diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,2699 +1,7175 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
-  "global_step": 446,
+  "epoch": 0.9995796553173603,
+  "global_step": 1189,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 4.747,
+      "learning_rate": 0.0,
+      "loss": 4.8663,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.9006,
+      "learning_rate": 0.0,
+      "loss": 5.0546,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.8744,
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 5.0571,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.8574,
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.8722,
       "step": 4
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.7861,
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.6058,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.8432,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.9662,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.8179,
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.7335,
       "step": 7
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.8564,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.7097,
       "step": 8
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.8332,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.9142,
       "step": 9
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.8e-06,
-      "loss": 4.7164,
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.7525,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.8208,
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.8107,
       "step": 11
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.2e-06,
-      "loss": 4.7747,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 4.6655,
       "step": 12
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.7053,
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.9272,
       "step": 13
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.6e-06,
-      "loss": 4.7023,
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 4.7069,
       "step": 14
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.5744,
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.5746,
       "step": 15
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3e-06,
-      "loss": 4.7191,
+      "epoch": 0.01,
+      "learning_rate": 2.6e-06,
+      "loss": 4.7888,
       "step": 16
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.6175,
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.7284,
       "step": 17
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.6957,
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 4.8545,
       "step": 18
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.6e-06,
-      "loss": 4.4988,
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.5886,
       "step": 19
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.8e-06,
-      "loss": 4.6422,
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.7927,
       "step": 20
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.6299,
+      "epoch": 0.02,
+      "learning_rate": 3.6e-06,
+      "loss": 4.7407,
       "step": 21
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.5586,
+      "epoch": 0.02,
+      "learning_rate": 3.8e-06,
+      "loss": 4.5483,
       "step": 22
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.4e-06,
-      "loss": 4.6489,
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.6978,
       "step": 23
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.6e-06,
-      "loss": 4.614,
+      "epoch": 0.02,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.7197,
       "step": 24
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.4035,
+      "epoch": 0.02,
+      "learning_rate": 4.4e-06,
+      "loss": 4.5343,
       "step": 25
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5e-06,
-      "loss": 4.6675,
+      "epoch": 0.02,
+      "learning_rate": 4.6e-06,
+      "loss": 4.6812,
       "step": 26
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.2e-06,
-      "loss": 4.5006,
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.5176,
       "step": 27
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.4e-06,
-      "loss": 4.3643,
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 4.3872,
       "step": 28
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.3192,
+      "epoch": 0.02,
+      "learning_rate": 5.2e-06,
+      "loss": 4.4441,
       "step": 29
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 5.8e-06,
-      "loss": 4.5311,
+      "epoch": 0.03,
+      "learning_rate": 5.4e-06,
+      "loss": 4.4824,
       "step": 30
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6e-06,
-      "loss": 4.5042,
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.5984,
       "step": 31
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.2e-06,
-      "loss": 4.3526,
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 4.3995,
       "step": 32
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.4255,
+      "epoch": 0.03,
+      "learning_rate": 6e-06,
+      "loss": 4.5001,
       "step": 33
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 6.6e-06,
-      "loss": 4.3723,
+      "epoch": 0.03,
+      "learning_rate": 6.2e-06,
+      "loss": 4.1968,
       "step": 34
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.4424,
+      "epoch": 0.03,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.3981,
       "step": 35
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 4.3992,
+      "epoch": 0.03,
+      "learning_rate": 6.6e-06,
+      "loss": 4.6696,
       "step": 36
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.2e-06,
-      "loss": 4.4579,
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.5091,
       "step": 37
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.4e-06,
-      "loss": 4.385,
+      "epoch": 0.03,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 4.3622,
       "step": 38
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.6e-06,
-      "loss": 4.5033,
+      "epoch": 0.03,
+      "learning_rate": 7.2e-06,
+      "loss": 4.8527,
       "step": 39
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.8e-06,
-      "loss": 4.3482,
+      "epoch": 0.03,
+      "learning_rate": 7.4e-06,
+      "loss": 4.5175,
       "step": 40
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.3762,
+      "epoch": 0.03,
+      "learning_rate": 7.6e-06,
+      "loss": 4.2974,
       "step": 41
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 4.4784,
+      "epoch": 0.04,
+      "learning_rate": 7.8e-06,
+      "loss": 4.3827,
       "step": 42
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.2417,
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.3609,
       "step": 43
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 4.4203,
+      "epoch": 0.04,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 4.4077,
       "step": 44
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.8e-06,
-      "loss": 4.5079,
+      "epoch": 0.04,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.5765,
       "step": 45
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9e-06,
-      "loss": 4.5547,
+      "epoch": 0.04,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 4.5387,
       "step": 46
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.2e-06,
-      "loss": 4.4638,
+      "epoch": 0.04,
+      "learning_rate": 8.8e-06,
+      "loss": 4.7018,
       "step": 47
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.4e-06,
-      "loss": 4.699,
+      "epoch": 0.04,
+      "learning_rate": 9e-06,
+      "loss": 4.9332,
       "step": 48
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.8242,
+      "epoch": 0.04,
+      "learning_rate": 9.2e-06,
+      "loss": 5.083,
       "step": 49
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.4994,
+      "epoch": 0.04,
+      "learning_rate": 9.2e-06,
+      "loss": 4.8089,
       "step": 50
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.3463,
+      "epoch": 0.04,
+      "learning_rate": 9.4e-06,
+      "loss": 4.3204,
       "step": 51
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1e-05,
-      "loss": 4.2201,
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.3233,
       "step": 52
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.02e-05,
-      "loss": 4.3763,
+      "epoch": 0.04,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.2036,
       "step": 53
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.04e-05,
-      "loss": 4.2946,
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 4.2543,
       "step": 54
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.06e-05,
-      "loss": 4.1038,
+      "epoch": 0.05,
+      "learning_rate": 1.02e-05,
+      "loss": 4.4195,
       "step": 55
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.08e-05,
-      "loss": 4.3279,
+      "epoch": 0.05,
+      "learning_rate": 1.04e-05,
+      "loss": 4.2281,
       "step": 56
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 4.3403,
+      "epoch": 0.05,
+      "learning_rate": 1.06e-05,
+      "loss": 4.4942,
       "step": 57
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 4.4244,
+      "epoch": 0.05,
+      "learning_rate": 1.08e-05,
+      "loss": 4.3782,
       "step": 58
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.351,
+      "epoch": 0.05,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 4.2534,
       "step": 59
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.16e-05,
-      "loss": 4.2072,
+      "epoch": 0.05,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 4.4094,
       "step": 60
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.18e-05,
-      "loss": 4.2895,
+      "epoch": 0.05,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.2115,
       "step": 61
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.2e-05,
-      "loss": 4.2445,
+      "epoch": 0.05,
+      "learning_rate": 1.16e-05,
+      "loss": 4.3804,
       "step": 62
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.22e-05,
-      "loss": 4.2112,
+      "epoch": 0.05,
+      "learning_rate": 1.18e-05,
+      "loss": 4.2626,
       "step": 63
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.24e-05,
-      "loss": 4.3095,
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 4.1308,
       "step": 64
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 4.2884,
+      "epoch": 0.05,
+      "learning_rate": 1.22e-05,
+      "loss": 4.1386,
       "step": 65
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 4.154,
+      "epoch": 0.06,
+      "learning_rate": 1.24e-05,
+      "loss": 4.2841,
       "step": 66
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 4.2997,
+      "epoch": 0.06,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 4.3423,
       "step": 67
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.32e-05,
-      "loss": 4.184,
+      "epoch": 0.06,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 4.2643,
       "step": 68
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 4.2295,
+      "epoch": 0.06,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 4.2433,
       "step": 69
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 4.1964,
+      "epoch": 0.06,
+      "learning_rate": 1.32e-05,
+      "loss": 4.2846,
       "step": 70
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.3184,
+      "epoch": 0.06,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 4.4448,
       "step": 71
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 4.3462,
+      "epoch": 0.06,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 4.3386,
       "step": 72
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.42e-05,
-      "loss": 4.1737,
+      "epoch": 0.06,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.2584,
       "step": 73
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.44e-05,
-      "loss": 4.2753,
+      "epoch": 0.06,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 4.3441,
       "step": 74
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 4.2565,
+      "epoch": 0.06,
+      "learning_rate": 1.42e-05,
+      "loss": 4.1907,
       "step": 75
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.48e-05,
-      "loss": 4.2009,
+      "epoch": 0.06,
+      "learning_rate": 1.44e-05,
+      "loss": 4.5775,
       "step": 76
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.5e-05,
-      "loss": 4.2576,
+      "epoch": 0.06,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 4.3469,
       "step": 77
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.52e-05,
-      "loss": 4.2042,
+      "epoch": 0.07,
+      "learning_rate": 1.48e-05,
+      "loss": 4.2015,
       "step": 78
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.54e-05,
-      "loss": 4.1836,
+      "epoch": 0.07,
+      "learning_rate": 1.5e-05,
+      "loss": 4.201,
       "step": 79
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.56e-05,
-      "loss": 4.3454,
+      "epoch": 0.07,
+      "learning_rate": 1.52e-05,
+      "loss": 4.3279,
       "step": 80
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.58e-05,
-      "loss": 4.2717,
+      "epoch": 0.07,
+      "learning_rate": 1.54e-05,
+      "loss": 4.3433,
       "step": 81
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 4.2216,
+      "epoch": 0.07,
+      "learning_rate": 1.56e-05,
+      "loss": 4.2784,
       "step": 82
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.62e-05,
-      "loss": 4.3399,
+      "epoch": 0.07,
+      "learning_rate": 1.58e-05,
+      "loss": 4.1945,
       "step": 83
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 4.2078,
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 4.3453,
       "step": 84
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.66e-05,
-      "loss": 4.2838,
+      "epoch": 0.07,
+      "learning_rate": 1.62e-05,
+      "loss": 4.2331,
       "step": 85
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.2483,
+      "epoch": 0.07,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 4.2471,
       "step": 86
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 4.328,
+      "epoch": 0.07,
+      "learning_rate": 1.66e-05,
+      "loss": 4.1374,
       "step": 87
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 4.2141,
+      "epoch": 0.07,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.3383,
       "step": 88
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.74e-05,
-      "loss": 4.3494,
+      "epoch": 0.07,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 4.3956,
       "step": 89
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.76e-05,
-      "loss": 4.335,
+      "epoch": 0.08,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 4.302,
       "step": 90
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.78e-05,
-      "loss": 4.2173,
+      "epoch": 0.08,
+      "learning_rate": 1.74e-05,
+      "loss": 4.3601,
       "step": 91
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8e-05,
-      "loss": 4.4714,
+      "epoch": 0.08,
+      "learning_rate": 1.76e-05,
+      "loss": 4.4232,
       "step": 92
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 4.3728,
+      "epoch": 0.08,
+      "learning_rate": 1.78e-05,
+      "loss": 4.0201,
       "step": 93
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.84e-05,
-      "loss": 4.3668,
+      "epoch": 0.08,
+      "learning_rate": 1.8e-05,
+      "loss": 4.406,
       "step": 94
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.86e-05,
-      "loss": 4.2791,
+      "epoch": 0.08,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 4.5938,
       "step": 95
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.88e-05,
-      "loss": 4.5144,
+      "epoch": 0.08,
+      "learning_rate": 1.84e-05,
+      "loss": 4.4084,
       "step": 96
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9e-05,
-      "loss": 4.4262,
+      "epoch": 0.08,
+      "learning_rate": 1.86e-05,
+      "loss": 4.3834,
       "step": 97
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.2953,
+      "epoch": 0.08,
+      "learning_rate": 1.88e-05,
+      "loss": 4.5423,
       "step": 98
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.94e-05,
-      "loss": 4.6353,
+      "epoch": 0.08,
+      "learning_rate": 1.9e-05,
+      "loss": 4.5836,
       "step": 99
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 4.4004,
+      "epoch": 0.08,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.2831,
       "step": 100
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 4.2493,
+      "epoch": 0.08,
+      "learning_rate": 1.94e-05,
+      "loss": 4.035,
       "step": 101
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2e-05,
-      "loss": 4.1879,
+      "epoch": 0.09,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 4.3696,
       "step": 102
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 4.0631,
+      "epoch": 0.09,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 4.2102,
       "step": 103
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2.04e-05,
-      "loss": 4.1411,
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 4.2823,
       "step": 104
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.06e-05,
-      "loss": 4.0814,
+      "epoch": 0.09,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 4.2054,
       "step": 105
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.08e-05,
-      "loss": 4.1362,
+      "epoch": 0.09,
+      "learning_rate": 2.04e-05,
+      "loss": 4.3419,
       "step": 106
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.1e-05,
-      "loss": 4.2139,
+      "epoch": 0.09,
+      "learning_rate": 2.06e-05,
+      "loss": 4.4399,
       "step": 107
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.12e-05,
-      "loss": 4.1156,
+      "epoch": 0.09,
+      "learning_rate": 2.08e-05,
+      "loss": 4.3932,
       "step": 108
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 4.1333,
+      "epoch": 0.09,
+      "learning_rate": 2.1e-05,
+      "loss": 4.3296,
       "step": 109
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.16e-05,
-      "loss": 4.1727,
+      "epoch": 0.09,
+      "learning_rate": 2.12e-05,
+      "loss": 4.0718,
       "step": 110
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.18e-05,
-      "loss": 4.2007,
+      "epoch": 0.09,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 4.3539,
       "step": 111
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 4.1927,
+      "epoch": 0.09,
+      "learning_rate": 2.16e-05,
+      "loss": 4.4107,
       "step": 112
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.22e-05,
-      "loss": 4.1808,
+      "epoch": 0.09,
+      "learning_rate": 2.18e-05,
+      "loss": 4.2035,
       "step": 113
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 4.1332,
+      "epoch": 0.1,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.3587,
       "step": 114
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.26e-05,
-      "loss": 4.1873,
+      "epoch": 0.1,
+      "learning_rate": 2.22e-05,
+      "loss": 4.296,
       "step": 115
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.2898,
+      "epoch": 0.1,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 4.4333,
       "step": 116
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 4.1585,
+      "epoch": 0.1,
+      "learning_rate": 2.26e-05,
+      "loss": 4.1893,
       "step": 117
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.32e-05,
-      "loss": 4.3155,
+      "epoch": 0.1,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.0561,
       "step": 118
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.216,
+      "epoch": 0.1,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.1667,
       "step": 119
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.36e-05,
-      "loss": 4.239,
+      "epoch": 0.1,
+      "learning_rate": 2.32e-05,
+      "loss": 4.2186,
       "step": 120
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.38e-05,
-      "loss": 4.0644,
+      "epoch": 0.1,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.4999,
       "step": 121
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.4e-05,
-      "loss": 4.1797,
+      "epoch": 0.1,
+      "learning_rate": 2.36e-05,
+      "loss": 4.214,
       "step": 122
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 4.1803,
+      "epoch": 0.1,
+      "learning_rate": 2.38e-05,
+      "loss": 4.32,
       "step": 123
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.44e-05,
-      "loss": 4.1742,
+      "epoch": 0.1,
+      "learning_rate": 2.4e-05,
+      "loss": 4.3299,
       "step": 124
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.46e-05,
-      "loss": 4.2787,
+      "epoch": 0.11,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 4.0869,
       "step": 125
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.48e-05,
-      "loss": 4.2894,
+      "epoch": 0.11,
+      "learning_rate": 2.44e-05,
+      "loss": 4.2637,
       "step": 126
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.5e-05,
-      "loss": 4.2008,
+      "epoch": 0.11,
+      "learning_rate": 2.46e-05,
+      "loss": 4.4578,
       "step": 127
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 4.1734,
+      "epoch": 0.11,
+      "learning_rate": 2.48e-05,
+      "loss": 4.1107,
       "step": 128
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.54e-05,
-      "loss": 4.0823,
+      "epoch": 0.11,
+      "learning_rate": 2.5e-05,
+      "loss": 4.0361,
       "step": 129
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 4.2369,
+      "epoch": 0.11,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 4.3053,
       "step": 130
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.58e-05,
-      "loss": 4.1488,
+      "epoch": 0.11,
+      "learning_rate": 2.54e-05,
+      "loss": 4.276,
       "step": 131
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 4.2138,
+      "epoch": 0.11,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 4.2046,
       "step": 132
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 4.1217,
+      "epoch": 0.11,
+      "learning_rate": 2.58e-05,
+      "loss": 4.4279,
       "step": 133
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.64e-05,
-      "loss": 4.1846,
+      "epoch": 0.11,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.2977,
       "step": 134
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 4.3267,
+      "epoch": 0.11,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 4.4145,
       "step": 135
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 4.184,
+      "epoch": 0.11,
+      "learning_rate": 2.64e-05,
+      "loss": 4.1056,
       "step": 136
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 4.263,
+      "epoch": 0.12,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 4.3979,
       "step": 137
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 4.3025,
+      "epoch": 0.12,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 4.3491,
       "step": 138
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 4.307,
+      "epoch": 0.12,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.1367,
       "step": 139
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 4.2831,
+      "epoch": 0.12,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 4.3595,
       "step": 140
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 4.3902,
+      "epoch": 0.12,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 4.3963,
       "step": 141
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 4.3271,
+      "epoch": 0.12,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.2004,
       "step": 142
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.2895,
+      "epoch": 0.12,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 4.1288,
       "step": 143
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.84e-05,
-      "loss": 4.3228,
+      "epoch": 0.12,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.4963,
       "step": 144
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.86e-05,
-      "loss": 4.2486,
+      "epoch": 0.12,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.2566,
       "step": 145
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.88e-05,
-      "loss": 4.3242,
+      "epoch": 0.12,
+      "learning_rate": 2.84e-05,
+      "loss": 4.5506,
       "step": 146
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.9e-05,
-      "loss": 4.595,
+      "epoch": 0.12,
+      "learning_rate": 2.86e-05,
+      "loss": 4.2999,
       "step": 147
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 4.6988,
+      "epoch": 0.12,
+      "learning_rate": 2.88e-05,
+      "loss": 4.1624,
       "step": 148
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.94e-05,
-      "loss": 4.4544,
+      "epoch": 0.13,
+      "learning_rate": 2.9e-05,
+      "loss": 4.417,
       "step": 149
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.96e-05,
-      "loss": 4.5323,
+      "epoch": 0.13,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 4.1794,
       "step": 150
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.98e-05,
-      "loss": 4.1013,
+      "epoch": 0.13,
+      "learning_rate": 2.94e-05,
+      "loss": 4.2763,
       "step": 151
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 3e-05,
-      "loss": 4.1979,
+      "epoch": 0.13,
+      "learning_rate": 2.96e-05,
+      "loss": 4.1685,
       "step": 152
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 3.02e-05,
-      "loss": 4.2243,
+      "epoch": 0.13,
+      "learning_rate": 2.98e-05,
+      "loss": 4.2241,
       "step": 153
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.04e-05,
-      "loss": 4.0781,
+      "epoch": 0.13,
+      "learning_rate": 3e-05,
+      "loss": 4.3586,
       "step": 154
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.06e-05,
-      "loss": 4.213,
+      "epoch": 0.13,
+      "learning_rate": 3.02e-05,
+      "loss": 4.2351,
       "step": 155
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.08e-05,
-      "loss": 4.0605,
+      "epoch": 0.13,
+      "learning_rate": 3.04e-05,
+      "loss": 4.0828,
       "step": 156
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.1e-05,
-      "loss": 4.0824,
+      "epoch": 0.13,
+      "learning_rate": 3.06e-05,
+      "loss": 4.2497,
       "step": 157
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.12e-05,
-      "loss": 4.1807,
+      "epoch": 0.13,
+      "learning_rate": 3.08e-05,
+      "loss": 4.4467,
       "step": 158
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 4.1912,
+      "epoch": 0.13,
+      "learning_rate": 3.1e-05,
+      "loss": 4.1817,
       "step": 159
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.16e-05,
-      "loss": 4.2013,
+      "epoch": 0.13,
+      "learning_rate": 3.12e-05,
+      "loss": 3.9492,
       "step": 160
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.18e-05,
-      "loss": 4.2653,
+      "epoch": 0.14,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 4.284,
       "step": 161
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 4.2088,
+      "epoch": 0.14,
+      "learning_rate": 3.16e-05,
+      "loss": 4.2965,
       "step": 162
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 4.1604,
+      "epoch": 0.14,
+      "learning_rate": 3.18e-05,
+      "loss": 4.2367,
       "step": 163
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.24e-05,
-      "loss": 4.218,
+      "epoch": 0.14,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.0118,
       "step": 164
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.26e-05,
-      "loss": 4.0929,
+      "epoch": 0.14,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 4.2444,
       "step": 165
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 4.1595,
+      "epoch": 0.14,
+      "learning_rate": 3.24e-05,
+      "loss": 3.8924,
       "step": 166
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.3e-05,
-      "loss": 4.0662,
+      "epoch": 0.14,
+      "learning_rate": 3.26e-05,
+      "loss": 4.2508,
       "step": 167
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.32e-05,
-      "loss": 4.1203,
+      "epoch": 0.14,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 4.1222,
       "step": 168
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 4.2015,
+      "epoch": 0.14,
+      "learning_rate": 3.3e-05,
+      "loss": 4.3227,
       "step": 169
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 4.2155,
+      "epoch": 0.14,
+      "learning_rate": 3.32e-05,
+      "loss": 4.1995,
       "step": 170
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.38e-05,
-      "loss": 4.2056,
+      "epoch": 0.14,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 4.2672,
       "step": 171
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.2715,
+      "epoch": 0.14,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 4.1165,
       "step": 172
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 4.0736,
+      "epoch": 0.15,
+      "learning_rate": 3.38e-05,
+      "loss": 4.0226,
       "step": 173
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 4.2239,
+      "epoch": 0.15,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.3942,
       "step": 174
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.46e-05,
-      "loss": 4.2676,
+      "epoch": 0.15,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 4.2474,
       "step": 175
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.48e-05,
-      "loss": 4.1391,
+      "epoch": 0.15,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 4.2374,
       "step": 176
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.5e-05,
-      "loss": 4.0946,
+      "epoch": 0.15,
+      "learning_rate": 3.46e-05,
+      "loss": 4.3093,
       "step": 177
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.52e-05,
-      "loss": 4.2066,
+      "epoch": 0.15,
+      "learning_rate": 3.48e-05,
+      "loss": 4.1673,
       "step": 178
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.54e-05,
-      "loss": 4.2802,
+      "epoch": 0.15,
+      "learning_rate": 3.5e-05,
+      "loss": 4.2325,
       "step": 179
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.56e-05,
-      "loss": 4.2446,
+      "epoch": 0.15,
+      "learning_rate": 3.52e-05,
+      "loss": 4.344,
       "step": 180
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.58e-05,
-      "loss": 4.1873,
+      "epoch": 0.15,
+      "learning_rate": 3.54e-05,
+      "loss": 4.3441,
       "step": 181
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.6e-05,
-      "loss": 4.2523,
+      "epoch": 0.15,
+      "learning_rate": 3.56e-05,
+      "loss": 4.3454,
       "step": 182
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.62e-05,
-      "loss": 4.2105,
+      "epoch": 0.15,
+      "learning_rate": 3.58e-05,
+      "loss": 4.1881,
       "step": 183
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 4.2264,
+      "epoch": 0.15,
+      "learning_rate": 3.6e-05,
+      "loss": 4.2848,
       "step": 184
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.66e-05,
-      "loss": 4.208,
+      "epoch": 0.16,
+      "learning_rate": 3.62e-05,
+      "loss": 4.0505,
       "step": 185
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.68e-05,
-      "loss": 4.0727,
+      "epoch": 0.16,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 4.2045,
       "step": 186
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.7e-05,
-      "loss": 4.0995,
+      "epoch": 0.16,
+      "learning_rate": 3.66e-05,
+      "loss": 4.4636,
       "step": 187
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.72e-05,
-      "loss": 4.0693,
+      "epoch": 0.16,
+      "learning_rate": 3.68e-05,
+      "loss": 4.2481,
       "step": 188
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.74e-05,
-      "loss": 4.1961,
+      "epoch": 0.16,
+      "learning_rate": 3.7e-05,
+      "loss": 4.4013,
       "step": 189
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.76e-05,
-      "loss": 4.3002,
+      "epoch": 0.16,
+      "learning_rate": 3.72e-05,
+      "loss": 4.0479,
       "step": 190
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 4.2077,
+      "epoch": 0.16,
+      "learning_rate": 3.74e-05,
+      "loss": 4.355,
       "step": 191
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.8e-05,
-      "loss": 4.1656,
+      "epoch": 0.16,
+      "learning_rate": 3.76e-05,
+      "loss": 4.549,
       "step": 192
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.82e-05,
-      "loss": 4.1497,
+      "epoch": 0.16,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 4.4815,
       "step": 193
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 4.28,
+      "epoch": 0.16,
+      "learning_rate": 3.8e-05,
+      "loss": 4.2467,
       "step": 194
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.86e-05,
-      "loss": 4.4085,
+      "epoch": 0.16,
+      "learning_rate": 3.82e-05,
+      "loss": 4.4543,
       "step": 195
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.88e-05,
-      "loss": 4.3136,
+      "epoch": 0.16,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 4.3041,
       "step": 196
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 4.4431,
+      "epoch": 0.17,
+      "learning_rate": 3.86e-05,
+      "loss": 4.3881,
       "step": 197
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 4.5063,
+      "epoch": 0.17,
+      "learning_rate": 3.88e-05,
+      "loss": 4.5193,
       "step": 198
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.94e-05,
-      "loss": 4.4454,
+      "epoch": 0.17,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 4.5835,
       "step": 199
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 4.436,
+      "epoch": 0.17,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 4.3379,
       "step": 200
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 4.2278,
+      "epoch": 0.17,
+      "learning_rate": 3.94e-05,
+      "loss": 4.3094,
       "step": 201
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 4e-05,
-      "loss": 4.1701,
+      "epoch": 0.17,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 4.2136,
       "step": 202
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.02e-05,
-      "loss": 4.1987,
+      "epoch": 0.17,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 4.3287,
       "step": 203
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 4.1717,
+      "epoch": 0.17,
+      "learning_rate": 4e-05,
+      "loss": 4.1841,
       "step": 204
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 4.2467,
+      "epoch": 0.17,
+      "learning_rate": 4.02e-05,
+      "loss": 4.1372,
       "step": 205
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.08e-05,
-      "loss": 4.2045,
+      "epoch": 0.17,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 4.2192,
       "step": 206
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.1e-05,
-      "loss": 4.0783,
+      "epoch": 0.17,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 4.2651,
       "step": 207
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.12e-05,
-      "loss": 4.1963,
+      "epoch": 0.17,
+      "learning_rate": 4.08e-05,
+      "loss": 4.5047,
       "step": 208
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.14e-05,
-      "loss": 4.2476,
+      "epoch": 0.18,
+      "learning_rate": 4.1e-05,
+      "loss": 4.3023,
       "step": 209
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.16e-05,
-      "loss": 4.197,
+      "epoch": 0.18,
+      "learning_rate": 4.12e-05,
+      "loss": 4.4317,
       "step": 210
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.18e-05,
-      "loss": 4.0622,
+      "epoch": 0.18,
+      "learning_rate": 4.14e-05,
+      "loss": 4.2699,
       "step": 211
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.2e-05,
-      "loss": 4.1235,
+      "epoch": 0.18,
+      "learning_rate": 4.16e-05,
+      "loss": 4.3623,
       "step": 212
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.22e-05,
-      "loss": 4.0617,
+      "epoch": 0.18,
+      "learning_rate": 4.18e-05,
+      "loss": 4.1939,
       "step": 213
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.24e-05,
-      "loss": 4.119,
+      "epoch": 0.18,
+      "learning_rate": 4.2e-05,
+      "loss": 4.258,
       "step": 214
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.26e-05,
-      "loss": 4.1679,
+      "epoch": 0.18,
+      "learning_rate": 4.22e-05,
+      "loss": 4.337,
       "step": 215
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 4.1364,
+      "epoch": 0.18,
+      "learning_rate": 4.24e-05,
+      "loss": 4.2375,
       "step": 216
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.3e-05,
-      "loss": 4.1803,
+      "epoch": 0.18,
+      "learning_rate": 4.26e-05,
+      "loss": 4.3783,
       "step": 217
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.32e-05,
-      "loss": 4.132,
+      "epoch": 0.18,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 4.248,
       "step": 218
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 4.0901,
+      "epoch": 0.18,
+      "learning_rate": 4.3e-05,
+      "loss": 3.9674,
       "step": 219
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.36e-05,
-      "loss": 4.1789,
+      "epoch": 0.18,
+      "learning_rate": 4.32e-05,
+      "loss": 4.2933,
       "step": 220
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.38e-05,
-      "loss": 4.1393,
+      "epoch": 0.19,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 4.1908,
       "step": 221
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 4.1346,
+      "epoch": 0.19,
+      "learning_rate": 4.36e-05,
+      "loss": 4.5806,
       "step": 222
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 4.1677,
+      "epoch": 0.19,
+      "learning_rate": 4.38e-05,
+      "loss": 4.3698,
       "step": 223
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.44e-05,
-      "loss": 4.1156,
+      "epoch": 0.19,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 4.3493,
       "step": 224
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.46e-05,
-      "loss": 4.27,
+      "epoch": 0.19,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 4.1244,
       "step": 225
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 4.159,
+      "epoch": 0.19,
+      "learning_rate": 4.44e-05,
+      "loss": 4.0751,
       "step": 226
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.5e-05,
-      "loss": 4.2819,
+      "epoch": 0.19,
+      "learning_rate": 4.46e-05,
+      "loss": 4.2677,
       "step": 227
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.52e-05,
-      "loss": 4.2656,
+      "epoch": 0.19,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 4.2542,
       "step": 228
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 4.1013,
+      "epoch": 0.19,
+      "learning_rate": 4.5e-05,
+      "loss": 4.1859,
       "step": 229
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 4.2321,
+      "epoch": 0.19,
+      "learning_rate": 4.52e-05,
+      "loss": 4.3379,
       "step": 230
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.58e-05,
-      "loss": 4.2513,
+      "epoch": 0.19,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 4.3722,
       "step": 231
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 4.0487,
+      "epoch": 0.2,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 4.236,
       "step": 232
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 4.1196,
+      "epoch": 0.2,
+      "learning_rate": 4.58e-05,
+      "loss": 4.2567,
       "step": 233
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.64e-05,
-      "loss": 4.1884,
+      "epoch": 0.2,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.1549,
       "step": 234
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 4.2405,
+      "epoch": 0.2,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 4.3044,
       "step": 235
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 4.1805,
+      "epoch": 0.2,
+      "learning_rate": 4.64e-05,
+      "loss": 4.4134,
       "step": 236
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.7e-05,
-      "loss": 4.1694,
+      "epoch": 0.2,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 4.137,
       "step": 237
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.72e-05,
-      "loss": 4.1361,
+      "epoch": 0.2,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 4.2326,
       "step": 238
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.74e-05,
-      "loss": 4.0898,
+      "epoch": 0.2,
+      "learning_rate": 4.7e-05,
+      "loss": 4.2931,
       "step": 239
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.76e-05,
-      "loss": 4.1756,
+      "epoch": 0.2,
+      "learning_rate": 4.72e-05,
+      "loss": 4.2976,
       "step": 240
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.78e-05,
-      "loss": 4.1759,
+      "epoch": 0.2,
+      "learning_rate": 4.74e-05,
+      "loss": 4.4488,
       "step": 241
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.8e-05,
-      "loss": 4.2335,
+      "epoch": 0.2,
+      "learning_rate": 4.76e-05,
+      "loss": 4.1721,
       "step": 242
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.82e-05,
-      "loss": 4.4165,
+      "epoch": 0.2,
+      "learning_rate": 4.78e-05,
+      "loss": 4.4607,
       "step": 243
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 4.2451,
+      "epoch": 0.21,
+      "learning_rate": 4.8e-05,
+      "loss": 4.3101,
       "step": 244
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.86e-05,
-      "loss": 4.2268,
+      "epoch": 0.21,
+      "learning_rate": 4.82e-05,
+      "loss": 4.4948,
       "step": 245
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.88e-05,
-      "loss": 4.0973,
-      "step": 246
+      "epoch": 0.21,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 4.1713,
+      "step": 246
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.9e-05,
-      "loss": 4.0274,
+      "epoch": 0.21,
+      "learning_rate": 4.86e-05,
+      "loss": 4.9295,
       "step": 247
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.92e-05,
-      "loss": 4.5398,
+      "epoch": 0.21,
+      "learning_rate": 4.88e-05,
+      "loss": 4.4184,
       "step": 248
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.94e-05,
-      "loss": 4.4893,
+      "epoch": 0.21,
+      "learning_rate": 4.9e-05,
+      "loss": 4.5421,
       "step": 249
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.96e-05,
-      "loss": 4.4634,
+      "epoch": 0.21,
+      "learning_rate": 4.92e-05,
+      "loss": 4.4627,
       "step": 250
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 4.2059,
+      "epoch": 0.21,
+      "learning_rate": 4.94e-05,
+      "loss": 4.3955,
       "step": 251
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5e-05,
-      "loss": 4.2231,
+      "epoch": 0.21,
+      "learning_rate": 4.96e-05,
+      "loss": 4.0872,
       "step": 252
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.02e-05,
-      "loss": 4.2464,
+      "epoch": 0.21,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 4.4067,
       "step": 253
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 4.0991,
+      "epoch": 0.21,
+      "learning_rate": 5e-05,
+      "loss": 4.0762,
       "step": 254
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 4.1931,
+      "epoch": 0.21,
+      "learning_rate": 5.02e-05,
+      "loss": 4.1029,
       "step": 255
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.08e-05,
-      "loss": 4.1384,
+      "epoch": 0.22,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 4.2593,
       "step": 256
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 4.1315,
+      "epoch": 0.22,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 4.1289,
       "step": 257
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 4.1089,
+      "epoch": 0.22,
+      "learning_rate": 5.08e-05,
+      "loss": 4.5927,
       "step": 258
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.14e-05,
-      "loss": 4.1466,
+      "epoch": 0.22,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.3014,
       "step": 259
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.16e-05,
-      "loss": 4.1282,
+      "epoch": 0.22,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 4.202,
       "step": 260
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 4.213,
+      "epoch": 0.22,
+      "learning_rate": 5.14e-05,
+      "loss": 4.2538,
       "step": 261
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 4.1021,
+      "epoch": 0.22,
+      "learning_rate": 5.16e-05,
+      "loss": 4.2557,
       "step": 262
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.22e-05,
-      "loss": 4.2691,
+      "epoch": 0.22,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 4.4381,
       "step": 263
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 4.1385,
+      "epoch": 0.22,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 4.3642,
       "step": 264
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 4.1956,
+      "epoch": 0.22,
+      "learning_rate": 5.22e-05,
+      "loss": 4.2546,
       "step": 265
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.28e-05,
-      "loss": 4.1191,
+      "epoch": 0.22,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 4.38,
       "step": 266
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 4.1614,
+      "epoch": 0.22,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 4.1872,
       "step": 267
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 4.0642,
+      "epoch": 0.23,
+      "learning_rate": 5.28e-05,
+      "loss": 4.373,
       "step": 268
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.3400000000000004e-05,
-      "loss": 4.2205,
+      "epoch": 0.23,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 4.2422,
       "step": 269
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.360000000000001e-05,
-      "loss": 4.0858,
+      "epoch": 0.23,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 4.1554,
       "step": 270
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.380000000000001e-05,
-      "loss": 4.0458,
+      "epoch": 0.23,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 4.4783,
       "step": 271
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 4.2021,
+      "epoch": 0.23,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 4.1543,
       "step": 272
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.420000000000001e-05,
-      "loss": 4.0632,
+      "epoch": 0.23,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 4.2668,
       "step": 273
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.440000000000001e-05,
-      "loss": 4.1333,
+      "epoch": 0.23,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 4.2967,
       "step": 274
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.4600000000000006e-05,
-      "loss": 4.2086,
+      "epoch": 0.23,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 4.2984,
       "step": 275
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.4800000000000004e-05,
-      "loss": 4.1195,
+      "epoch": 0.23,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 4.0842,
       "step": 276
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 4.08,
+      "epoch": 0.23,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 4.1497,
       "step": 277
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.520000000000001e-05,
-      "loss": 4.1521,
+      "epoch": 0.23,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 4.2626,
       "step": 278
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.5400000000000005e-05,
-      "loss": 4.1239,
+      "epoch": 0.23,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.3448,
       "step": 279
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.560000000000001e-05,
-      "loss": 4.2814,
+      "epoch": 0.24,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 4.2987,
       "step": 280
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.580000000000001e-05,
-      "loss": 4.1766,
+      "epoch": 0.24,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 4.2392,
       "step": 281
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 4.0358,
+      "epoch": 0.24,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 4.3573,
       "step": 282
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.620000000000001e-05,
-      "loss": 4.0526,
+      "epoch": 0.24,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 4.4735,
       "step": 283
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.1761,
+      "epoch": 0.24,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 4.1439,
       "step": 284
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.66e-05,
-      "loss": 4.2562,
+      "epoch": 0.24,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 4.3813,
       "step": 285
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.68e-05,
-      "loss": 4.1342,
+      "epoch": 0.24,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.1915,
       "step": 286
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.2481,
+      "epoch": 0.24,
+      "learning_rate": 5.66e-05,
+      "loss": 4.2068,
       "step": 287
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.72e-05,
-      "loss": 4.1786,
+      "epoch": 0.24,
+      "learning_rate": 5.68e-05,
+      "loss": 4.2236,
       "step": 288
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.74e-05,
-      "loss": 4.0954,
+      "epoch": 0.24,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.4435,
       "step": 289
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.76e-05,
-      "loss": 4.1567,
+      "epoch": 0.24,
+      "learning_rate": 5.72e-05,
+      "loss": 4.324,
       "step": 290
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.7799999999999995e-05,
-      "loss": 4.44,
+      "epoch": 0.24,
+      "learning_rate": 5.74e-05,
+      "loss": 4.3704,
       "step": 291
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.8e-05,
-      "loss": 4.2732,
+      "epoch": 0.25,
+      "learning_rate": 5.76e-05,
+      "loss": 4.6098,
       "step": 292
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.82e-05,
-      "loss": 4.3449,
+      "epoch": 0.25,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 4.1998,
       "step": 293
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.8399999999999997e-05,
-      "loss": 4.3422,
+      "epoch": 0.25,
+      "learning_rate": 5.8e-05,
+      "loss": 4.5113,
       "step": 294
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.86e-05,
-      "loss": 4.2886,
+      "epoch": 0.25,
+      "learning_rate": 5.82e-05,
+      "loss": 4.4031,
       "step": 295
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.88e-05,
-      "loss": 4.3468,
+      "epoch": 0.25,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 5.0759,
       "step": 296
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.9e-05,
-      "loss": 4.1907,
+      "epoch": 0.25,
+      "learning_rate": 5.86e-05,
+      "loss": 4.3472,
       "step": 297
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.92e-05,
-      "loss": 4.5633,
+      "epoch": 0.25,
+      "learning_rate": 5.88e-05,
+      "loss": 4.8672,
       "step": 298
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.94e-05,
-      "loss": 4.2836,
+      "epoch": 0.25,
+      "learning_rate": 5.9e-05,
+      "loss": 5.1552,
       "step": 299
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.96e-05,
-      "loss": 4.6411,
+      "epoch": 0.25,
+      "learning_rate": 5.92e-05,
+      "loss": 4.9012,
       "step": 300
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.9800000000000003e-05,
-      "loss": 4.2357,
+      "epoch": 0.25,
+      "learning_rate": 5.94e-05,
+      "loss": 4.1857,
       "step": 301
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6e-05,
-      "loss": 4.2123,
+      "epoch": 0.25,
+      "learning_rate": 5.96e-05,
+      "loss": 4.2613,
       "step": 302
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6.02e-05,
-      "loss": 4.1736,
+      "epoch": 0.25,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 4.3109,
       "step": 303
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6.04e-05,
-      "loss": 4.199,
+      "epoch": 0.26,
+      "learning_rate": 6e-05,
+      "loss": 4.0199,
       "step": 304
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6.06e-05,
-      "loss": 4.1657,
+      "epoch": 0.26,
+      "learning_rate": 6.02e-05,
+      "loss": 4.1863,
       "step": 305
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.08e-05,
-      "loss": 4.1201,
+      "epoch": 0.26,
+      "learning_rate": 6.04e-05,
+      "loss": 4.2041,
       "step": 306
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.1e-05,
-      "loss": 4.1695,
+      "epoch": 0.26,
+      "learning_rate": 6.06e-05,
+      "loss": 4.3666,
       "step": 307
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.12e-05,
-      "loss": 4.1435,
+      "epoch": 0.26,
+      "learning_rate": 6.08e-05,
+      "loss": 4.0859,
       "step": 308
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.14e-05,
-      "loss": 4.1498,
+      "epoch": 0.26,
+      "learning_rate": 6.1e-05,
+      "loss": 4.04,
       "step": 309
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.16e-05,
-      "loss": 4.2333,
+      "epoch": 0.26,
+      "learning_rate": 6.12e-05,
+      "loss": 4.2237,
       "step": 310
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.18e-05,
-      "loss": 4.0113,
+      "epoch": 0.26,
+      "learning_rate": 6.14e-05,
+      "loss": 4.5948,
       "step": 311
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.2e-05,
-      "loss": 4.2104,
+      "epoch": 0.26,
+      "learning_rate": 6.16e-05,
+      "loss": 4.1543,
       "step": 312
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.220000000000001e-05,
-      "loss": 4.3056,
+      "epoch": 0.26,
+      "learning_rate": 6.18e-05,
+      "loss": 4.0795,
       "step": 313
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.24e-05,
-      "loss": 4.1282,
+      "epoch": 0.26,
+      "learning_rate": 6.2e-05,
+      "loss": 4.2446,
       "step": 314
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.26e-05,
-      "loss": 4.1415,
+      "epoch": 0.26,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 4.0431,
       "step": 315
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.280000000000001e-05,
-      "loss": 4.1358,
+      "epoch": 0.27,
+      "learning_rate": 6.24e-05,
+      "loss": 4.4321,
       "step": 316
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.3e-05,
-      "loss": 4.1366,
+      "epoch": 0.27,
+      "learning_rate": 6.26e-05,
+      "loss": 4.4274,
       "step": 317
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.32e-05,
-      "loss": 4.1256,
+      "epoch": 0.27,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 4.4757,
       "step": 318
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.340000000000001e-05,
-      "loss": 4.1378,
+      "epoch": 0.27,
+      "learning_rate": 6.3e-05,
+      "loss": 4.2332,
       "step": 319
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.36e-05,
-      "loss": 4.2278,
+      "epoch": 0.27,
+      "learning_rate": 6.32e-05,
+      "loss": 4.3185,
       "step": 320
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.38e-05,
-      "loss": 4.1062,
+      "epoch": 0.27,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 4.361,
       "step": 321
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 3.9692,
+      "epoch": 0.27,
+      "learning_rate": 6.36e-05,
+      "loss": 4.1371,
       "step": 322
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.42e-05,
-      "loss": 4.1761,
+      "epoch": 0.27,
+      "learning_rate": 6.38e-05,
+      "loss": 4.151,
       "step": 323
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.440000000000001e-05,
-      "loss": 4.0297,
+      "epoch": 0.27,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 4.3845,
       "step": 324
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.460000000000001e-05,
-      "loss": 4.196,
+      "epoch": 0.27,
+      "learning_rate": 6.42e-05,
+      "loss": 4.2488,
       "step": 325
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.48e-05,
-      "loss": 4.1788,
+      "epoch": 0.27,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 4.5731,
       "step": 326
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 4.0838,
+      "epoch": 0.27,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 4.6346,
       "step": 327
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.52e-05,
-      "loss": 4.0498,
+      "epoch": 0.28,
+      "learning_rate": 6.48e-05,
+      "loss": 4.3665,
       "step": 328
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.54e-05,
-      "loss": 4.0674,
+      "epoch": 0.28,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.1868,
       "step": 329
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.560000000000001e-05,
-      "loss": 4.2265,
+      "epoch": 0.28,
+      "learning_rate": 6.52e-05,
+      "loss": 4.4949,
       "step": 330
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.58e-05,
-      "loss": 4.2266,
+      "epoch": 0.28,
+      "learning_rate": 6.54e-05,
+      "loss": 4.1534,
       "step": 331
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.6e-05,
-      "loss": 4.0707,
+      "epoch": 0.28,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 4.4734,
       "step": 332
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.620000000000001e-05,
-      "loss": 4.2441,
+      "epoch": 0.28,
+      "learning_rate": 6.58e-05,
+      "loss": 4.312,
       "step": 333
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.64e-05,
-      "loss": 4.2306,
+      "epoch": 0.28,
+      "learning_rate": 6.6e-05,
+      "loss": 4.0929,
       "step": 334
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.66e-05,
-      "loss": 4.3024,
+      "epoch": 0.28,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 4.4961,
       "step": 335
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.680000000000001e-05,
-      "loss": 4.1858,
+      "epoch": 0.28,
+      "learning_rate": 6.64e-05,
+      "loss": 4.5378,
       "step": 336
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.7e-05,
-      "loss": 4.1005,
+      "epoch": 0.28,
+      "learning_rate": 6.66e-05,
+      "loss": 4.2957,
       "step": 337
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.720000000000001e-05,
-      "loss": 4.2044,
+      "epoch": 0.28,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 4.4982,
       "step": 338
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.740000000000001e-05,
-      "loss": 4.2649,
+      "epoch": 0.28,
+      "learning_rate": 6.7e-05,
+      "loss": 4.4253,
       "step": 339
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.76e-05,
-      "loss": 4.267,
+      "epoch": 0.29,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 4.3843,
       "step": 340
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.780000000000001e-05,
-      "loss": 4.1318,
+      "epoch": 0.29,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 4.4761,
       "step": 341
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 4.3481,
+      "epoch": 0.29,
+      "learning_rate": 6.76e-05,
+      "loss": 4.4225,
       "step": 342
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.82e-05,
-      "loss": 4.2143,
+      "epoch": 0.29,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 4.4168,
       "step": 343
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.840000000000001e-05,
-      "loss": 4.336,
+      "epoch": 0.29,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.8081,
       "step": 344
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.860000000000001e-05,
-      "loss": 4.0903,
+      "epoch": 0.29,
+      "learning_rate": 6.82e-05,
+      "loss": 4.3226,
       "step": 345
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.879999999999999e-05,
-      "loss": 4.2886,
+      "epoch": 0.29,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 4.2473,
       "step": 346
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.9e-05,
-      "loss": 4.3189,
+      "epoch": 0.29,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 4.7237,
       "step": 347
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.92e-05,
-      "loss": 4.2777,
+      "epoch": 0.29,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 4.8503,
       "step": 348
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.939999999999999e-05,
-      "loss": 4.3815,
+      "epoch": 0.29,
+      "learning_rate": 6.9e-05,
+      "loss": 4.8351,
       "step": 349
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.96e-05,
-      "loss": 4.2971,
+      "epoch": 0.29,
+      "learning_rate": 6.92e-05,
+      "loss": 4.4137,
       "step": 350
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 6.98e-05,
-      "loss": 4.2663,
+      "epoch": 0.3,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 4.7367,
       "step": 351
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 7e-05,
-      "loss": 4.1588,
+      "epoch": 0.3,
+      "learning_rate": 6.96e-05,
+      "loss": 4.3287,
       "step": 352
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 7.02e-05,
-      "loss": 4.0897,
+      "epoch": 0.3,
+      "learning_rate": 6.98e-05,
+      "loss": 4.4298,
       "step": 353
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 7.04e-05,
-      "loss": 4.0429,
+      "epoch": 0.3,
+      "learning_rate": 7e-05,
+      "loss": 4.3341,
       "step": 354
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.06e-05,
-      "loss": 4.1905,
+      "epoch": 0.3,
+      "learning_rate": 7.02e-05,
+      "loss": 4.2487,
       "step": 355
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.08e-05,
-      "loss": 4.2756,
+      "epoch": 0.3,
+      "learning_rate": 7.04e-05,
+      "loss": 4.4111,
       "step": 356
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.1e-05,
-      "loss": 4.2279,
+      "epoch": 0.3,
+      "learning_rate": 7.06e-05,
+      "loss": 4.3982,
       "step": 357
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.12e-05,
-      "loss": 4.2192,
+      "epoch": 0.3,
+      "learning_rate": 7.08e-05,
+      "loss": 4.0899,
       "step": 358
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.14e-05,
-      "loss": 4.0772,
+      "epoch": 0.3,
+      "learning_rate": 7.1e-05,
+      "loss": 4.3019,
       "step": 359
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.16e-05,
-      "loss": 4.162,
+      "epoch": 0.3,
+      "learning_rate": 7.12e-05,
+      "loss": 4.4083,
       "step": 360
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.18e-05,
-      "loss": 4.0911,
+      "epoch": 0.3,
+      "learning_rate": 7.14e-05,
+      "loss": 4.2763,
       "step": 361
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.2e-05,
-      "loss": 4.1129,
+      "epoch": 0.3,
+      "learning_rate": 7.16e-05,
+      "loss": 4.6319,
       "step": 362
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.22e-05,
-      "loss": 4.0596,
+      "epoch": 0.31,
+      "learning_rate": 7.18e-05,
+      "loss": 4.2293,
       "step": 363
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.24e-05,
-      "loss": 4.1996,
+      "epoch": 0.31,
+      "learning_rate": 7.2e-05,
+      "loss": 4.2944,
       "step": 364
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.26e-05,
-      "loss": 4.2779,
+      "epoch": 0.31,
+      "learning_rate": 7.22e-05,
+      "loss": 4.4939,
       "step": 365
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.280000000000001e-05,
-      "loss": 4.1437,
+      "epoch": 0.31,
+      "learning_rate": 7.24e-05,
+      "loss": 4.552,
       "step": 366
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.3e-05,
-      "loss": 4.2111,
+      "epoch": 0.31,
+      "learning_rate": 7.26e-05,
+      "loss": 4.208,
       "step": 367
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.32e-05,
-      "loss": 4.1603,
+      "epoch": 0.31,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 4.5707,
       "step": 368
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.340000000000001e-05,
-      "loss": 4.0618,
+      "epoch": 0.31,
+      "learning_rate": 7.3e-05,
+      "loss": 4.5746,
       "step": 369
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.36e-05,
-      "loss": 4.084,
+      "epoch": 0.31,
+      "learning_rate": 7.32e-05,
+      "loss": 4.4998,
       "step": 370
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.38e-05,
-      "loss": 4.0958,
+      "epoch": 0.31,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 4.2445,
       "step": 371
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.4e-05,
-      "loss": 4.1815,
+      "epoch": 0.31,
+      "learning_rate": 7.36e-05,
+      "loss": 4.2634,
       "step": 372
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.42e-05,
-      "loss": 4.161,
+      "epoch": 0.31,
+      "learning_rate": 7.38e-05,
+      "loss": 4.1849,
       "step": 373
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.44e-05,
-      "loss": 3.9419,
+      "epoch": 0.31,
+      "learning_rate": 7.4e-05,
+      "loss": 4.181,
       "step": 374
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.46e-05,
-      "loss": 4.2128,
+      "epoch": 0.32,
+      "learning_rate": 7.42e-05,
+      "loss": 4.2052,
       "step": 375
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.48e-05,
-      "loss": 4.0403,
+      "epoch": 0.32,
+      "learning_rate": 7.44e-05,
+      "loss": 4.353,
       "step": 376
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 4.1399,
+      "epoch": 0.32,
+      "learning_rate": 7.46e-05,
+      "loss": 4.2325,
       "step": 377
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.52e-05,
-      "loss": 4.2712,
+      "epoch": 0.32,
+      "learning_rate": 7.48e-05,
+      "loss": 4.2946,
       "step": 378
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.54e-05,
-      "loss": 4.141,
+      "epoch": 0.32,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.1815,
       "step": 379
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.560000000000001e-05,
-      "loss": 4.0592,
+      "epoch": 0.32,
+      "learning_rate": 7.52e-05,
+      "loss": 4.3885,
       "step": 380
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.58e-05,
-      "loss": 4.1622,
+      "epoch": 0.32,
+      "learning_rate": 7.54e-05,
+      "loss": 4.1504,
       "step": 381
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.6e-05,
-      "loss": 4.221,
+      "epoch": 0.32,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 4.29,
       "step": 382
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.620000000000001e-05,
-      "loss": 4.1675,
+      "epoch": 0.32,
+      "learning_rate": 7.58e-05,
+      "loss": 4.1227,
       "step": 383
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.64e-05,
-      "loss": 4.0643,
+      "epoch": 0.32,
+      "learning_rate": 7.6e-05,
+      "loss": 4.429,
       "step": 384
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.66e-05,
-      "loss": 4.2672,
+      "epoch": 0.32,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 4.4384,
       "step": 385
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.680000000000001e-05,
-      "loss": 4.0361,
+      "epoch": 0.32,
+      "learning_rate": 7.64e-05,
+      "loss": 4.4201,
       "step": 386
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.7e-05,
-      "loss": 4.1967,
+      "epoch": 0.33,
+      "learning_rate": 7.66e-05,
+      "loss": 4.4538,
       "step": 387
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.72e-05,
-      "loss": 4.2709,
+      "epoch": 0.33,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 4.2859,
       "step": 388
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.740000000000001e-05,
-      "loss": 4.1456,
+      "epoch": 0.33,
+      "learning_rate": 7.7e-05,
+      "loss": 4.4742,
       "step": 389
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.76e-05,
-      "loss": 4.1303,
+      "epoch": 0.33,
+      "learning_rate": 7.72e-05,
+      "loss": 4.4294,
       "step": 390
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.780000000000001e-05,
-      "loss": 4.3029,
+      "epoch": 0.33,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 4.4404,
       "step": 391
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 4.103,
+      "epoch": 0.33,
+      "learning_rate": 7.76e-05,
+      "loss": 4.6172,
       "step": 392
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.82e-05,
-      "loss": 4.0453,
+      "epoch": 0.33,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 4.3909,
       "step": 393
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.840000000000001e-05,
-      "loss": 4.2824,
+      "epoch": 0.33,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 4.5848,
       "step": 394
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.860000000000001e-05,
-      "loss": 4.376,
+      "epoch": 0.33,
+      "learning_rate": 7.82e-05,
+      "loss": 4.1815,
       "step": 395
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.88e-05,
-      "loss": 4.2418,
+      "epoch": 0.33,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 4.7384,
       "step": 396
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.900000000000001e-05,
-      "loss": 4.5705,
+      "epoch": 0.33,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 4.8338,
       "step": 397
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.920000000000001e-05,
-      "loss": 4.3234,
+      "epoch": 0.33,
+      "learning_rate": 7.88e-05,
+      "loss": 4.8058,
       "step": 398
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.94e-05,
-      "loss": 4.7077,
+      "epoch": 0.34,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 4.7478,
       "step": 399
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 7.960000000000001e-05,
-      "loss": 4.3861,
+      "epoch": 0.34,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 4.5749,
       "step": 400
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 7.98e-05,
-      "loss": 4.2475,
+      "epoch": 0.34,
+      "learning_rate": 7.94e-05,
+      "loss": 4.2179,
       "step": 401
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8e-05,
-      "loss": 4.1823,
+      "epoch": 0.34,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 4.3339,
       "step": 402
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.020000000000001e-05,
-      "loss": 4.2527,
+      "epoch": 0.34,
+      "learning_rate": 7.98e-05,
+      "loss": 4.2771,
       "step": 403
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.04e-05,
-      "loss": 4.2234,
+      "epoch": 0.34,
+      "learning_rate": 8e-05,
+      "loss": 4.3949,
       "step": 404
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.060000000000001e-05,
-      "loss": 4.0678,
+      "epoch": 0.34,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 4.2464,
       "step": 405
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.080000000000001e-05,
-      "loss": 4.2486,
+      "epoch": 0.34,
+      "learning_rate": 8.04e-05,
+      "loss": 4.324,
       "step": 406
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.1e-05,
-      "loss": 4.1372,
+      "epoch": 0.34,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 4.2594,
       "step": 407
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.120000000000001e-05,
-      "loss": 4.2349,
+      "epoch": 0.34,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 4.4186,
       "step": 408
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.14e-05,
-      "loss": 4.1296,
+      "epoch": 0.34,
+      "learning_rate": 8.1e-05,
+      "loss": 4.4327,
       "step": 409
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.16e-05,
-      "loss": 4.1051,
+      "epoch": 0.34,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 4.4526,
       "step": 410
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.18e-05,
-      "loss": 4.3548,
+      "epoch": 0.35,
+      "learning_rate": 8.14e-05,
+      "loss": 4.5927,
       "step": 411
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.2e-05,
-      "loss": 4.1184,
+      "epoch": 0.35,
+      "learning_rate": 8.16e-05,
+      "loss": 4.2916,
       "step": 412
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.22e-05,
-      "loss": 4.0538,
+      "epoch": 0.35,
+      "learning_rate": 8.18e-05,
+      "loss": 4.4595,
       "step": 413
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.24e-05,
-      "loss": 4.2076,
+      "epoch": 0.35,
+      "learning_rate": 8.2e-05,
+      "loss": 4.1215,
       "step": 414
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.26e-05,
-      "loss": 4.3461,
+      "epoch": 0.35,
+      "learning_rate": 8.22e-05,
+      "loss": 4.3378,
       "step": 415
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.28e-05,
-      "loss": 4.0961,
+      "epoch": 0.35,
+      "learning_rate": 8.24e-05,
+      "loss": 4.4134,
       "step": 416
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.3e-05,
-      "loss": 4.2111,
+      "epoch": 0.35,
+      "learning_rate": 8.26e-05,
+      "loss": 4.0101,
       "step": 417
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.32e-05,
-      "loss": 4.0412,
+      "epoch": 0.35,
+      "learning_rate": 8.28e-05,
+      "loss": 4.2038,
       "step": 418
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.34e-05,
-      "loss": 4.2169,
+      "epoch": 0.35,
+      "learning_rate": 8.3e-05,
+      "loss": 4.7062,
       "step": 419
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.36e-05,
-      "loss": 4.0514,
+      "epoch": 0.35,
+      "learning_rate": 8.32e-05,
+      "loss": 4.2984,
       "step": 420
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.38e-05,
-      "loss": 4.3228,
+      "epoch": 0.35,
+      "learning_rate": 8.34e-05,
+      "loss": 4.5583,
       "step": 421
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 8.4e-05,
-      "loss": 4.0805,
+      "epoch": 0.35,
+      "learning_rate": 8.36e-05,
+      "loss": 4.3735,
       "step": 422
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 8.42e-05,
-      "loss": 4.1906,
+      "epoch": 0.36,
+      "learning_rate": 8.38e-05,
+      "loss": 4.3283,
       "step": 423
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 8.44e-05,
-      "loss": 4.2743,
+      "epoch": 0.36,
+      "learning_rate": 8.4e-05,
+      "loss": 4.2344,
       "step": 424
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 8.46e-05,
-      "loss": 4.2114,
+      "epoch": 0.36,
+      "learning_rate": 8.42e-05,
+      "loss": 4.9612,
       "step": 425
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 8.48e-05,
-      "loss": 4.1866,
+      "epoch": 0.36,
+      "learning_rate": 8.44e-05,
+      "loss": 4.2467,
       "step": 426
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 8.5e-05,
-      "loss": 4.1642,
-      "step": 427
+      "epoch": 0.36,
+      "learning_rate": 8.46e-05,
+      "loss": 4.1715,
+      "step": 427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.48e-05,
+      "loss": 4.1792,
+      "step": 428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.5e-05,
+      "loss": 4.7391,
+      "step": 429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.52e-05,
+      "loss": 4.7239,
+      "step": 430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.54e-05,
+      "loss": 4.4149,
+      "step": 431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 4.384,
+      "step": 432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.58e-05,
+      "loss": 4.5157,
+      "step": 433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.6e-05,
+      "loss": 4.1937,
+      "step": 434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 4.4241,
+      "step": 435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.64e-05,
+      "loss": 4.1436,
+      "step": 436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.66e-05,
+      "loss": 4.2146,
+      "step": 437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 4.1829,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.7e-05,
+      "loss": 4.5748,
+      "step": 439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.72e-05,
+      "loss": 4.4626,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 5.1397,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.76e-05,
+      "loss": 4.4431,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.78e-05,
+      "loss": 4.2565,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 4.3701,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.82e-05,
+      "loss": 4.2751,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 4.5515,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.86e-05,
+      "loss": 5.7108,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.88e-05,
+      "loss": 4.5228,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 4.4874,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.92e-05,
+      "loss": 5.0072,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.94e-05,
+      "loss": 4.2581,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 4.1373,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.98e-05,
+      "loss": 4.2505,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9e-05,
+      "loss": 4.5861,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 4.7848,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.04e-05,
+      "loss": 4.7952,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.06e-05,
+      "loss": 4.5531,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 4.9711,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.1e-05,
+      "loss": 4.386,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 4.3429,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 4.5334,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.16e-05,
+      "loss": 4.3281,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 4.2986,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 4.5071,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.22e-05,
+      "loss": 4.4139,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 4.401,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 4.2929,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.28e-05,
+      "loss": 4.3226,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 4.2224,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 4.5265,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 4.2618,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 4.4912,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.38e-05,
+      "loss": 4.0883,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.4e-05,
+      "loss": 4.3745,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.42e-05,
+      "loss": 4.5709,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.44e-05,
+      "loss": 4.5467,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.46e-05,
+      "loss": 4.0715,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.48e-05,
+      "loss": 4.3925,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.5e-05,
+      "loss": 4.4213,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.52e-05,
+      "loss": 4.4422,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.54e-05,
+      "loss": 4.194,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.56e-05,
+      "loss": 4.3123,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.58e-05,
+      "loss": 4.7589,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.6e-05,
+      "loss": 4.439,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 4.4865,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.64e-05,
+      "loss": 4.2871,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.66e-05,
+      "loss": 4.132,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 4.4013,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.7e-05,
+      "loss": 4.3354,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.72e-05,
+      "loss": 4.2273,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.74e-05,
+      "loss": 4.5767,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.76e-05,
+      "loss": 4.775,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.78e-05,
+      "loss": 4.5842,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.8e-05,
+      "loss": 4.503,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.82e-05,
+      "loss": 4.9986,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.84e-05,
+      "loss": 4.7382,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.86e-05,
+      "loss": 4.4995,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.88e-05,
+      "loss": 4.6277,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 4.4889,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.92e-05,
+      "loss": 4.8067,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 4.671250343322754,
+      "eval_runtime": 785.3163,
+      "eval_samples_per_second": 3.364,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 1.9355811186037286,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.94e-05,
+      "loss": 4.4979,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 4.65,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.98e-05,
+      "loss": 4.4584,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001,
+      "loss": 4.4008,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.985486211901307e-05,
+      "loss": 4.4309,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.970972423802612e-05,
+      "loss": 4.658,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.956458635703919e-05,
+      "loss": 4.7666,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.941944847605225e-05,
+      "loss": 4.4283,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.927431059506532e-05,
+      "loss": 4.4507,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.912917271407838e-05,
+      "loss": 4.9356,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.898403483309145e-05,
+      "loss": 4.4599,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.883889695210451e-05,
+      "loss": 4.3327,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.869375907111757e-05,
+      "loss": 4.5705,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.854862119013063e-05,
+      "loss": 4.1369,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.84034833091437e-05,
+      "loss": 4.1613,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.825834542815675e-05,
+      "loss": 4.3556,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.811320754716981e-05,
+      "loss": 4.4996,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.796806966618288e-05,
+      "loss": 4.3395,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.782293178519594e-05,
+      "loss": 4.7805,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.767779390420901e-05,
+      "loss": 4.2884,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.753265602322207e-05,
+      "loss": 4.4185,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.738751814223513e-05,
+      "loss": 4.3528,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.724238026124819e-05,
+      "loss": 4.6593,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.709724238026126e-05,
+      "loss": 4.5369,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.695210449927431e-05,
+      "loss": 4.3855,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.680696661828737e-05,
+      "loss": 4.0847,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.666182873730044e-05,
+      "loss": 4.3983,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.65166908563135e-05,
+      "loss": 4.6859,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.637155297532656e-05,
+      "loss": 4.5279,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.622641509433963e-05,
+      "loss": 4.4907,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.60812772133527e-05,
+      "loss": 4.592,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.593613933236575e-05,
+      "loss": 4.5528,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.579100145137882e-05,
+      "loss": 4.3375,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.564586357039188e-05,
+      "loss": 4.5083,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.550072568940493e-05,
+      "loss": 4.2869,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.5355587808418e-05,
+      "loss": 4.2989,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.521044992743106e-05,
+      "loss": 4.665,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.506531204644412e-05,
+      "loss": 4.4183,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.492017416545718e-05,
+      "loss": 4.7253,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.477503628447025e-05,
+      "loss": 4.5867,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.462989840348333e-05,
+      "loss": 4.7706,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.448476052249638e-05,
+      "loss": 4.5015,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.433962264150944e-05,
+      "loss": 4.2097,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.419448476052251e-05,
+      "loss": 4.5979,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.404934687953556e-05,
+      "loss": 4.8058,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.390420899854863e-05,
+      "loss": 4.4497,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.375907111756169e-05,
+      "loss": 4.8668,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.361393323657474e-05,
+      "loss": 4.8064,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.346879535558781e-05,
+      "loss": 4.7208,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.332365747460087e-05,
+      "loss": 5.4097,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.317851959361394e-05,
+      "loss": 4.5472,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.3033381712627e-05,
+      "loss": 4.5633,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.288824383164007e-05,
+      "loss": 4.7518,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.274310595065312e-05,
+      "loss": 4.6081,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.259796806966619e-05,
+      "loss": 4.5756,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.245283018867925e-05,
+      "loss": 4.5762,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 4.2182,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.216255442670537e-05,
+      "loss": 4.2532,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.201741654571843e-05,
+      "loss": 4.5495,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.18722786647315e-05,
+      "loss": 4.6059,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.172714078374456e-05,
+      "loss": 4.5139,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.158200290275763e-05,
+      "loss": 4.188,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.14368650217707e-05,
+      "loss": 4.9361,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.129172714078375e-05,
+      "loss": 4.3878,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.114658925979681e-05,
+      "loss": 4.5775,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.100145137880988e-05,
+      "loss": 4.3581,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.085631349782293e-05,
+      "loss": 4.5878,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.0711175616836e-05,
+      "loss": 4.4821,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.056603773584906e-05,
+      "loss": 4.19,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.042089985486212e-05,
+      "loss": 4.3486,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.027576197387519e-05,
+      "loss": 4.2667,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.013062409288826e-05,
+      "loss": 4.6684,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.998548621190132e-05,
+      "loss": 4.643,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.984034833091437e-05,
+      "loss": 4.4937,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.969521044992744e-05,
+      "loss": 4.6833,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.95500725689405e-05,
+      "loss": 4.6399,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.940493468795355e-05,
+      "loss": 4.8046,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.925979680696662e-05,
+      "loss": 4.3357,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.911465892597968e-05,
+      "loss": 4.521,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.896952104499274e-05,
+      "loss": 4.4679,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.882438316400582e-05,
+      "loss": 4.5289,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.867924528301888e-05,
+      "loss": 4.5471,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.853410740203193e-05,
+      "loss": 4.7883,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.8388969521045e-05,
+      "loss": 4.672,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.824383164005806e-05,
+      "loss": 4.7275,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.809869375907113e-05,
+      "loss": 4.4835,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.795355587808418e-05,
+      "loss": 4.6237,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.780841799709725e-05,
+      "loss": 4.3526,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.766328011611031e-05,
+      "loss": 4.8443,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.751814223512336e-05,
+      "loss": 4.644,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.737300435413643e-05,
+      "loss": 4.393,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.722786647314949e-05,
+      "loss": 4.1996,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.708272859216256e-05,
+      "loss": 5.0021,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.693759071117562e-05,
+      "loss": 4.3636,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.679245283018869e-05,
+      "loss": 4.6045,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.664731494920174e-05,
+      "loss": 4.1766,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.65021770682148e-05,
+      "loss": 4.326,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.635703918722787e-05,
+      "loss": 5.4465,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.621190130624092e-05,
+      "loss": 5.0553,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.606676342525399e-05,
+      "loss": 5.1031,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.592162554426705e-05,
+      "loss": 4.3411,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.577648766328012e-05,
+      "loss": 4.6369,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.563134978229318e-05,
+      "loss": 4.6698,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.548621190130625e-05,
+      "loss": 4.1823,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.534107402031931e-05,
+      "loss": 4.4147,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.519593613933237e-05,
+      "loss": 4.525,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.505079825834543e-05,
+      "loss": 4.527,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.49056603773585e-05,
+      "loss": 4.5922,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.476052249637155e-05,
+      "loss": 4.4272,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 4.3691,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.447024673439768e-05,
+      "loss": 4.4813,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.432510885341074e-05,
+      "loss": 4.0996,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.417997097242381e-05,
+      "loss": 4.967,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.403483309143688e-05,
+      "loss": 4.219,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.388969521044994e-05,
+      "loss": 4.3202,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.374455732946299e-05,
+      "loss": 4.5202,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.359941944847606e-05,
+      "loss": 4.3135,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.345428156748912e-05,
+      "loss": 4.4754,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.330914368650217e-05,
+      "loss": 4.5077,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.316400580551524e-05,
+      "loss": 4.388,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.30188679245283e-05,
+      "loss": 4.1367,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.287373004354137e-05,
+      "loss": 4.0775,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.272859216255444e-05,
+      "loss": 5.3717,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.25834542815675e-05,
+      "loss": 4.6448,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.243831640058055e-05,
+      "loss": 4.436,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.229317851959362e-05,
+      "loss": 4.9726,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.214804063860668e-05,
+      "loss": 4.894,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.200290275761974e-05,
+      "loss": 4.294,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.18577648766328e-05,
+      "loss": 4.5383,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.171262699564587e-05,
+      "loss": 4.4367,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.156748911465893e-05,
+      "loss": 4.9741,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.142235123367198e-05,
+      "loss": 4.6308,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.127721335268506e-05,
+      "loss": 4.6356,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.113207547169813e-05,
+      "loss": 4.3661,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.098693759071118e-05,
+      "loss": 4.5903,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.084179970972424e-05,
+      "loss": 4.2958,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.069666182873731e-05,
+      "loss": 4.1866,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.055152394775036e-05,
+      "loss": 4.2055,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.040638606676343e-05,
+      "loss": 4.3585,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.026124818577649e-05,
+      "loss": 4.4288,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.011611030478954e-05,
+      "loss": 4.3023,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.997097242380261e-05,
+      "loss": 4.5236,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.982583454281567e-05,
+      "loss": 5.2219,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.968069666182875e-05,
+      "loss": 4.4817,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.95355587808418e-05,
+      "loss": 4.33,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.939042089985487e-05,
+      "loss": 4.5421,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.924528301886794e-05,
+      "loss": 4.0658,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.910014513788099e-05,
+      "loss": 4.763,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.895500725689405e-05,
+      "loss": 4.7283,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.880986937590712e-05,
+      "loss": 4.2155,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.866473149492017e-05,
+      "loss": 4.5573,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.851959361393323e-05,
+      "loss": 4.3935,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.83744557329463e-05,
+      "loss": 4.4796,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.822931785195937e-05,
+      "loss": 4.6213,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.808417997097243e-05,
+      "loss": 4.4425,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.79390420899855e-05,
+      "loss": 4.376,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.779390420899855e-05,
+      "loss": 4.5437,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.764876632801161e-05,
+      "loss": 4.5835,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.750362844702468e-05,
+      "loss": 4.3881,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.735849056603774e-05,
+      "loss": 4.3723,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.72133526850508e-05,
+      "loss": 4.3128,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.706821480406386e-05,
+      "loss": 4.5692,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 4.4778,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.677793904208999e-05,
+      "loss": 4.229,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.663280116110306e-05,
+      "loss": 4.7161,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.648766328011612e-05,
+      "loss": 4.4985,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.634252539912917e-05,
+      "loss": 4.4632,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.619738751814224e-05,
+      "loss": 4.2736,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.60522496371553e-05,
+      "loss": 4.5092,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.590711175616836e-05,
+      "loss": 4.5335,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.576197387518142e-05,
+      "loss": 4.6834,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.561683599419449e-05,
+      "loss": 4.5439,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.547169811320755e-05,
+      "loss": 4.4682,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.532656023222062e-05,
+      "loss": 4.3431,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.518142235123368e-05,
+      "loss": 4.4662,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.503628447024675e-05,
+      "loss": 4.1951,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.48911465892598e-05,
+      "loss": 4.7141,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.474600870827286e-05,
+      "loss": 4.4372,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.460087082728593e-05,
+      "loss": 4.5797,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.445573294629898e-05,
+      "loss": 4.649,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.431059506531205e-05,
+      "loss": 4.1883,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.416545718432511e-05,
+      "loss": 4.2755,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.402031930333816e-05,
+      "loss": 4.3368,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.387518142235124e-05,
+      "loss": 4.1737,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.373004354136431e-05,
+      "loss": 4.6367,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.358490566037736e-05,
+      "loss": 4.5282,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.343976777939043e-05,
+      "loss": 4.3373,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.329462989840349e-05,
+      "loss": 4.2469,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.314949201741656e-05,
+      "loss": 4.4494,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.300435413642961e-05,
+      "loss": 4.4833,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.285921625544267e-05,
+      "loss": 4.7888,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.271407837445574e-05,
+      "loss": 4.501,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.256894049346879e-05,
+      "loss": 4.7076,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.242380261248185e-05,
+      "loss": 4.9082,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.227866473149493e-05,
+      "loss": 4.4738,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.213352685050799e-05,
+      "loss": 4.2174,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.198838896952105e-05,
+      "loss": 4.5306,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.184325108853412e-05,
+      "loss": 4.6403,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.169811320754717e-05,
+      "loss": 4.1501,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.155297532656023e-05,
+      "loss": 5.2222,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.14078374455733e-05,
+      "loss": 4.7203,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.126269956458636e-05,
+      "loss": 4.1881,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.111756168359942e-05,
+      "loss": 4.3717,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.097242380261248e-05,
+      "loss": 4.1561,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.082728592162555e-05,
+      "loss": 4.172,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.068214804063861e-05,
+      "loss": 4.2777,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.053701015965168e-05,
+      "loss": 4.4528,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.039187227866474e-05,
+      "loss": 4.3679,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.02467343976778e-05,
+      "loss": 4.4153,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.010159651669086e-05,
+      "loss": 4.8161,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.995645863570392e-05,
+      "loss": 4.0886,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.981132075471698e-05,
+      "loss": 4.4465,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.966618287373004e-05,
+      "loss": 4.3911,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.95210449927431e-05,
+      "loss": 4.2539,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937590711175617e-05,
+      "loss": 4.5202,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 4.1821,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.90856313497823e-05,
+      "loss": 4.2851,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.894049346879537e-05,
+      "loss": 4.4501,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.879535558780842e-05,
+      "loss": 4.2739,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865021770682148e-05,
+      "loss": 4.2233,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850507982583455e-05,
+      "loss": 4.2593,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83599419448476e-05,
+      "loss": 4.106,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821480406386067e-05,
+      "loss": 4.4189,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.806966618287373e-05,
+      "loss": 4.2338,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.79245283018868e-05,
+      "loss": 4.417,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.777939042089986e-05,
+      "loss": 4.7648,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.763425253991293e-05,
+      "loss": 4.3824,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.748911465892598e-05,
+      "loss": 4.5066,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.734397677793905e-05,
+      "loss": 4.0404,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.719883889695211e-05,
+      "loss": 4.3723,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.705370101596516e-05,
+      "loss": 4.7243,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690856313497823e-05,
+      "loss": 4.4089,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676342525399129e-05,
+      "loss": 4.7071,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661828737300436e-05,
+      "loss": 4.3241,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647314949201742e-05,
+      "loss": 4.4823,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632801161103049e-05,
+      "loss": 4.4137,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618287373004355e-05,
+      "loss": 4.4928,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60377358490566e-05,
+      "loss": 4.4986,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589259796806967e-05,
+      "loss": 4.5346,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574746008708274e-05,
+      "loss": 4.4273,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560232220609579e-05,
+      "loss": 4.4057,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545718432510885e-05,
+      "loss": 4.4471,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531204644412192e-05,
+      "loss": 4.436,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.516690856313497e-05,
+      "loss": 4.392,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502177068214804e-05,
+      "loss": 4.7445,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487663280116111e-05,
+      "loss": 4.6203,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473149492017418e-05,
+      "loss": 4.5648,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458635703918723e-05,
+      "loss": 5.1078,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44412191582003e-05,
+      "loss": 4.3228,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429608127721336e-05,
+      "loss": 4.5166,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415094339622641e-05,
+      "loss": 4.4591,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400580551523948e-05,
+      "loss": 4.5516,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386066763425254e-05,
+      "loss": 4.4334,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37155297532656e-05,
+      "loss": 4.1155,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357039187227866e-05,
+      "loss": 4.5459,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.342525399129173e-05,
+      "loss": 4.2456,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.328011611030479e-05,
+      "loss": 4.1891,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.313497822931786e-05,
+      "loss": 4.1167,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.298984034833092e-05,
+      "loss": 4.181,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.284470246734397e-05,
+      "loss": 4.7468,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.269956458635704e-05,
+      "loss": 4.0969,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.25544267053701e-05,
+      "loss": 4.3632,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.240928882438317e-05,
+      "loss": 4.3691,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.226415094339622e-05,
+      "loss": 4.244,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211901306240929e-05,
+      "loss": 4.4817,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197387518142235e-05,
+      "loss": 4.2074,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182873730043542e-05,
+      "loss": 4.4864,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.168359941944848e-05,
+      "loss": 4.0788,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 4.3257,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.13933236574746e-05,
+      "loss": 4.4328,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.124818577648767e-05,
+      "loss": 4.2076,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.110304789550073e-05,
+      "loss": 4.3631,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.095791001451378e-05,
+      "loss": 4.1903,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0812772133526855e-05,
+      "loss": 4.3872,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.066763425253992e-05,
+      "loss": 4.4072,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0522496371552985e-05,
+      "loss": 4.1748,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.037735849056604e-05,
+      "loss": 4.256,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.02322206095791e-05,
+      "loss": 4.4656,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.008708272859217e-05,
+      "loss": 4.1983,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9941944847605226e-05,
+      "loss": 4.3142,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.979680696661829e-05,
+      "loss": 4.1147,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.965166908563136e-05,
+      "loss": 4.4216,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.950653120464441e-05,
+      "loss": 4.1337,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9361393323657474e-05,
+      "loss": 4.3844,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9216255442670546e-05,
+      "loss": 4.7322,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.90711175616836e-05,
+      "loss": 4.5113,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.892597968069666e-05,
+      "loss": 4.6318,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.878084179970973e-05,
+      "loss": 4.4316,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.863570391872279e-05,
+      "loss": 4.3428,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.849056603773585e-05,
+      "loss": 4.2122,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.834542815674892e-05,
+      "loss": 4.7253,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.820029027576198e-05,
+      "loss": 4.5355,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8055152394775034e-05,
+      "loss": 4.7701,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.79100145137881e-05,
+      "loss": 4.1898,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7764876632801165e-05,
+      "loss": 4.6791,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7619738751814224e-05,
+      "loss": 4.0532,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.747460087082729e-05,
+      "loss": 4.1239,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7329462989840354e-05,
+      "loss": 4.4782,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.718432510885341e-05,
+      "loss": 5.0331,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.703918722786648e-05,
+      "loss": 4.3246,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.689404934687954e-05,
+      "loss": 4.2909,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6748911465892595e-05,
+      "loss": 4.2879,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.660377358490566e-05,
+      "loss": 4.2837,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6458635703918726e-05,
+      "loss": 4.2727,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.631349782293179e-05,
+      "loss": 4.085,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.616835994194485e-05,
+      "loss": 4.2453,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6023222060957915e-05,
+      "loss": 4.463,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587808417997098e-05,
+      "loss": 4.1147,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.573294629898403e-05,
+      "loss": 4.3113,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.55878084179971e-05,
+      "loss": 4.3341,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.544267053701017e-05,
+      "loss": 4.2755,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.529753265602322e-05,
+      "loss": 4.5046,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5152394775036286e-05,
+      "loss": 4.1929,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.500725689404935e-05,
+      "loss": 4.585,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.486211901306241e-05,
+      "loss": 4.1662,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4716981132075475e-05,
+      "loss": 4.3452,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.457184325108854e-05,
+      "loss": 4.3158,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.442670537010159e-05,
+      "loss": 4.1048,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.428156748911466e-05,
+      "loss": 4.1298,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.413642960812772e-05,
+      "loss": 4.3732,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.399129172714079e-05,
+      "loss": 4.477,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 4.0158,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.370101596516691e-05,
+      "loss": 4.3635,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.355587808417998e-05,
+      "loss": 4.4987,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3410740203193036e-05,
+      "loss": 4.2948,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.32656023222061e-05,
+      "loss": 4.4931,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.3120464441219166e-05,
+      "loss": 4.2689,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.297532656023222e-05,
+      "loss": 4.3268,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.283018867924528e-05,
+      "loss": 4.172,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.268505079825835e-05,
+      "loss": 4.2568,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.253991291727141e-05,
+      "loss": 4.4393,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.239477503628447e-05,
+      "loss": 4.3127,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.224963715529754e-05,
+      "loss": 4.274,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.21044992743106e-05,
+      "loss": 4.3069,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1959361393323655e-05,
+      "loss": 4.2938,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.181422351233673e-05,
+      "loss": 4.4433,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.166908563134979e-05,
+      "loss": 4.1943,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1523947750362844e-05,
+      "loss": 4.1027,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.137880986937591e-05,
+      "loss": 4.3145,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1233671988388975e-05,
+      "loss": 4.01,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108853410740203e-05,
+      "loss": 4.3126,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.09433962264151e-05,
+      "loss": 4.4548,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0798258345428164e-05,
+      "loss": 4.7485,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0653120464441215e-05,
+      "loss": 4.7877,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.050798258345428e-05,
+      "loss": 4.2618,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0362844702467346e-05,
+      "loss": 4.1136,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0217706821480404e-05,
+      "loss": 4.4953,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.007256894049347e-05,
+      "loss": 4.0708,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9927431059506535e-05,
+      "loss": 4.5171,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9782293178519594e-05,
+      "loss": 4.4652,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.963715529753266e-05,
+      "loss": 4.5819,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9492017416545724e-05,
+      "loss": 4.2691,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.934687953555878e-05,
+      "loss": 4.3695,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.920174165457185e-05,
+      "loss": 4.3197,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9056603773584906e-05,
+      "loss": 4.22,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.891146589259797e-05,
+      "loss": 4.1893,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.876632801161104e-05,
+      "loss": 4.2481,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8621190130624096e-05,
+      "loss": 4.0581,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8476052249637154e-05,
+      "loss": 4.2424,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833091436865022e-05,
+      "loss": 4.3292,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.818577648766328e-05,
+      "loss": 3.9997,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.804063860667635e-05,
+      "loss": 4.3297,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.789550072568941e-05,
+      "loss": 4.278,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.775036284470247e-05,
+      "loss": 4.2033,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.760522496371553e-05,
+      "loss": 4.3005,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.746008708272859e-05,
+      "loss": 4.2794,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.731494920174166e-05,
+      "loss": 4.0489,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.716981132075472e-05,
+      "loss": 4.3107,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.702467343976778e-05,
+      "loss": 4.5245,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6879535558780845e-05,
+      "loss": 4.0362,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6734397677793904e-05,
+      "loss": 3.8953,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.658925979680697e-05,
+      "loss": 4.086,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6444121915820034e-05,
+      "loss": 4.2383,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.629898403483309e-05,
+      "loss": 4.1686,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 4.2718,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.600870827285922e-05,
+      "loss": 4.1237,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.586357039187228e-05,
+      "loss": 4.1422,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.571843251088535e-05,
+      "loss": 4.608,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5573294629898406e-05,
+      "loss": 4.3258,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5428156748911464e-05,
+      "loss": 4.4232,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528301886792453e-05,
+      "loss": 4.0813,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5137880986937595e-05,
+      "loss": 4.3399,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.499274310595066e-05,
+      "loss": 4.1706,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.484760522496372e-05,
+      "loss": 4.2197,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.470246734397678e-05,
+      "loss": 4.1379,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.455732946298984e-05,
+      "loss": 4.4075,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.441219158200291e-05,
+      "loss": 4.3793,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.4267053701015966e-05,
+      "loss": 4.4398,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.412191582002903e-05,
+      "loss": 4.4393,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.397677793904209e-05,
+      "loss": 4.4366,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3831640058055155e-05,
+      "loss": 4.6884,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3686502177068214e-05,
+      "loss": 4.2599,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.354136429608128e-05,
+      "loss": 4.2376,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3396226415094345e-05,
+      "loss": 4.4408,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.32510885341074e-05,
+      "loss": 4.4194,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.310595065312046e-05,
+      "loss": 4.4417,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.296081277213353e-05,
+      "loss": 4.5396,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.281567489114659e-05,
+      "loss": 4.4014,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.267053701015966e-05,
+      "loss": 4.3996,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2525399129172716e-05,
+      "loss": 4.352,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2380261248185774e-05,
+      "loss": 4.2127,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.223512336719884e-05,
+      "loss": 4.3542,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2089985486211905e-05,
+      "loss": 4.3284,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.194484760522497e-05,
+      "loss": 4.362,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.179970972423803e-05,
+      "loss": 4.1521,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.165457184325109e-05,
+      "loss": 4.5356,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.150943396226415e-05,
+      "loss": 4.3468,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.136429608127722e-05,
+      "loss": 4.5989,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1219158200290276e-05,
+      "loss": 4.2448,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.107402031930334e-05,
+      "loss": 4.5103,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.09288824383164e-05,
+      "loss": 4.1048,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0783744557329466e-05,
+      "loss": 4.0682,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.063860667634253e-05,
+      "loss": 4.2307,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.049346879535559e-05,
+      "loss": 4.257,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0348330914368655e-05,
+      "loss": 4.418,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.020319303338171e-05,
+      "loss": 4.2088,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.005805515239477e-05,
+      "loss": 4.1863,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.991291727140784e-05,
+      "loss": 4.2222,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.97677793904209e-05,
+      "loss": 4.4224,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.962264150943397e-05,
+      "loss": 4.2765,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9477503628447026e-05,
+      "loss": 4.1756,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9332365747460085e-05,
+      "loss": 4.396,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.918722786647315e-05,
+      "loss": 4.0434,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9042089985486215e-05,
+      "loss": 4.1402,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.8896952104499274e-05,
+      "loss": 4.0033,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.875181422351234e-05,
+      "loss": 4.1007,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.86066763425254e-05,
+      "loss": 4.2354,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 4.2295,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.831640058055153e-05,
+      "loss": 4.0592,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.817126269956459e-05,
+      "loss": 4.0171,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.802612481857765e-05,
+      "loss": 4.3427,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.788098693759071e-05,
+      "loss": 4.2882,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 4.2127,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.759071117561684e-05,
+      "loss": 4.2399,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.74455732946299e-05,
+      "loss": 4.3821,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7300435413642965e-05,
+      "loss": 4.3813,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7155297532656023e-05,
+      "loss": 4.3656,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.701015965166908e-05,
+      "loss": 4.3246,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6865021770682154e-05,
+      "loss": 4.1945,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.671988388969521e-05,
+      "loss": 4.4663,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.657474600870828e-05,
+      "loss": 4.145,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6429608127721336e-05,
+      "loss": 4.3821,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6284470246734395e-05,
+      "loss": 3.9873,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.613933236574747e-05,
+      "loss": 4.2039,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.5994194484760525e-05,
+      "loss": 4.3081,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5849056603773584e-05,
+      "loss": 4.0279,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.570391872278665e-05,
+      "loss": 4.36,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.555878084179971e-05,
+      "loss": 4.3751,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.541364296081277e-05,
+      "loss": 4.4976,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.526850507982584e-05,
+      "loss": 4.0955,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.51233671988389e-05,
+      "loss": 4.0744,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.497822931785196e-05,
+      "loss": 4.4292,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.483309143686502e-05,
+      "loss": 4.1521,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4687953555878086e-05,
+      "loss": 3.9292,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.454281567489115e-05,
+      "loss": 4.267,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.439767779390421e-05,
+      "loss": 4.2647,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4252539912917275e-05,
+      "loss": 4.5288,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4107402031930334e-05,
+      "loss": 4.2394,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.39622641509434e-05,
+      "loss": 4.2262,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3817126269956464e-05,
+      "loss": 4.1043,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.367198838896952e-05,
+      "loss": 4.1359,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.352685050798258e-05,
+      "loss": 4.3121,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3381712626995646e-05,
+      "loss": 3.9559,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.323657474600871e-05,
+      "loss": 4.3733,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.309143686502178e-05,
+      "loss": 4.2911,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2946298984034836e-05,
+      "loss": 3.9729,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2801161103047894e-05,
+      "loss": 4.2171,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.265602322206096e-05,
+      "loss": 4.286,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.251088534107402e-05,
+      "loss": 3.8745,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.236574746008709e-05,
+      "loss": 4.1338,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.222060957910015e-05,
+      "loss": 3.9965,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207547169811321e-05,
+      "loss": 4.0498,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.193033381712627e-05,
+      "loss": 4.3367,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.178519593613933e-05,
+      "loss": 4.102,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1640058055152396e-05,
+      "loss": 4.2302,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.149492017416546e-05,
+      "loss": 4.2506,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.134978229317852e-05,
+      "loss": 4.1119,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1204644412191585e-05,
+      "loss": 4.1239,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1059506531204644e-05,
+      "loss": 4.0464,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.091436865021771e-05,
+      "loss": 4.2744,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 4.7272,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.062409288824383e-05,
+      "loss": 4.459,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.047895500725689e-05,
+      "loss": 4.0873,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.033381712626996e-05,
+      "loss": 4.343,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.018867924528302e-05,
+      "loss": 4.0366,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0043541364296084e-05,
+      "loss": 4.0833,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9898403483309146e-05,
+      "loss": 4.2542,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9753265602322204e-05,
+      "loss": 4.1514,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9608127721335273e-05,
+      "loss": 4.288,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.946298984034833e-05,
+      "loss": 4.1444,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9317851959361393e-05,
+      "loss": 4.3545,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.917271407837446e-05,
+      "loss": 4.1981,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9027576197387517e-05,
+      "loss": 3.8983,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8882438316400583e-05,
+      "loss": 4.2837,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8737300435413644e-05,
+      "loss": 4.2541,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8592162554426706e-05,
+      "loss": 5.0162,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.844702467343977e-05,
+      "loss": 4.5207,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 4.211,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8156748911465895e-05,
+      "loss": 4.367,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8011611030478957e-05,
+      "loss": 4.425,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 4.305517673492432,
+      "eval_runtime": 784.9234,
+      "eval_samples_per_second": 3.366,
+      "eval_steps_per_second": 0.282,
+      "eval_wer": 1.9549980166600556,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7866473149492016e-05,
+      "loss": 4.2513,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7721335268505085e-05,
+      "loss": 4.2343,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7576197387518143e-05,
+      "loss": 4.1411,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7431059506531205e-05,
+      "loss": 4.2902,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.728592162554427e-05,
+      "loss": 4.2419,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.714078374455733e-05,
+      "loss": 4.1095,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6995645863570394e-05,
+      "loss": 4.1266,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6850507982583456e-05,
+      "loss": 4.1181,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6705370101596518e-05,
+      "loss": 4.1899,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6560232220609583e-05,
+      "loss": 4.2387,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 4.0676,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6269956458635704e-05,
+      "loss": 4.1863,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.612481857764877e-05,
+      "loss": 4.0467,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5979680696661827e-05,
+      "loss": 4.1038,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5834542815674896e-05,
+      "loss": 4.173,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5689404934687955e-05,
+      "loss": 4.3439,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5544267053701017e-05,
+      "loss": 4.0359,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5399129172714082e-05,
+      "loss": 4.3483,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.525399129172714e-05,
+      "loss": 4.3164,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5108853410740202e-05,
+      "loss": 3.976,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4963715529753268e-05,
+      "loss": 4.1258,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.481857764876633e-05,
+      "loss": 4.4515,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.467343976777939e-05,
+      "loss": 4.6965,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4528301886792453e-05,
+      "loss": 3.9486,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.438316400580552e-05,
+      "loss": 4.2184,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4238026124818577e-05,
+      "loss": 4.0248,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.409288824383164e-05,
+      "loss": 4.2345,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3947750362844704e-05,
+      "loss": 4.0615,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3802612481857766e-05,
+      "loss": 4.058,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.365747460087083e-05,
+      "loss": 3.9174,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.351233671988389e-05,
+      "loss": 4.1625,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3367198838896952e-05,
+      "loss": 4.1846,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3222060957910017e-05,
+      "loss": 4.3224,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 4.0142,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.293178519593614e-05,
+      "loss": 4.017,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2786647314949203e-05,
+      "loss": 3.9842,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 4.0666,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.249637155297533e-05,
+      "loss": 4.3419,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.235123367198839e-05,
+      "loss": 4.1773,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2206095791001454e-05,
+      "loss": 4.1623,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2060957910014516e-05,
+      "loss": 4.3386,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1915820029027578e-05,
+      "loss": 4.1176,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.177068214804064e-05,
+      "loss": 4.1972,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.16255442670537e-05,
+      "loss": 4.3177,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1480406386066763e-05,
+      "loss": 4.4122,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.133526850507983e-05,
+      "loss": 4.0535,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1190130624092887e-05,
+      "loss": 4.1067,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1044992743105953e-05,
+      "loss": 4.3154,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0899854862119014e-05,
+      "loss": 4.7161,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0754716981132076e-05,
+      "loss": 4.3966,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0609579100145138e-05,
+      "loss": 4.1921,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.04644412191582e-05,
+      "loss": 4.0835,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0319303338171265e-05,
+      "loss": 4.0379,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0174165457184327e-05,
+      "loss": 4.2727,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0029027576197386e-05,
+      "loss": 4.2015,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.988388969521045e-05,
+      "loss": 4.3016,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9738751814223513e-05,
+      "loss": 4.2272,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9593613933236575e-05,
+      "loss": 3.9746,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9448476052249637e-05,
+      "loss": 4.2535,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.93033381712627e-05,
+      "loss": 4.3878,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9158200290275764e-05,
+      "loss": 4.2485,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9013062409288826e-05,
+      "loss": 4.0494,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 4.2857,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.872278664731495e-05,
+      "loss": 4.1066,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8577648766328012e-05,
+      "loss": 4.0084,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8432510885341077e-05,
+      "loss": 4.044,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828737300435414e-05,
+      "loss": 4.1652,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8142235123367197e-05,
+      "loss": 4.0841,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7997097242380263e-05,
+      "loss": 4.1078,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7851959361393325e-05,
+      "loss": 4.2325,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7706821480406387e-05,
+      "loss": 4.1537,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.756168359941945e-05,
+      "loss": 4.1244,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.741654571843251e-05,
+      "loss": 3.8679,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7271407837445576e-05,
+      "loss": 4.0223,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7126269956458638e-05,
+      "loss": 4.0643,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 4.098,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.683599419448476e-05,
+      "loss": 4.5115,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6690856313497823e-05,
+      "loss": 4.2147,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.654571843251089e-05,
+      "loss": 4.154,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6400580551523947e-05,
+      "loss": 3.933,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.625544267053701e-05,
+      "loss": 4.313,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6110304789550074e-05,
+      "loss": 4.1882,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5965166908563136e-05,
+      "loss": 4.1372,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5820029027576198e-05,
+      "loss": 4.1378,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.567489114658926e-05,
+      "loss": 4.1729,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5529753265602322e-05,
+      "loss": 4.0639,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 4.1382,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5239477503628446e-05,
+      "loss": 4.1558,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 4.0708,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4949201741654573e-05,
+      "loss": 4.1351,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4804063860667636e-05,
+      "loss": 4.0931,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4658925979680697e-05,
+      "loss": 4.0607,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4513788098693759e-05,
+      "loss": 4.0852,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4368650217706822e-05,
+      "loss": 4.2926,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4223512336719886e-05,
+      "loss": 4.1314,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4078374455732948e-05,
+      "loss": 4.6991,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3933236574746008e-05,
+      "loss": 4.2806,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3788098693759072e-05,
+      "loss": 4.4607,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3642960812772135e-05,
+      "loss": 4.5837,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3497822931785197e-05,
+      "loss": 4.1257,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3352685050798259e-05,
+      "loss": 4.4883,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 4.1343,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3062409288824384e-05,
+      "loss": 4.1905,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2917271407837448e-05,
+      "loss": 3.9722,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2772133526850508e-05,
+      "loss": 4.1061,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.262699564586357e-05,
+      "loss": 3.9549,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2481857764876634e-05,
+      "loss": 4.3045,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2336719883889696e-05,
+      "loss": 4.2633,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.219158200290276e-05,
+      "loss": 4.2909,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.204644412191582e-05,
+      "loss": 4.0318,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1901306240928883e-05,
+      "loss": 4.129,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1756168359941945e-05,
+      "loss": 4.3877,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1611030478955009e-05,
+      "loss": 4.1647,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.146589259796807e-05,
+      "loss": 3.8328,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 4.2986,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1175616835994194e-05,
+      "loss": 4.3847,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1030478955007258e-05,
+      "loss": 4.3198,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.088534107402032e-05,
+      "loss": 3.9998,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0740203193033382e-05,
+      "loss": 3.8145,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0595065312046444e-05,
+      "loss": 4.2237,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0449927431059507e-05,
+      "loss": 4.2272,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0304789550072569e-05,
+      "loss": 4.3583,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0159651669085633e-05,
+      "loss": 3.9839,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0014513788098693e-05,
+      "loss": 4.1836,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.869375907111757e-06,
+      "loss": 4.3229,
+      "step": 1125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.724238026124818e-06,
+      "loss": 4.1746,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.579100145137882e-06,
+      "loss": 4.2639,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 4.0946,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.288824383164006e-06,
+      "loss": 4.1368,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.14368650217707e-06,
+      "loss": 4.3802,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.998548621190131e-06,
+      "loss": 3.8914,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.853410740203193e-06,
+      "loss": 4.0122,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 4.3009,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.563134978229319e-06,
+      "loss": 4.0771,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.41799709724238e-06,
+      "loss": 4.3787,
+      "step": 1135
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.52e-05,
-      "loss": 4.1608,
-      "step": 428
+      "learning_rate": 8.272859216255444e-06,
+      "loss": 4.0165,
+      "step": 1136
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.54e-05,
-      "loss": 4.0655,
-      "step": 429
+      "learning_rate": 8.127721335268504e-06,
+      "loss": 4.0955,
+      "step": 1137
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.560000000000001e-05,
-      "loss": 4.2002,
-      "step": 430
+      "learning_rate": 7.982583454281568e-06,
+      "loss": 4.0912,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.83744557329463e-06,
+      "loss": 4.0026,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 4.0917,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 4.414,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.402031930333818e-06,
+      "loss": 4.1832,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.256894049346879e-06,
+      "loss": 4.4302,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111756168359943e-06,
+      "loss": 4.3394,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.966618287373004e-06,
+      "loss": 4.0713,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.8214804063860676e-06,
+      "loss": 4.5296,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.6763425253991295e-06,
+      "loss": 4.8123,
+      "step": 1147
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.58e-05,
-      "loss": 4.1811,
-      "step": 431
+      "learning_rate": 6.531204644412192e-06,
+      "loss": 4.2825,
+      "step": 1148
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.6e-05,
-      "loss": 4.3704,
-      "step": 432
+      "learning_rate": 6.386066763425254e-06,
+      "loss": 4.5205,
+      "step": 1149
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.620000000000001e-05,
-      "loss": 4.1261,
-      "step": 433
+      "learning_rate": 6.240928882438317e-06,
+      "loss": 4.2223,
+      "step": 1150
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.64e-05,
-      "loss": 4.1063,
-      "step": 434
+      "learning_rate": 6.09579100145138e-06,
+      "loss": 4.0581,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9506531204644415e-06,
+      "loss": 4.0707,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.805515239477504e-06,
+      "loss": 4.0264,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 4.0049,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.515239477503629e-06,
+      "loss": 4.2789,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.370101596516691e-06,
+      "loss": 4.1889,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.224963715529754e-06,
+      "loss": 4.1622,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.079825834542816e-06,
+      "loss": 4.1381,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934687953555878e-06,
+      "loss": 3.9792,
+      "step": 1159
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.66e-05,
-      "loss": 4.1748,
-      "step": 435
+      "learning_rate": 4.789550072568941e-06,
+      "loss": 4.2352,
+      "step": 1160
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.680000000000001e-05,
-      "loss": 4.0705,
-      "step": 436
+      "learning_rate": 4.644412191582003e-06,
+      "loss": 4.0867,
+      "step": 1161
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.7e-05,
-      "loss": 4.1026,
-      "step": 437
+      "learning_rate": 4.499274310595066e-06,
+      "loss": 4.2896,
+      "step": 1162
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.72e-05,
-      "loss": 4.3491,
-      "step": 438
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 4.2937,
+      "step": 1163
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 4.2061,
-      "step": 439
+      "learning_rate": 4.20899854862119e-06,
+      "loss": 4.0065,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.063860667634252e-06,
+      "loss": 4.138,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.918722786647315e-06,
+      "loss": 4.2409,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 3.8939,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6284470246734397e-06,
+      "loss": 4.0644,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.483309143686502e-06,
+      "loss": 4.1003,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3381712626995647e-06,
+      "loss": 4.1696,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.193033381712627e-06,
+      "loss": 4.0985,
+      "step": 1171
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.76e-05,
-      "loss": 4.2527,
-      "step": 440
+      "learning_rate": 3.04789550072569e-06,
+      "loss": 3.9381,
+      "step": 1172
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.78e-05,
-      "loss": 4.3031,
-      "step": 441
+      "learning_rate": 2.902757619738752e-06,
+      "loss": 4.1201,
+      "step": 1173
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 4.3801,
-      "step": 442
+      "learning_rate": 2.7576197387518145e-06,
+      "loss": 4.1123,
+      "step": 1174
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.82e-05,
-      "loss": 4.467,
-      "step": 443
+      "learning_rate": 2.612481857764877e-06,
+      "loss": 3.9832,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.467343976777939e-06,
+      "loss": 4.1147,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3222060957910015e-06,
+      "loss": 4.1293,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.177068214804064e-06,
+      "loss": 4.4824,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.031930333817126e-06,
+      "loss": 3.9593,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8867924528301887e-06,
+      "loss": 4.263,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.741654571843251e-06,
+      "loss": 4.1339,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5965166908563135e-06,
+      "loss": 4.0985,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.451378809869376e-06,
+      "loss": 4.1598,
+      "step": 1183
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.840000000000001e-05,
-      "loss": 4.3737,
-      "step": 444
+      "learning_rate": 1.3062409288824384e-06,
+      "loss": 4.1815,
+      "step": 1184
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.86e-05,
-      "loss": 4.5371,
-      "step": 445
+      "learning_rate": 1.1611030478955007e-06,
+      "loss": 4.1236,
+      "step": 1185
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.88e-05,
-      "loss": 4.1867,
-      "step": 446
+      "learning_rate": 1.015965166908563e-06,
+      "loss": 4.2243,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.708272859216255e-07,
+      "loss": 4.1232,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.25689404934688e-07,
+      "loss": 4.1697,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.805515239477504e-07,
+      "loss": 4.7853,
+      "step": 1189
     },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 4.256010225535508,
-      "train_runtime": 5087.4578,
-      "train_samples_per_second": 5.609,
-      "train_steps_per_second": 0.088
+      "train_loss": 4.355411059521546,
+      "train_runtime": 6633.4843,
+      "train_samples_per_second": 4.302,
+      "train_steps_per_second": 0.179
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,