "loss": 4.3035,
-      "step": 407
+      "loss": 4.2486,
+      "step": 406
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.1e-05,
-      "loss": 4.3669,
-      "step": 408
+      "loss": 4.1372,
+      "step": 407
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.120000000000001e-05,
-      "loss": 4.4913,
-      "step": 409
+      "loss": 4.2349,
+      "step": 408
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.14e-05,
-      "loss": 4.4242,
-      "step": 410
+      "loss": 4.1296,
+      "step": 409
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.16e-05,
-      "loss": 4.2835,
-      "step": 411
+      "loss": 4.1051,
+      "step": 410
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.18e-05,
-      "loss": 4.2727,
-      "step": 412
+      "loss": 4.3548,
+      "step": 411
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.2e-05,
-      "loss": 4.5275,
-      "step": 413
+      "loss": 4.1184,
+      "step": 412
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 8.22e-05,
-      "loss": 4.1393,
-      "step": 414
+      "loss": 4.0538,
+      "step": 413
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.24e-05,
-      "loss": 4.2596,
-      "step": 415
+      "loss": 4.2076,
+      "step": 414
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.26e-05,
-      "loss": 4.3531,
-      "step": 416
+      "loss": 4.3461,
+      "step": 415
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.28e-05,
-      "loss": 4.1844,
-      "step": 417
+      "loss": 4.0961,
+      "step": 416
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.3e-05,
-      "loss": 4.4706,
-      "step": 418
+      "loss": 4.2111,
+      "step": 417
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.32e-05,
-      "loss": 4.2229,
-      "step": 419
+      "loss": 4.0412,
+      "step": 418
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.34e-05,
-      "loss": 4.2738,
-      "step": 420
+      "loss": 4.2169,
+      "step": 419
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.36e-05,
-      "loss": 4.3399,
-      "step": 421
+      "loss": 4.0514,
+      "step": 420
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.38e-05,
-      "loss": 4.4168,
-      "step": 422
+      "loss": 4.3228,
+      "step": 421
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 8.4e-05,
-      "loss": 4.33,
-      "step": 423
+      "loss": 4.0805,
+      "step": 422
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.42e-05,
-      "loss": 4.7077,
-      "step": 424
+      "loss": 4.1906,
+      "step": 423
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.44e-05,
-      "loss": 4.4407,
-      "step": 425
+      "loss": 4.2743,
+      "step": 424
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.46e-05,
-      "loss": 4.4444,
-      "step": 426
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.48e-05,
-      "loss": 4.328,
-      "step": 427
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.5e-05,
-      "loss": 4.682,
-      "step": 428
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.52e-05,
-      "loss": 4.3094,
-      "step": 429
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.54e-05,
-      "loss": 4.4006,
-      "step": 430
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.560000000000001e-05,
-      "loss": 4.5697,
-      "step": 431
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.58e-05,
-      "loss": 4.3803,
-      "step": 432
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.6e-05,
-      "loss": 4.238,
-      "step": 433
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.620000000000001e-05,
-      "loss": 4.455,
-      "step": 434
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.64e-05,
-      "loss": 4.274,
-      "step": 435
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.66e-05,
-      "loss": 4.538,
-      "step": 436
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.680000000000001e-05,
-      "loss": 4.6364,
-      "step": 437
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.7e-05,
-      "loss": 4.4224,
-      "step": 438
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.72e-05,
-      "loss": 4.4031,
-      "step": 439
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 4.4081,
-      "step": 440
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.76e-05,
-      "loss": 4.5072,
-      "step": 441
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.78e-05,
-      "loss": 4.2758,
-      "step": 442
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 4.5096,
-      "step": 443
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.82e-05,
-      "loss": 4.6415,
-      "step": 444
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.840000000000001e-05,
-      "loss": 4.557,
-      "step": 445
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.86e-05,
-      "loss": 4.1481,
-      "step": 446
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.88e-05,
-      "loss": 4.2754,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.900000000000001e-05,
-      "loss": 4.545,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.92e-05,
-      "loss": 4.7661,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.94e-05,
-      "loss": 5.6649,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.960000000000001e-05,
-      "loss": 4.4274,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.98e-05,
-      "loss": 4.4116,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9e-05,
-      "loss": 4.7116,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.020000000000001e-05,
-      "loss": 4.2829,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.04e-05,
-      "loss": 4.4765,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.06e-05,
-      "loss": 4.4396,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.080000000000001e-05,
-      "loss": 4.3675,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.1e-05,
-      "loss": 4.3934,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.120000000000001e-05,
-      "loss": 4.391,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.140000000000001e-05,
-      "loss": 4.3971,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.16e-05,
-      "loss": 4.164,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.180000000000001e-05,
-      "loss": 4.3476,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 4.505,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.22e-05,
-      "loss": 4.1572,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.240000000000001e-05,
-      "loss": 4.2912,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.260000000000001e-05,
-      "loss": 4.5576,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.28e-05,
-      "loss": 4.4265,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.300000000000001e-05,
-      "loss": 4.4557,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.320000000000002e-05,
-      "loss": 4.3606,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.340000000000001e-05,
-      "loss": 4.3158,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.360000000000001e-05,
-      "loss": 4.1388,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.38e-05,
-      "loss": 4.2528,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.4e-05,
-      "loss": 4.833,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.42e-05,
-      "loss": 4.438,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.44e-05,
-      "loss": 4.3888,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.46e-05,
-      "loss": 4.7277,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.48e-05,
-      "loss": 4.658,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.5e-05,
-      "loss": 4.2405,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.52e-05,
-      "loss": 4.1582,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.54e-05,
-      "loss": 4.3365,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.56e-05,
-      "loss": 4.2135,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.58e-05,
-      "loss": 4.1966,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.6e-05,
-      "loss": 4.4412,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.620000000000001e-05,
-      "loss": 4.4778,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.64e-05,
-      "loss": 4.5488,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.66e-05,
-      "loss": 4.267,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.680000000000001e-05,
-      "loss": 4.2701,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.7e-05,
-      "loss": 4.2323,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.72e-05,
-      "loss": 4.4346,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.74e-05,
-      "loss": 4.5228,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.76e-05,
-      "loss": 4.1844,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.78e-05,
-      "loss": 4.6989,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.8e-05,
-      "loss": 4.673,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.82e-05,
-      "loss": 4.3557,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.84e-05,
-      "loss": 4.6377,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.86e-05,
-      "loss": 4.1909,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.88e-05,
-      "loss": 4.1532,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.900000000000001e-05,
-      "loss": 4.8486,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.92e-05,
-      "loss": 4.766,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.94e-05,
-      "loss": 4.7566,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 4.591245651245117,
-      "eval_runtime": 931.3082,
-      "eval_samples_per_second": 2.837,
-      "eval_steps_per_second": 0.355,
-      "eval_wer": 1.96449821499405,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.960000000000001e-05,
-      "loss": 4.5959,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.98e-05,
-      "loss": 4.1684,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0001,
-      "loss": 4.6255,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.974489795918368e-05,
-      "loss": 4.3689,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.948979591836736e-05,
-      "loss": 4.2557,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.923469387755102e-05,
-      "loss": 4.326,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.897959183673469e-05,
-      "loss": 4.596,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.872448979591837e-05,
-      "loss": 4.5114,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.846938775510204e-05,
-      "loss": 4.468,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.821428571428572e-05,
-      "loss": 4.5804,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.79591836734694e-05,
-      "loss": 4.4155,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.770408163265307e-05,
-      "loss": 4.4633,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.744897959183674e-05,
-      "loss": 4.9784,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.719387755102042e-05,
-      "loss": 4.3901,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.693877551020408e-05,
-      "loss": 4.4349,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.668367346938776e-05,
-      "loss": 4.2123,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.642857142857143e-05,
-      "loss": 4.5308,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.617346938775511e-05,
-      "loss": 4.3677,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.591836734693878e-05,
-      "loss": 4.7828,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.566326530612246e-05,
-      "loss": 4.4524,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.540816326530613e-05,
-      "loss": 4.354,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.515306122448981e-05,
-      "loss": 4.2494,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.489795918367348e-05,
-      "loss": 4.5313,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.464285714285715e-05,
-      "loss": 4.4845,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.438775510204082e-05,
-      "loss": 4.4332,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.41326530612245e-05,
-      "loss": 4.3977,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.387755102040817e-05,
-      "loss": 4.326,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.362244897959183e-05,
-      "loss": 4.4473,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.336734693877551e-05,
-      "loss": 4.7425,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.311224489795918e-05,
-      "loss": 4.6354,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.285714285714286e-05,
-      "loss": 4.5632,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.260204081632653e-05,
-      "loss": 4.317,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.234693877551021e-05,
-      "loss": 4.4167,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.209183673469388e-05,
-      "loss": 4.2363,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.183673469387756e-05,
-      "loss": 4.5831,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.158163265306124e-05,
-      "loss": 4.7649,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.13265306122449e-05,
-      "loss": 4.5653,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.107142857142857e-05,
-      "loss": 4.6266,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.081632653061225e-05,
-      "loss": 4.2722,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.056122448979592e-05,
-      "loss": 4.5449,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.030612244897958e-05,
-      "loss": 4.5396,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.005102040816327e-05,
-      "loss": 4.4967,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.979591836734695e-05,
-      "loss": 4.5855,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.954081632653062e-05,
-      "loss": 4.7686,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.92857142857143e-05,
-      "loss": 4.7957,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.903061224489796e-05,
-      "loss": 4.9872,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.877551020408164e-05,
-      "loss": 5.2598,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.852040816326531e-05,
-      "loss": 4.8536,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.826530612244899e-05,
-      "loss": 4.2885,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.801020408163265e-05,
-      "loss": 5.2163,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.775510204081632e-05,
-      "loss": 4.4,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.75e-05,
-      "loss": 4.5511,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.724489795918367e-05,
-      "loss": 4.5211,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.698979591836735e-05,
-      "loss": 4.2617,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.673469387755102e-05,
-      "loss": 4.2831,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.64795918367347e-05,
-      "loss": 4.3512,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.622448979591838e-05,
-      "loss": 4.1824,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.596938775510205e-05,
-      "loss": 4.2188,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.571428571428571e-05,
-      "loss": 4.3538,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.545918367346939e-05,
-      "loss": 4.2245,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.520408163265306e-05,
-      "loss": 4.5255,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.494897959183674e-05,
-      "loss": 4.1717,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.469387755102041e-05,
-      "loss": 4.3354,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.443877551020409e-05,
-      "loss": 4.3604,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.418367346938776e-05,
-      "loss": 4.4324,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.392857142857144e-05,
-      "loss": 4.424,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.367346938775511e-05,
-      "loss": 4.344,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.341836734693878e-05,
-      "loss": 4.3265,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.316326530612245e-05,
-      "loss": 4.3604,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.290816326530613e-05,
-      "loss": 4.4938,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.26530612244898e-05,
-      "loss": 4.2892,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.239795918367348e-05,
-      "loss": 4.3934,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.214285714285714e-05,
-      "loss": 4.4141,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.188775510204081e-05,
-      "loss": 4.1242,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.163265306122449e-05,
-      "loss": 4.5223,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.137755102040817e-05,
-      "loss": 4.3807,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.112244897959184e-05,
-      "loss": 4.355,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.086734693877552e-05,
-      "loss": 4.7096,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.061224489795919e-05,
-      "loss": 3.9647,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.035714285714287e-05,
-      "loss": 4.5312,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.010204081632653e-05,
-      "loss": 4.4768,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.98469387755102e-05,
-      "loss": 4.1437,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.959183673469388e-05,
-      "loss": 4.3517,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.933673469387755e-05,
-      "loss": 4.3499,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.908163265306123e-05,
-      "loss": 4.5748,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.882653061224489e-05,
-      "loss": 4.4299,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.857142857142858e-05,
-      "loss": 4.6094,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.831632653061226e-05,
-      "loss": 4.4316,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.806122448979593e-05,
-      "loss": 4.2616,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.780612244897959e-05,
-      "loss": 4.4071,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.755102040816327e-05,
-      "loss": 4.3172,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.729591836734694e-05,
-      "loss": 4.5913,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.704081632653062e-05,
-      "loss": 4.6859,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.67857142857143e-05,
-      "loss": 4.3994,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.653061224489796e-05,
-      "loss": 4.6602,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.627551020408163e-05,
-      "loss": 4.316,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.60204081632653e-05,
-      "loss": 4.3982,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.576530612244898e-05,
-      "loss": 4.7021,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.551020408163266e-05,
-      "loss": 4.6899,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.525510204081633e-05,
-      "loss": 4.3161,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 4.4666,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.474489795918368e-05,
-      "loss": 4.3151,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.448979591836736e-05,
-      "loss": 4.3528,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.423469387755102e-05,
-      "loss": 4.3529,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.39795918367347e-05,
-      "loss": 4.2619,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.372448979591837e-05,
-      "loss": 4.2584,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.346938775510205e-05,
-      "loss": 4.4214,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.321428571428571e-05,
-      "loss": 4.2431,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.29591836734694e-05,
-      "loss": 4.2831,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.270408163265307e-05,
-      "loss": 4.4768,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.244897959183675e-05,
-      "loss": 4.4395,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.219387755102042e-05,
-      "loss": 4.4869,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.193877551020408e-05,
-      "loss": 4.2931,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.168367346938776e-05,
-      "loss": 4.664,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.142857142857143e-05,
-      "loss": 4.6411,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.117346938775511e-05,
-      "loss": 4.3625,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.091836734693877e-05,
-      "loss": 4.217,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.066326530612245e-05,
-      "loss": 4.2248,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.040816326530612e-05,
-      "loss": 4.1776,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.01530612244898e-05,
-      "loss": 4.3362,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.989795918367347e-05,
-      "loss": 4.4147,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.964285714285715e-05,
-      "loss": 4.2179,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.938775510204082e-05,
-      "loss": 4.4038,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.91326530612245e-05,
-      "loss": 4.4807,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.887755102040817e-05,
-      "loss": 4.7165,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.862244897959184e-05,
-      "loss": 4.4327,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.836734693877551e-05,
-      "loss": 4.2958,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.811224489795919e-05,
-      "loss": 4.2159,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.785714285714286e-05,
-      "loss": 4.2481,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.760204081632652e-05,
-      "loss": 4.6183,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.73469387755102e-05,
-      "loss": 4.3344,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.709183673469389e-05,
-      "loss": 4.2177,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.683673469387756e-05,
-      "loss": 4.3506,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.658163265306124e-05,
-      "loss": 4.2825,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.63265306122449e-05,
-      "loss": 4.6362,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.607142857142857e-05,
-      "loss": 4.3973,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.581632653061225e-05,
-      "loss": 4.3994,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.556122448979592e-05,
-      "loss": 4.2977,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.530612244897959e-05,
-      "loss": 4.5307,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.505102040816326e-05,
-      "loss": 4.2997,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.479591836734694e-05,
-      "loss": 4.267,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.454081632653061e-05,
-      "loss": 4.6337,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.428571428571429e-05,
-      "loss": 4.9211,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.403061224489796e-05,
-      "loss": 4.2347,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.377551020408164e-05,
-      "loss": 4.0541,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.352040816326531e-05,
-      "loss": 4.3861,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.326530612244899e-05,
-      "loss": 4.2551,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.301020408163265e-05,
-      "loss": 4.2855,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.275510204081633e-05,
-      "loss": 4.648,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.25e-05,
-      "loss": 4.6074,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.224489795918368e-05,
-      "loss": 4.4537,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.198979591836735e-05,
-      "loss": 4.4171,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.173469387755101e-05,
-      "loss": 4.3433,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.14795918367347e-05,
-      "loss": 4.4906,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.122448979591838e-05,
-      "loss": 4.3407,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.0969387755102046e-05,
-      "loss": 4.2764,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.0714285714285715e-05,
-      "loss": 4.3259,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.045918367346939e-05,
-      "loss": 4.5668,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.0204081632653065e-05,
-      "loss": 4.2393,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.994897959183674e-05,
-      "loss": 4.4674,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.9693877551020416e-05,
-      "loss": 4.145,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.9438775510204084e-05,
-      "loss": 4.061,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.918367346938776e-05,
-      "loss": 4.5757,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.8928571428571435e-05,
-      "loss": 4.1992,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.867346938775511e-05,
-      "loss": 4.1435,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.841836734693877e-05,
-      "loss": 4.4722,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.816326530612245e-05,
-      "loss": 4.1846,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.790816326530612e-05,
-      "loss": 4.4433,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.7653061224489805e-05,
-      "loss": 4.417,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.739795918367348e-05,
-      "loss": 4.3992,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.714285714285714e-05,
-      "loss": 4.1534,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.688775510204082e-05,
-      "loss": 4.2826,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.663265306122449e-05,
-      "loss": 4.1582,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.637755102040817e-05,
-      "loss": 4.0745,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.6122448979591836e-05,
-      "loss": 4.1668,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.586734693877551e-05,
-      "loss": 4.3828,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.561224489795919e-05,
-      "loss": 4.116,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.535714285714286e-05,
-      "loss": 4.0616,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.510204081632653e-05,
-      "loss": 4.2286,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.4846938775510206e-05,
-      "loss": 4.2636,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.459183673469388e-05,
-      "loss": 4.6022,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.4336734693877556e-05,
-      "loss": 4.103,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.408163265306123e-05,
-      "loss": 4.3669,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.382653061224489e-05,
-      "loss": 4.6126,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.3571428571428575e-05,
-      "loss": 4.4961,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.331632653061225e-05,
-      "loss": 4.2818,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.3061224489795926e-05,
-      "loss": 4.2816,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.280612244897959e-05,
-      "loss": 4.4225,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.255102040816326e-05,
-      "loss": 4.4492,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.229591836734694e-05,
-      "loss": 4.4852,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.2040816326530614e-05,
-      "loss": 4.2335,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1785714285714296e-05,
-      "loss": 4.4825,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.153061224489796e-05,
-      "loss": 4.1698,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.127551020408163e-05,
-      "loss": 4.7683,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.102040816326531e-05,
-      "loss": 4.7465,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.076530612244898e-05,
-      "loss": 4.4719,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.051020408163265e-05,
-      "loss": 4.7418,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.025510204081633e-05,
-      "loss": 4.3636,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5e-05,
-      "loss": 4.4641,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 4.974489795918368e-05,
-      "loss": 4.5259,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.9489795918367346e-05,
-      "loss": 4.1317,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.923469387755102e-05,
-      "loss": 4.3063,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.89795918367347e-05,
-      "loss": 4.1669,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.872448979591837e-05,
-      "loss": 4.1128,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.846938775510204e-05,
-      "loss": 4.3668,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.8214285714285716e-05,
-      "loss": 4.5481,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.795918367346939e-05,
-      "loss": 4.4199,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.7704081632653066e-05,
-      "loss": 4.3895,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.744897959183674e-05,
-      "loss": 4.3989,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.719387755102041e-05,
-      "loss": 4.3812,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.6938775510204086e-05,
-      "loss": 4.1986,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.6683673469387754e-05,
-      "loss": 4.1714,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.642857142857143e-05,
-      "loss": 4.2361,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.6173469387755105e-05,
-      "loss": 4.4285,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.591836734693878e-05,
-      "loss": 4.3898,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.566326530612245e-05,
-      "loss": 4.2292,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.5408163265306124e-05,
-      "loss": 4.5586,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.515306122448979e-05,
-      "loss": 4.1922,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.4897959183673474e-05,
-      "loss": 4.0818,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.464285714285715e-05,
-      "loss": 4.2948,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.438775510204082e-05,
-      "loss": 4.208,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.4132653061224493e-05,
-      "loss": 4.1011,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.387755102040816e-05,
-      "loss": 4.3333,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.362244897959184e-05,
-      "loss": 4.338,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.336734693877551e-05,
-      "loss": 4.1581,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.311224489795919e-05,
-      "loss": 4.347,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.2857142857142856e-05,
-      "loss": 4.2181,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.260204081632653e-05,
-      "loss": 4.2544,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.234693877551021e-05,
-      "loss": 4.1822,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.209183673469388e-05,
-      "loss": 4.3612,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.183673469387756e-05,
-      "loss": 4.7029,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.1581632653061226e-05,
-      "loss": 4.3051,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.13265306122449e-05,
-      "loss": 4.1392,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.107142857142857e-05,
-      "loss": 4.4285,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.0816326530612245e-05,
-      "loss": 4.0742,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.056122448979592e-05,
-      "loss": 4.1706,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.0306122448979596e-05,
-      "loss": 4.3423,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.0051020408163264e-05,
-      "loss": 4.2188,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.979591836734694e-05,
-      "loss": 4.1841,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.9540816326530615e-05,
-      "loss": 4.3884,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.928571428571429e-05,
-      "loss": 4.5054,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.9030612244897965e-05,
-      "loss": 4.1764,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.8775510204081634e-05,
-      "loss": 4.0358,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.852040816326531e-05,
-      "loss": 4.2615,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.826530612244898e-05,
-      "loss": 4.3648,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.801020408163265e-05,
-      "loss": 4.4862,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.775510204081633e-05,
-      "loss": 4.2654,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7500000000000003e-05,
-      "loss": 4.4289,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.724489795918368e-05,
-      "loss": 4.8704,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.698979591836735e-05,
-      "loss": 4.3239,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.673469387755102e-05,
-      "loss": 4.3253,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.64795918367347e-05,
-      "loss": 4.2496,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.622448979591837e-05,
-      "loss": 4.3051,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.596938775510204e-05,
-      "loss": 4.2689,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.571428571428572e-05,
-      "loss": 4.4432,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5459183673469385e-05,
-      "loss": 4.3211,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.520408163265306e-05,
-      "loss": 4.3471,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4948979591836736e-05,
-      "loss": 4.3602,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.469387755102041e-05,
-      "loss": 4.1401,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.443877551020409e-05,
-      "loss": 4.2309,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4183673469387755e-05,
-      "loss": 4.3356,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.392857142857143e-05,
-      "loss": 4.236,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.36734693877551e-05,
-      "loss": 4.1982,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.341836734693878e-05,
-      "loss": 4.1095,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.316326530612245e-05,
-      "loss": 4.348,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2908163265306125e-05,
-      "loss": 4.1909,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.265306122448979e-05,
-      "loss": 4.172,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.239795918367347e-05,
-      "loss": 4.2519,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2142857142857144e-05,
-      "loss": 4.2602,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.188775510204082e-05,
-      "loss": 4.3414,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.1632653061224494e-05,
-      "loss": 4.1333,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.137755102040816e-05,
-      "loss": 4.37,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.112244897959184e-05,
-      "loss": 4.4521,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.086734693877551e-05,
-      "loss": 4.0012,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.061224489795919e-05,
-      "loss": 4.0225,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0357142857142857e-05,
-      "loss": 4.4277,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0102040816326533e-05,
-      "loss": 3.9975,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.9846938775510208e-05,
-      "loss": 4.3869,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.959183673469388e-05,
-      "loss": 4.1808,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.9336734693877555e-05,
-      "loss": 4.2401,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.9081632653061224e-05,
-      "loss": 4.0887,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8826530612244902e-05,
-      "loss": 4.3455,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.857142857142857e-05,
-      "loss": 4.3743,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8316326530612246e-05,
-      "loss": 4.1141,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8061224489795918e-05,
-      "loss": 4.2681,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7806122448979593e-05,
-      "loss": 4.0444,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7551020408163265e-05,
-      "loss": 4.4677,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.729591836734694e-05,
-      "loss": 4.295,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7040816326530616e-05,
-      "loss": 4.1983,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6785714285714288e-05,
-      "loss": 4.3102,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6530612244897963e-05,
-      "loss": 4.2775,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.627551020408163e-05,
-      "loss": 4.1654,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6020408163265307e-05,
-      "loss": 4.1957,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.576530612244898e-05,
-      "loss": 4.3824,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5510204081632654e-05,
-      "loss": 4.3644,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5255102040816326e-05,
-      "loss": 4.179,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5e-05,
-      "loss": 4.3939,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.4744897959183673e-05,
-      "loss": 4.4674,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.448979591836735e-05,
-      "loss": 4.8266,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.423469387755102e-05,
-      "loss": 4.6286,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3979591836734696e-05,
-      "loss": 4.2513,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.372448979591837e-05,
-      "loss": 4.1684,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3469387755102043e-05,
-      "loss": 4.1856,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3214285714285715e-05,
-      "loss": 4.2883,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.295918367346939e-05,
-      "loss": 4.0168,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.2704081632653062e-05,
-      "loss": 4.2353,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.2448979591836737e-05,
-      "loss": 4.1928,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.219387755102041e-05,
-      "loss": 4.071,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.193877551020408e-05,
-      "loss": 4.3744,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1683673469387756e-05,
-      "loss": 4.0712,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1428571428571428e-05,
-      "loss": 4.4198,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1173469387755103e-05,
-      "loss": 4.2845,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.091836734693878e-05,
-      "loss": 4.0229,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.066326530612245e-05,
-      "loss": 4.1555,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0408163265306123e-05,
-      "loss": 4.1574,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0153061224489798e-05,
-      "loss": 4.1121,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.989795918367347e-05,
-      "loss": 4.1488,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.9642857142857145e-05,
-      "loss": 4.1536,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.9387755102040817e-05,
-      "loss": 4.1018,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.913265306122449e-05,
-      "loss": 4.2319,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8877551020408164e-05,
-      "loss": 4.3041,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.862244897959184e-05,
-      "loss": 4.0551,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.836734693877551e-05,
-      "loss": 4.1893,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8112244897959187e-05,
-      "loss": 4.2017,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.785714285714286e-05,
-      "loss": 4.434,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.760204081632653e-05,
-      "loss": 4.3849,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7346938775510206e-05,
-      "loss": 4.176,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7091836734693878e-05,
-      "loss": 4.0998,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.683673469387755e-05,
-      "loss": 4.1019,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6581632653061225e-05,
-      "loss": 4.3222,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6326530612244897e-05,
-      "loss": 4.0529,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6071428571428572e-05,
-      "loss": 4.08,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.5816326530612247e-05,
-      "loss": 4.2724,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.556122448979592e-05,
-      "loss": 4.1382,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5306122448979594e-05,
-      "loss": 4.1915,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5051020408163266e-05,
-      "loss": 4.1702,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.479591836734694e-05,
-      "loss": 4.1094,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4540816326530612e-05,
-      "loss": 4.4106,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4285714285714285e-05,
-      "loss": 4.0532,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4030612244897959e-05,
-      "loss": 4.1847,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3775510204081633e-05,
-      "loss": 4.5441,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3520408163265308e-05,
-      "loss": 4.198,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3265306122448982e-05,
-      "loss": 4.4227,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3010204081632653e-05,
-      "loss": 4.3613,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2755102040816327e-05,
-      "loss": 4.5263,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.25e-05,
-      "loss": 4.6235,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2244897959183674e-05,
-      "loss": 4.2742,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1989795918367348e-05,
-      "loss": 5.0247,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1734693877551021e-05,
-      "loss": 4.5391,
-      "step": 849
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1479591836734695e-05,
-      "loss": 4.3746,
-      "step": 850
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1224489795918369e-05,
-      "loss": 4.2074,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.096938775510204e-05,
-      "loss": 4.1277,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0714285714285714e-05,
-      "loss": 4.1905,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.045918367346939e-05,
-      "loss": 4.1016,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0204081632653061e-05,
-      "loss": 4.1227,
-      "step": 855
+      "loss": 4.2114,
+      "step": 425
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.948979591836735e-06,
-      "loss": 4.158,
-      "step": 856
+      "learning_rate": 8.48e-05,
+      "loss": 4.1866,
+      "step": 426
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.693877551020408e-06,
-      "loss": 4.2657,
-      "step": 857
+      "learning_rate": 8.5e-05,
+      "loss": 4.1642,
+      "step": 427
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.438775510204082e-06,
-      "loss": 4.2059,
-      "step": 858
+      "learning_rate": 8.52e-05,
+      "loss": 4.1608,
+      "step": 428
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.183673469387756e-06,
-      "loss": 4.0288,
-      "step": 859
+      "learning_rate": 8.54e-05,
+      "loss": 4.0655,
+      "step": 429
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.92857142857143e-06,
-      "loss": 4.195,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.673469387755103e-06,
-      "loss": 4.4489,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.418367346938775e-06,
-      "loss": 4.1907,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.163265306122448e-06,
-      "loss": 4.2896,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.908163265306124e-06,
-      "loss": 4.5305,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.653061224489797e-06,
-      "loss": 4.2739,
-      "step": 865
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 4.2002,
+      "step": 430
     },
     {
       "epoch": 0.97,
-      "learning_rate": 7.39795918367347e-06,
-      "loss": 4.0737,
-      "step": 866
+      "learning_rate": 8.58e-05,
+      "loss": 4.1811,
+      "step": 431
     },
     {
       "epoch": 0.97,
-      "learning_rate": 7.142857142857143e-06,
-      "loss": 4.1886,
-      "step": 867
+      "learning_rate": 8.6e-05,
+      "loss": 4.3704,
+      "step": 432
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.887755102040816e-06,
-      "loss": 4.1191,
-      "step": 868
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 4.1261,
+      "step": 433
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.632653061224491e-06,
-      "loss": 4.2065,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.3775510204081635e-06,
-      "loss": 4.3539,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.122448979591837e-06,
-      "loss": 4.0906,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.867346938775511e-06,
-      "loss": 4.3103,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.612244897959184e-06,
-      "loss": 4.2243,
-      "step": 873
+      "learning_rate": 8.64e-05,
+      "loss": 4.1063,
+      "step": 434
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.357142857142857e-06,
-      "loss": 4.2663,
-      "step": 874
+      "learning_rate": 8.66e-05,
+      "loss": 4.1748,
+      "step": 435
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.102040816326531e-06,
-      "loss": 4.3513,
-      "step": 875
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 4.0705,
+      "step": 436
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.846938775510204e-06,
-      "loss": 4.1566,
-      "step": 876
+      "learning_rate": 8.7e-05,
+      "loss": 4.1026,
+      "step": 437
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.591836734693878e-06,
-      "loss": 4.1534,
-      "step": 877
+      "learning_rate": 8.72e-05,
+      "loss": 4.3491,
+      "step": 438
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.336734693877551e-06,
-      "loss": 4.1904,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.081632653061224e-06,
-      "loss": 4.2973,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.826530612244899e-06,
-      "loss": 4.5052,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.5714285714285714e-06,
-      "loss": 4.065,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.3163265306122454e-06,
-      "loss": 4.219,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.0612244897959185e-06,
-      "loss": 4.2132,
-      "step": 883
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 4.2061,
+      "step": 439
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.806122448979592e-06,
-      "loss": 4.2218,
-      "step": 884
+      "learning_rate": 8.76e-05,
+      "loss": 4.2527,
+      "step": 440
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.5510204081632653e-06,
-      "loss": 4.2322,
-      "step": 885
+      "learning_rate": 8.78e-05,
+      "loss": 4.3031,
+      "step": 441
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.295918367346939e-06,
-      "loss": 4.3062,
-      "step": 886
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 4.3801,
+      "step": 442
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.040816326530612e-06,
-      "loss": 4.0873,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.7857142857142857e-06,
-      "loss": 4.3274,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.5306122448979593e-06,
-      "loss": 4.4504,
-      "step": 889
+      "learning_rate": 8.82e-05,
+      "loss": 4.467,
+      "step": 443
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.2755102040816327e-06,
-      "loss": 4.1995,
-      "step": 890
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 4.3737,
+      "step": 444
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.020408163265306e-06,
-      "loss": 4.4025,
-      "step": 891
+      "learning_rate": 8.86e-05,
+      "loss": 4.5371,
+      "step": 445
     },
     {
       "epoch": 1.0,
-      "learning_rate": 7.653061224489796e-07,
-      "loss": 3.9803,
-      "step": 892
+      "learning_rate": 8.88e-05,
+      "loss": 4.1867,
+      "step": 446
     },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 4.35051649381227,
-      "train_runtime": 6248.9422,
-      "train_samples_per_second": 4.567,
-      "train_steps_per_second": 0.143
+      "train_loss": 4.256010225535508,
+      "train_runtime": 5087.4578,
+      "train_samples_per_second": 5.609,
+      "train_steps_per_second": 0.088
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,