diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9995796553173603,
-  "global_step": 1189,
+  "epoch": 0.9987389659520807,
+  "global_step": 594,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,7166 +10,3587 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 4.8825,
+      "loss": 4.8042,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.0739,
+      "learning_rate": 2e-08,
+      "loss": 4.9828,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 2e-08,
-      "loss": 5.0609,
+      "epoch": 0.01,
+      "learning_rate": 4e-08,
+      "loss": 4.8735,
       "step": 3
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 4e-08,
-      "loss": 4.8813,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 4.8238,
       "step": 4
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 4.6007,
+      "epoch": 0.01,
+      "learning_rate": 8e-08,
+      "loss": 4.7747,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8e-08,
-      "loss": 4.9746,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 4.7918,
       "step": 6
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 4.743,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.8798,
       "step": 7
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.7243,
+      "learning_rate": 1.4e-07,
+      "loss": 4.7786,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.9582,
+      "epoch": 0.02,
+      "learning_rate": 1.6e-07,
+      "loss": 4.9028,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4e-07,
-      "loss": 4.8108,
+      "epoch": 0.02,
+      "learning_rate": 1.8e-07,
+      "loss": 4.8233,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6e-07,
-      "loss": 4.8627,
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.8176,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-07,
-      "loss": 4.7085,
+      "epoch": 0.02,
+      "learning_rate": 2.2e-07,
+      "loss": 4.8817,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 5.0112,
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.8089,
       "step": 13
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.2e-07,
-      "loss": 4.7932,
+      "epoch": 0.02,
+      "learning_rate": 2.6e-07,
+      "loss": 4.7724,
       "step": 14
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.6951,
+      "epoch": 0.03,
+      "learning_rate": 2.8e-07,
+      "loss": 4.662,
       "step": 15
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.6e-07,
-      "loss": 4.9232,
+      "epoch": 0.03,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.8532,
       "step": 16
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.8e-07,
-      "loss": 4.8666,
+      "epoch": 0.03,
+      "learning_rate": 3.2e-07,
+      "loss": 4.8698,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 5.0363,
+      "epoch": 0.03,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 4.9494,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2e-07,
-      "loss": 4.7808,
+      "epoch": 0.03,
+      "learning_rate": 3.6e-07,
+      "loss": 4.7211,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 5.0227,
+      "epoch": 0.03,
+      "learning_rate": 3.8e-07,
+      "loss": 4.9558,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-07,
-      "loss": 4.9847,
+      "epoch": 0.04,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.9076,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.8e-07,
-      "loss": 4.7493,
+      "epoch": 0.04,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 4.8373,
       "step": 22
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.9432,
+      "epoch": 0.04,
+      "learning_rate": 4.4e-07,
+      "loss": 4.8605,
       "step": 23
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 5.0013,
+      "epoch": 0.04,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 4.9537,
       "step": 24
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.4e-07,
-      "loss": 4.7983,
+      "epoch": 0.04,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.7198,
       "step": 25
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.6000000000000004e-07,
-      "loss": 4.9729,
+      "epoch": 0.04,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 5.1327,
       "step": 26
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 4.8627,
+      "epoch": 0.05,
+      "learning_rate": 5.2e-07,
+      "loss": 4.8064,
       "step": 27
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 4.6941,
+      "epoch": 0.05,
+      "learning_rate": 5.4e-07,
+      "loss": 4.7093,
       "step": 28
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.2e-07,
-      "loss": 4.7796,
+      "epoch": 0.05,
+      "learning_rate": 5.4e-07,
+      "loss": 4.7345,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-07,
-      "loss": 4.8117,
+      "epoch": 0.05,
+      "learning_rate": 5.6e-07,
+      "loss": 4.8727,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.6e-07,
-      "loss": 4.9358,
+      "epoch": 0.05,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 4.8892,
       "step": 31
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.800000000000001e-07,
-      "loss": 4.7575,
+      "epoch": 0.05,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.7728,
       "step": 32
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.9002,
+      "epoch": 0.06,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 4.78,
       "step": 33
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.200000000000001e-07,
-      "loss": 4.5657,
+      "epoch": 0.06,
+      "learning_rate": 6.4e-07,
+      "loss": 4.6903,
       "step": 34
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.4e-07,
-      "loss": 4.7417,
+      "epoch": 0.06,
+      "learning_rate": 6.6e-07,
+      "loss": 4.7677,
       "step": 35
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.6e-07,
-      "loss": 5.0702,
+      "epoch": 0.06,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 4.9281,
       "step": 36
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.800000000000001e-07,
-      "loss": 4.8929,
+      "epoch": 0.06,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 4.7253,
       "step": 37
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.000000000000001e-07,
-      "loss": 4.745,
+      "epoch": 0.06,
+      "learning_rate": 7.2e-07,
+      "loss": 4.8761,
       "step": 38
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.2e-07,
-      "loss": 5.2413,
+      "epoch": 0.07,
+      "learning_rate": 7.4e-07,
+      "loss": 5.0624,
       "step": 39
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.4e-07,
-      "loss": 4.8466,
+      "epoch": 0.07,
+      "learning_rate": 7.6e-07,
+      "loss": 4.6726,
       "step": 40
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.6e-07,
-      "loss": 4.702,
+      "epoch": 0.07,
+      "learning_rate": 7.8e-07,
+      "loss": 4.7354,
       "step": 41
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.8e-07,
-      "loss": 4.7698,
+      "epoch": 0.07,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.987,
       "step": 42
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.795,
+      "epoch": 0.07,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 4.739,
       "step": 43
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.200000000000001e-07,
-      "loss": 4.8656,
+      "epoch": 0.07,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 4.8959,
       "step": 44
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.400000000000001e-07,
-      "loss": 5.0811,
+      "epoch": 0.08,
+      "learning_rate": 8.6e-07,
+      "loss": 5.0197,
       "step": 45
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.6e-07,
-      "loss": 5.0927,
+      "epoch": 0.08,
+      "learning_rate": 8.8e-07,
+      "loss": 5.1426,
       "step": 46
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.8e-07,
-      "loss": 5.3102,
+      "epoch": 0.08,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 5.0508,
       "step": 47
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.000000000000001e-07,
-      "loss": 5.5342,
+      "epoch": 0.08,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 5.4743,
       "step": 48
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.6641,
+      "epoch": 0.08,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 5.5167,
       "step": 49
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.6525,
+      "epoch": 0.08,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 5.3789,
       "step": 50
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.400000000000001e-07,
-      "loss": 4.624,
+      "epoch": 0.09,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 4.5449,
       "step": 51
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 4.6762,
+      "epoch": 0.09,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 4.5448,
       "step": 52
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.800000000000001e-07,
-      "loss": 4.4921,
+      "epoch": 0.09,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.609,
       "step": 53
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.5681,
+      "epoch": 0.09,
+      "learning_rate": 1.02e-06,
+      "loss": 4.6805,
       "step": 54
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.02e-06,
-      "loss": 4.7185,
+      "epoch": 0.09,
+      "learning_rate": 1.04e-06,
+      "loss": 4.5197,
       "step": 55
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.04e-06,
-      "loss": 4.4741,
+      "epoch": 0.09,
+      "learning_rate": 1.06e-06,
+      "loss": 4.7056,
       "step": 56
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.06e-06,
-      "loss": 4.7435,
+      "epoch": 0.1,
+      "learning_rate": 1.08e-06,
+      "loss": 4.5713,
       "step": 57
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.08e-06,
-      "loss": 4.6779,
+      "epoch": 0.1,
+      "learning_rate": 1.1e-06,
+      "loss": 4.8102,
       "step": 58
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.1e-06,
-      "loss": 4.5611,
+      "epoch": 0.1,
+      "learning_rate": 1.12e-06,
+      "loss": 4.6042,
       "step": 59
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.12e-06,
-      "loss": 4.7428,
+      "epoch": 0.1,
+      "learning_rate": 1.14e-06,
+      "loss": 4.3496,
       "step": 60
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.14e-06,
-      "loss": 4.4808,
+      "epoch": 0.1,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 4.6199,
       "step": 61
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.1600000000000001e-06,
-      "loss": 4.6633,
+      "epoch": 0.1,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 4.675,
       "step": 62
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.1800000000000001e-06,
-      "loss": 4.5837,
+      "epoch": 0.11,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.5535,
       "step": 63
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.4583,
+      "epoch": 0.11,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 4.4975,
       "step": 64
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.2200000000000002e-06,
-      "loss": 4.406,
+      "epoch": 0.11,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 4.5554,
       "step": 65
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.2400000000000002e-06,
-      "loss": 4.5595,
+      "epoch": 0.11,
+      "learning_rate": 1.26e-06,
+      "loss": 4.4926,
       "step": 66
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.26e-06,
-      "loss": 4.72,
+      "epoch": 0.11,
+      "learning_rate": 1.28e-06,
+      "loss": 4.4971,
       "step": 67
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.28e-06,
-      "loss": 4.5178,
+      "epoch": 0.11,
+      "learning_rate": 1.3e-06,
+      "loss": 4.4337,
       "step": 68
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3e-06,
-      "loss": 4.4877,
+      "epoch": 0.12,
+      "learning_rate": 1.32e-06,
+      "loss": 4.5496,
       "step": 69
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.32e-06,
-      "loss": 4.5414,
+      "epoch": 0.12,
+      "learning_rate": 1.34e-06,
+      "loss": 4.5592,
       "step": 70
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.34e-06,
-      "loss": 4.7282,
+      "epoch": 0.12,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 4.6406,
       "step": 71
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3600000000000001e-06,
-      "loss": 4.5943,
+      "epoch": 0.12,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 4.5955,
       "step": 72
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3800000000000001e-06,
-      "loss": 4.5429,
+      "epoch": 0.12,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.3985,
       "step": 73
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.63,
+      "epoch": 0.12,
+      "learning_rate": 1.42e-06,
+      "loss": 4.582,
       "step": 74
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.42e-06,
-      "loss": 4.4177,
+      "epoch": 0.13,
+      "learning_rate": 1.44e-06,
+      "loss": 4.5618,
       "step": 75
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.44e-06,
-      "loss": 4.9119,
+      "epoch": 0.13,
+      "learning_rate": 1.46e-06,
+      "loss": 4.561,
       "step": 76
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.46e-06,
-      "loss": 4.5759,
+      "epoch": 0.13,
+      "learning_rate": 1.48e-06,
+      "loss": 4.5283,
       "step": 77
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.48e-06,
-      "loss": 4.4309,
+      "epoch": 0.13,
+      "learning_rate": 1.5e-06,
+      "loss": 4.5355,
       "step": 78
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.5e-06,
-      "loss": 4.4874,
+      "epoch": 0.13,
+      "learning_rate": 1.52e-06,
+      "loss": 4.412,
       "step": 79
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.52e-06,
-      "loss": 4.5647,
+      "epoch": 0.13,
+      "learning_rate": 1.54e-06,
+      "loss": 4.5652,
       "step": 80
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.54e-06,
-      "loss": 4.5978,
+      "epoch": 0.14,
+      "learning_rate": 1.56e-06,
+      "loss": 4.5097,
       "step": 81
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.56e-06,
-      "loss": 4.5589,
+      "epoch": 0.14,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 4.5247,
       "step": 82
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.5800000000000001e-06,
-      "loss": 4.4532,
+      "epoch": 0.14,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.5395,
       "step": 83
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.5699,
+      "epoch": 0.14,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 4.5561,
       "step": 84
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6200000000000002e-06,
-      "loss": 4.4818,
+      "epoch": 0.14,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 4.4865,
       "step": 85
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6400000000000002e-06,
-      "loss": 4.4766,
+      "epoch": 0.14,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 4.3575,
       "step": 86
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6600000000000002e-06,
-      "loss": 4.3995,
+      "epoch": 0.15,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 4.6712,
       "step": 87
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6800000000000002e-06,
-      "loss": 4.5932,
+      "epoch": 0.15,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 4.5597,
       "step": 88
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.7000000000000002e-06,
-      "loss": 4.6113,
+      "epoch": 0.15,
+      "learning_rate": 1.72e-06,
+      "loss": 4.5239,
       "step": 89
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.72e-06,
-      "loss": 4.5185,
+      "epoch": 0.15,
+      "learning_rate": 1.74e-06,
+      "loss": 4.568,
       "step": 90
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.74e-06,
-      "loss": 4.5992,
+      "epoch": 0.15,
+      "learning_rate": 1.76e-06,
+      "loss": 4.5041,
       "step": 91
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.76e-06,
-      "loss": 4.6589,
+      "epoch": 0.15,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 4.6208,
       "step": 92
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.7800000000000001e-06,
-      "loss": 4.2828,
+      "epoch": 0.16,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 4.4879,
       "step": 93
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8000000000000001e-06,
-      "loss": 4.6509,
+      "epoch": 0.16,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 4.7273,
       "step": 94
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8200000000000002e-06,
-      "loss": 4.8917,
+      "epoch": 0.16,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 4.6565,
       "step": 95
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8400000000000002e-06,
-      "loss": 4.7116,
+      "epoch": 0.16,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 4.694,
       "step": 96
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8600000000000002e-06,
-      "loss": 4.6296,
+      "epoch": 0.16,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 4.7673,
       "step": 97
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 4.8917,
+      "epoch": 0.16,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 4.5242,
       "step": 98
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9000000000000002e-06,
-      "loss": 4.9009,
+      "epoch": 0.17,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 4.7109,
       "step": 99
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.7752,
+      "epoch": 0.17,
+      "learning_rate": 1.94e-06,
+      "loss": 4.4853,
       "step": 100
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.94e-06,
-      "loss": 4.1803,
+      "epoch": 0.17,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 4.5437,
       "step": 101
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9600000000000003e-06,
-      "loss": 4.4801,
+      "epoch": 0.17,
+      "learning_rate": 1.98e-06,
+      "loss": 4.3413,
       "step": 102
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.98e-06,
-      "loss": 4.3671,
+      "epoch": 0.17,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.3155,
       "step": 103
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.4625,
+      "epoch": 0.17,
+      "learning_rate": 2.02e-06,
+      "loss": 4.2955,
       "step": 104
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.02e-06,
-      "loss": 4.3437,
+      "epoch": 0.18,
+      "learning_rate": 2.04e-06,
+      "loss": 4.2886,
       "step": 105
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.04e-06,
-      "loss": 4.4997,
+      "epoch": 0.18,
+      "learning_rate": 2.06e-06,
+      "loss": 4.325,
       "step": 106
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.06e-06,
-      "loss": 4.615,
+      "epoch": 0.18,
+      "learning_rate": 2.08e-06,
+      "loss": 4.4709,
       "step": 107
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.08e-06,
-      "loss": 4.5304,
+      "epoch": 0.18,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.4975,
       "step": 108
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.4706,
+      "epoch": 0.18,
+      "learning_rate": 2.12e-06,
+      "loss": 4.5738,
       "step": 109
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.12e-06,
-      "loss": 4.1576,
+      "epoch": 0.18,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 4.3586,
       "step": 110
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.1400000000000003e-06,
-      "loss": 4.4798,
+      "epoch": 0.19,
+      "learning_rate": 2.16e-06,
+      "loss": 4.3516,
       "step": 111
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.16e-06,
-      "loss": 4.5613,
+      "epoch": 0.19,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 4.4207,
       "step": 112
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.1800000000000003e-06,
-      "loss": 4.3502,
+      "epoch": 0.19,
+      "learning_rate": 2.2e-06,
+      "loss": 4.4198,
       "step": 113
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.2e-06,
-      "loss": 4.5198,
+      "epoch": 0.19,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 4.441,
       "step": 114
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.2200000000000003e-06,
-      "loss": 4.4196,
+      "epoch": 0.19,
+      "learning_rate": 2.24e-06,
+      "loss": 4.3794,
       "step": 115
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.24e-06,
-      "loss": 4.5957,
+      "epoch": 0.2,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 4.3902,
       "step": 116
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.2600000000000004e-06,
-      "loss": 4.3273,
+      "epoch": 0.2,
+      "learning_rate": 2.28e-06,
+      "loss": 4.4838,
       "step": 117
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.28e-06,
-      "loss": 4.1643,
+      "epoch": 0.2,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 4.2504,
       "step": 118
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.3000000000000004e-06,
-      "loss": 4.269,
+      "epoch": 0.2,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 4.228,
       "step": 119
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.3200000000000002e-06,
-      "loss": 4.3791,
+      "epoch": 0.2,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 4.2372,
       "step": 120
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.3400000000000005e-06,
-      "loss": 4.6536,
+      "epoch": 0.2,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 4.4673,
       "step": 121
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.3600000000000003e-06,
-      "loss": 4.4041,
+      "epoch": 0.21,
+      "learning_rate": 2.38e-06,
+      "loss": 4.553,
       "step": 122
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.38e-06,
-      "loss": 4.4627,
+      "epoch": 0.21,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.315,
       "step": 123
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.4345,
+      "epoch": 0.21,
+      "learning_rate": 2.42e-06,
+      "loss": 4.3808,
       "step": 124
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.42e-06,
-      "loss": 4.224,
+      "epoch": 0.21,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 4.3912,
       "step": 125
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.4400000000000004e-06,
-      "loss": 4.4019,
+      "epoch": 0.21,
+      "learning_rate": 2.46e-06,
+      "loss": 4.2075,
       "step": 126
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.46e-06,
-      "loss": 4.6445,
+      "epoch": 0.21,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 4.266,
       "step": 127
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.4800000000000004e-06,
-      "loss": 4.2698,
+      "epoch": 0.22,
+      "learning_rate": 2.5e-06,
+      "loss": 4.2219,
       "step": 128
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.5e-06,
-      "loss": 4.13,
+      "epoch": 0.22,
+      "learning_rate": 2.52e-06,
+      "loss": 4.3792,
       "step": 129
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.52e-06,
-      "loss": 4.4367,
+      "epoch": 0.22,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 4.4051,
       "step": 130
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.5400000000000002e-06,
-      "loss": 4.3729,
+      "epoch": 0.22,
+      "learning_rate": 2.56e-06,
+      "loss": 4.3441,
       "step": 131
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.56e-06,
-      "loss": 4.3179,
+      "epoch": 0.22,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 4.2955,
       "step": 132
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.5800000000000003e-06,
-      "loss": 4.499,
+      "epoch": 0.22,
+      "learning_rate": 2.6e-06,
+      "loss": 4.3969,
       "step": 133
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.6e-06,
-      "loss": 4.3976,
+      "epoch": 0.23,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 4.402,
       "step": 134
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.6200000000000003e-06,
-      "loss": 4.5379,
+      "epoch": 0.23,
+      "learning_rate": 2.64e-06,
+      "loss": 4.4528,
       "step": 135
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.64e-06,
-      "loss": 4.193,
+      "epoch": 0.23,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 4.3123,
       "step": 136
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.6600000000000004e-06,
-      "loss": 4.4928,
+      "epoch": 0.23,
+      "learning_rate": 2.68e-06,
+      "loss": 4.3315,
       "step": 137
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.68e-06,
-      "loss": 4.4997,
+      "epoch": 0.23,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 4.329,
       "step": 138
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7000000000000004e-06,
-      "loss": 4.2407,
+      "epoch": 0.23,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 4.3091,
       "step": 139
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7200000000000002e-06,
-      "loss": 4.4741,
+      "epoch": 0.24,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 4.5347,
       "step": 140
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7400000000000004e-06,
-      "loss": 4.5195,
+      "epoch": 0.24,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 4.453,
       "step": 141
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7600000000000003e-06,
-      "loss": 4.3883,
+      "epoch": 0.24,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 4.4934,
       "step": 142
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7800000000000005e-06,
-      "loss": 4.246,
+      "epoch": 0.24,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.5347,
       "step": 143
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.6217,
+      "epoch": 0.24,
+      "learning_rate": 2.82e-06,
+      "loss": 4.3281,
       "step": 144
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.82e-06,
-      "loss": 4.3828,
+      "epoch": 0.24,
+      "learning_rate": 2.84e-06,
+      "loss": 4.4223,
       "step": 145
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.84e-06,
-      "loss": 4.6801,
+      "epoch": 0.25,
+      "learning_rate": 2.86e-06,
+      "loss": 4.7379,
       "step": 146
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.86e-06,
-      "loss": 4.444,
+      "epoch": 0.25,
+      "learning_rate": 2.88e-06,
+      "loss": 4.7251,
       "step": 147
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.88e-06,
-      "loss": 4.2831,
+      "epoch": 0.25,
+      "learning_rate": 2.9e-06,
+      "loss": 4.7122,
       "step": 148
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.9e-06,
-      "loss": 4.5017,
+      "epoch": 0.25,
+      "learning_rate": 2.92e-06,
+      "loss": 4.6427,
       "step": 149
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.92e-06,
-      "loss": 4.3597,
+      "epoch": 0.25,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.7168,
       "step": 150
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.3668,
+      "epoch": 0.25,
+      "learning_rate": 2.96e-06,
+      "loss": 4.3009,
       "step": 151
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.96e-06,
-      "loss": 4.2143,
+      "epoch": 0.26,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 4.2353,
       "step": 152
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.9800000000000003e-06,
-      "loss": 4.3089,
+      "epoch": 0.26,
+      "learning_rate": 3e-06,
+      "loss": 4.2273,
       "step": 153
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3e-06,
-      "loss": 4.3931,
+      "epoch": 0.26,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 4.1507,
       "step": 154
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.0200000000000003e-06,
-      "loss": 4.3049,
+      "epoch": 0.26,
+      "learning_rate": 3.04e-06,
+      "loss": 4.1878,
       "step": 155
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.04e-06,
-      "loss": 4.1502,
+      "epoch": 0.26,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 4.1888,
       "step": 156
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.0600000000000003e-06,
-      "loss": 4.2988,
+      "epoch": 0.26,
+      "learning_rate": 3.08e-06,
+      "loss": 4.3586,
       "step": 157
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.08e-06,
-      "loss": 4.505,
+      "epoch": 0.27,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 4.1796,
       "step": 158
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.1000000000000004e-06,
-      "loss": 4.2524,
+      "epoch": 0.27,
+      "learning_rate": 3.12e-06,
+      "loss": 4.1439,
       "step": 159
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.12e-06,
-      "loss": 4.0503,
+      "epoch": 0.27,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 4.2612,
       "step": 160
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.1400000000000004e-06,
-      "loss": 4.3716,
+      "epoch": 0.27,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 4.3082,
       "step": 161
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.1600000000000002e-06,
-      "loss": 4.3695,
+      "epoch": 0.27,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 4.1563,
       "step": 162
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.1800000000000005e-06,
-      "loss": 4.264,
+      "epoch": 0.27,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.2179,
       "step": 163
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.0612,
+      "epoch": 0.28,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 4.2514,
       "step": 164
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2200000000000005e-06,
-      "loss": 4.295,
+      "epoch": 0.28,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 4.2873,
       "step": 165
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2400000000000003e-06,
-      "loss": 4.0107,
+      "epoch": 0.28,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 4.3925,
       "step": 166
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2600000000000006e-06,
-      "loss": 4.3229,
+      "epoch": 0.28,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 4.2303,
       "step": 167
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2800000000000004e-06,
-      "loss": 4.1692,
+      "epoch": 0.28,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 4.4668,
       "step": 168
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.3000000000000006e-06,
-      "loss": 4.4363,
+      "epoch": 0.28,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 4.2762,
       "step": 169
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.3200000000000004e-06,
-      "loss": 4.2784,
+      "epoch": 0.29,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 4.3279,
       "step": 170
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.3400000000000006e-06,
-      "loss": 4.3623,
+      "epoch": 0.29,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 4.1924,
       "step": 171
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.3600000000000004e-06,
-      "loss": 4.1539,
+      "epoch": 0.29,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 4.1893,
       "step": 172
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.3800000000000007e-06,
-      "loss": 4.1177,
+      "epoch": 0.29,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.2197,
       "step": 173
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.4365,
+      "epoch": 0.29,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 4.1982,
       "step": 174
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.4200000000000007e-06,
-      "loss": 4.3485,
+      "epoch": 0.29,
+      "learning_rate": 3.44e-06,
+      "loss": 4.3419,
       "step": 175
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.44e-06,
-      "loss": 4.3067,
+      "epoch": 0.3,
+      "learning_rate": 3.46e-06,
+      "loss": 4.3705,
       "step": 176
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.46e-06,
-      "loss": 4.4658,
+      "epoch": 0.3,
+      "learning_rate": 3.48e-06,
+      "loss": 4.3055,
       "step": 177
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.48e-06,
-      "loss": 4.2413,
+      "epoch": 0.3,
+      "learning_rate": 3.5e-06,
+      "loss": 4.1993,
       "step": 178
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.5e-06,
-      "loss": 4.2862,
+      "epoch": 0.3,
+      "learning_rate": 3.52e-06,
+      "loss": 4.3358,
       "step": 179
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.52e-06,
-      "loss": 4.4141,
+      "epoch": 0.3,
+      "learning_rate": 3.54e-06,
+      "loss": 4.1873,
       "step": 180
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.54e-06,
-      "loss": 4.3774,
+      "epoch": 0.3,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 4.2644,
       "step": 181
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.5600000000000002e-06,
-      "loss": 4.4269,
+      "epoch": 0.31,
+      "learning_rate": 3.58e-06,
+      "loss": 4.2642,
       "step": 182
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.58e-06,
-      "loss": 4.2487,
+      "epoch": 0.31,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 4.2311,
       "step": 183
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.6000000000000003e-06,
-      "loss": 4.3575,
+      "epoch": 0.31,
+      "learning_rate": 3.62e-06,
+      "loss": 4.1317,
       "step": 184
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.62e-06,
-      "loss": 4.1065,
+      "epoch": 0.31,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 4.4816,
       "step": 185
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.6400000000000003e-06,
-      "loss": 4.2595,
+      "epoch": 0.31,
+      "learning_rate": 3.66e-06,
+      "loss": 4.268,
       "step": 186
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.66e-06,
-      "loss": 4.5998,
+      "epoch": 0.31,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 4.3489,
       "step": 187
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.6800000000000003e-06,
-      "loss": 4.3651,
+      "epoch": 0.32,
+      "learning_rate": 3.7e-06,
+      "loss": 4.3619,
       "step": 188
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.7e-06,
-      "loss": 4.4821,
+      "epoch": 0.32,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 4.5345,
       "step": 189
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.7200000000000004e-06,
-      "loss": 4.1233,
+      "epoch": 0.32,
+      "learning_rate": 3.74e-06,
+      "loss": 4.2482,
       "step": 190
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.74e-06,
-      "loss": 4.3625,
-      "step": 191
+      "epoch": 0.32,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 4.5152,
+      "step": 191
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.7600000000000004e-06,
-      "loss": 4.569,
+      "epoch": 0.32,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.4431,
       "step": 192
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.5782,
+      "epoch": 0.32,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 4.3379,
       "step": 193
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.8000000000000005e-06,
-      "loss": 4.3513,
+      "epoch": 0.33,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 4.4033,
       "step": 194
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.820000000000001e-06,
-      "loss": 4.5295,
+      "epoch": 0.33,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.3606,
       "step": 195
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.4018,
+      "epoch": 0.33,
+      "learning_rate": 3.86e-06,
+      "loss": 4.2561,
       "step": 196
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.86e-06,
-      "loss": 4.5647,
+      "epoch": 0.33,
+      "learning_rate": 3.88e-06,
+      "loss": 4.6605,
       "step": 197
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.88e-06,
-      "loss": 4.6384,
+      "epoch": 0.33,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 4.6509,
       "step": 198
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.900000000000001e-06,
-      "loss": 4.588,
+      "epoch": 0.33,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 4.5575,
       "step": 199
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.920000000000001e-06,
-      "loss": 4.3682,
+      "epoch": 0.34,
+      "learning_rate": 3.94e-06,
+      "loss": 4.7293,
       "step": 200
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.94e-06,
-      "loss": 4.3844,
+      "epoch": 0.34,
+      "learning_rate": 3.96e-06,
+      "loss": 4.119,
       "step": 201
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.96e-06,
-      "loss": 4.2155,
+      "epoch": 0.34,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 4.2329,
       "step": 202
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.980000000000001e-06,
-      "loss": 4.2721,
+      "epoch": 0.34,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.2809,
       "step": 203
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.2226,
+      "epoch": 0.34,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.1888,
       "step": 204
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.1393,
+      "epoch": 0.34,
+      "learning_rate": 4.04e-06,
+      "loss": 4.2201,
       "step": 205
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.04e-06,
-      "loss": 4.2505,
+      "epoch": 0.35,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 4.1575,
       "step": 206
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.060000000000001e-06,
-      "loss": 4.3549,
+      "epoch": 0.35,
+      "learning_rate": 4.08e-06,
+      "loss": 4.2356,
       "step": 207
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.08e-06,
-      "loss": 4.5281,
+      "epoch": 0.35,
+      "learning_rate": 4.1e-06,
+      "loss": 4.1999,
       "step": 208
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.1e-06,
-      "loss": 4.4086,
+      "epoch": 0.35,
+      "learning_rate": 4.12e-06,
+      "loss": 4.2149,
       "step": 209
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.12e-06,
-      "loss": 4.4068,
+      "epoch": 0.35,
+      "learning_rate": 4.14e-06,
+      "loss": 4.3318,
       "step": 210
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.14e-06,
-      "loss": 4.2584,
+      "epoch": 0.35,
+      "learning_rate": 4.16e-06,
+      "loss": 4.299,
       "step": 211
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.16e-06,
-      "loss": 4.3476,
+      "epoch": 0.36,
+      "learning_rate": 4.18e-06,
+      "loss": 4.2981,
       "step": 212
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.18e-06,
-      "loss": 4.2793,
+      "epoch": 0.36,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.2578,
       "step": 213
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.2831,
+      "epoch": 0.36,
+      "learning_rate": 4.22e-06,
+      "loss": 4.353,
       "step": 214
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.22e-06,
-      "loss": 4.3532,
+      "epoch": 0.36,
+      "learning_rate": 4.24e-06,
+      "loss": 4.2134,
       "step": 215
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.24e-06,
-      "loss": 4.3104,
+      "epoch": 0.36,
+      "learning_rate": 4.26e-06,
+      "loss": 4.24,
       "step": 216
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.26e-06,
-      "loss": 4.399,
+      "epoch": 0.36,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 4.1222,
       "step": 217
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.2800000000000005e-06,
-      "loss": 4.2388,
+      "epoch": 0.37,
+      "learning_rate": 4.3e-06,
+      "loss": 4.2403,
       "step": 218
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.3e-06,
-      "loss": 3.9632,
+      "epoch": 0.37,
+      "learning_rate": 4.32e-06,
+      "loss": 4.2177,
       "step": 219
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.32e-06,
-      "loss": 4.2871,
+      "epoch": 0.37,
+      "learning_rate": 4.34e-06,
+      "loss": 4.3008,
       "step": 220
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.34e-06,
-      "loss": 4.2568,
+      "epoch": 0.37,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 4.3402,
       "step": 221
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.360000000000001e-06,
-      "loss": 4.566,
+      "epoch": 0.37,
+      "learning_rate": 4.38e-06,
+      "loss": 4.3275,
       "step": 222
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.38e-06,
-      "loss": 4.3124,
+      "epoch": 0.37,
+      "learning_rate": 4.4e-06,
+      "loss": 4.1779,
       "step": 223
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.4e-06,
-      "loss": 4.3912,
+      "epoch": 0.38,
+      "learning_rate": 4.42e-06,
+      "loss": 4.2863,
       "step": 224
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.42e-06,
-      "loss": 4.1658,
+      "epoch": 0.38,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 4.4608,
       "step": 225
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.440000000000001e-06,
-      "loss": 4.109,
+      "epoch": 0.38,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 4.2403,
       "step": 226
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.4600000000000005e-06,
-      "loss": 4.2459,
+      "epoch": 0.38,
+      "learning_rate": 4.48e-06,
+      "loss": 4.1853,
       "step": 227
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.48e-06,
-      "loss": 4.3081,
+      "epoch": 0.38,
+      "learning_rate": 4.5e-06,
+      "loss": 4.2403,
       "step": 228
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.5e-06,
-      "loss": 4.1719,
+      "epoch": 0.39,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 4.3736,
       "step": 229
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.520000000000001e-06,
-      "loss": 4.3651,
+      "epoch": 0.39,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 4.3332,
       "step": 230
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.540000000000001e-06,
-      "loss": 4.3436,
+      "epoch": 0.39,
+      "learning_rate": 4.56e-06,
+      "loss": 4.2064,
       "step": 231
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.56e-06,
-      "loss": 4.3034,
+      "epoch": 0.39,
+      "learning_rate": 4.58e-06,
+      "loss": 4.2761,
       "step": 232
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.58e-06,
-      "loss": 4.2627,
+      "epoch": 0.39,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 4.3082,
       "step": 233
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.600000000000001e-06,
-      "loss": 4.2283,
+      "epoch": 0.39,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 4.2818,
       "step": 234
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.620000000000001e-06,
-      "loss": 4.2872,
+      "epoch": 0.4,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 4.4403,
       "step": 235
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.6400000000000005e-06,
-      "loss": 4.4579,
+      "epoch": 0.4,
+      "learning_rate": 4.66e-06,
+      "loss": 4.2043,
       "step": 236
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.66e-06,
-      "loss": 4.1832,
+      "epoch": 0.4,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 4.2007,
       "step": 237
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.680000000000001e-06,
-      "loss": 4.3195,
+      "epoch": 0.4,
+      "learning_rate": 4.7e-06,
+      "loss": 4.2285,
       "step": 238
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.7e-06,
-      "loss": 4.2828,
+      "epoch": 0.4,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 4.1442,
       "step": 239
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.7200000000000005e-06,
-      "loss": 4.341,
+      "epoch": 0.4,
+      "learning_rate": 4.74e-06,
+      "loss": 4.3297,
       "step": 240
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.74e-06,
-      "loss": 4.4422,
+      "epoch": 0.41,
+      "learning_rate": 4.76e-06,
+      "loss": 4.253,
       "step": 241
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.76e-06,
-      "loss": 4.272,
+      "epoch": 0.41,
+      "learning_rate": 4.78e-06,
+      "loss": 4.2696,
       "step": 242
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.78e-06,
-      "loss": 4.4976,
+      "epoch": 0.41,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.2858,
       "step": 243
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.3036,
+      "epoch": 0.41,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 4.3687,
       "step": 244
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.8200000000000004e-06,
-      "loss": 4.3239,
+      "epoch": 0.41,
+      "learning_rate": 4.84e-06,
+      "loss": 4.4333,
       "step": 245
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.84e-06,
-      "loss": 4.2632,
+      "epoch": 0.41,
+      "learning_rate": 4.86e-06,
+      "loss": 4.476,
       "step": 246
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.86e-06,
-      "loss": 4.818,
+      "epoch": 0.42,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 4.4581,
       "step": 247
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 4.469,
+      "epoch": 0.42,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 4.5935,
       "step": 248
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 4.3937,
+      "epoch": 0.42,
+      "learning_rate": 4.92e-06,
+      "loss": 4.5562,
       "step": 249
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.92e-06,
-      "loss": 4.4808,
+      "epoch": 0.42,
+      "learning_rate": 4.94e-06,
+      "loss": 4.5708,
       "step": 250
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.94e-06,
-      "loss": 4.3932,
+      "epoch": 0.42,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 4.2715,
       "step": 251
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.960000000000001e-06,
-      "loss": 4.0489,
+      "epoch": 0.42,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.2646,
       "step": 252
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.3453,
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 4.2305,
       "step": 253
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5e-06,
-      "loss": 4.077,
+      "epoch": 0.43,
+      "learning_rate": 5.02e-06,
+      "loss": 4.1202,
       "step": 254
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.02e-06,
-      "loss": 4.0752,
+      "epoch": 0.43,
+      "learning_rate": 5.04e-06,
+      "loss": 4.2932,
       "step": 255
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.04e-06,
-      "loss": 4.2665,
+      "epoch": 0.43,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 4.2446,
       "step": 256
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.060000000000001e-06,
-      "loss": 4.0714,
+      "epoch": 0.43,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 4.1774,
       "step": 257
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.0800000000000005e-06,
-      "loss": 4.5038,
+      "epoch": 0.43,
+      "learning_rate": 5.1e-06,
+      "loss": 4.203,
       "step": 258
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.1e-06,
-      "loss": 4.3117,
+      "epoch": 0.44,
+      "learning_rate": 5.12e-06,
+      "loss": 4.2629,
       "step": 259
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.12e-06,
-      "loss": 4.2644,
+      "epoch": 0.44,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 4.2674,
       "step": 260
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.140000000000001e-06,
-      "loss": 4.2386,
+      "epoch": 0.44,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 4.2434,
       "step": 261
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.1600000000000006e-06,
-      "loss": 4.2262,
+      "epoch": 0.44,
+      "learning_rate": 5.18e-06,
+      "loss": 4.1899,
       "step": 262
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.18e-06,
-      "loss": 4.3571,
+      "epoch": 0.44,
+      "learning_rate": 5.2e-06,
+      "loss": 4.2941,
       "step": 263
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.2e-06,
-      "loss": 4.3429,
+      "epoch": 0.44,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 4.0175,
       "step": 264
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.220000000000001e-06,
-      "loss": 4.2446,
+      "epoch": 0.45,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 4.1044,
       "step": 265
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.240000000000001e-06,
-      "loss": 4.3092,
+      "epoch": 0.45,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 4.2199,
       "step": 266
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.2600000000000005e-06,
-      "loss": 4.1785,
+      "epoch": 0.45,
+      "learning_rate": 5.28e-06,
+      "loss": 4.2392,
       "step": 267
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.28e-06,
-      "loss": 4.2174,
+      "epoch": 0.45,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 4.1779,
       "step": 268
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.300000000000001e-06,
-      "loss": 4.2146,
+      "epoch": 0.45,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 4.2911,
       "step": 269
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.320000000000001e-06,
-      "loss": 4.1565,
+      "epoch": 0.45,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 4.2804,
       "step": 270
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.3400000000000005e-06,
-      "loss": 4.4837,
+      "epoch": 0.46,
+      "learning_rate": 5.36e-06,
+      "loss": 4.2093,
       "step": 271
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.36e-06,
-      "loss": 4.1145,
+      "epoch": 0.46,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 4.3397,
       "step": 272
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.380000000000001e-06,
-      "loss": 4.2404,
+      "epoch": 0.46,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 4.2085,
       "step": 273
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.400000000000001e-06,
-      "loss": 4.2718,
+      "epoch": 0.46,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 4.2158,
       "step": 274
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.420000000000001e-06,
-      "loss": 4.3222,
+      "epoch": 0.46,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 4.2876,
       "step": 275
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.4400000000000004e-06,
-      "loss": 4.0561,
+      "epoch": 0.46,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 4.2118,
       "step": 276
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.460000000000001e-06,
-      "loss": 4.0345,
+      "epoch": 0.47,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 4.3481,
       "step": 277
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.480000000000001e-06,
-      "loss": 4.21,
+      "epoch": 0.47,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 4.1363,
       "step": 278
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 4.3329,
+      "epoch": 0.47,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 4.3209,
       "step": 279
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.5200000000000005e-06,
-      "loss": 4.1826,
+      "epoch": 0.47,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 4.3781,
       "step": 280
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.540000000000001e-06,
-      "loss": 4.216,
+      "epoch": 0.47,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 4.3857,
       "step": 281
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.560000000000001e-06,
-      "loss": 4.3664,
+      "epoch": 0.47,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 4.2594,
       "step": 282
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.580000000000001e-06,
-      "loss": 4.331,
+      "epoch": 0.48,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.161,
       "step": 283
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.1788,
+      "epoch": 0.48,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 4.1539,
       "step": 284
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.620000000000001e-06,
-      "loss": 4.3907,
+      "epoch": 0.48,
+      "learning_rate": 5.64e-06,
+      "loss": 4.4102,
       "step": 285
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.64e-06,
-      "loss": 4.2185,
+      "epoch": 0.48,
+      "learning_rate": 5.66e-06,
+      "loss": 4.166,
       "step": 286
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.66e-06,
-      "loss": 4.1472,
+      "epoch": 0.48,
+      "learning_rate": 5.68e-06,
+      "loss": 4.5148,
       "step": 287
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.68e-06,
-      "loss": 4.2194,
+      "epoch": 0.48,
+      "learning_rate": 5.7e-06,
+      "loss": 4.0003,
       "step": 288
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.7e-06,
-      "loss": 4.4897,
+      "epoch": 0.49,
+      "learning_rate": 5.72e-06,
+      "loss": 4.2587,
       "step": 289
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.72e-06,
-      "loss": 4.3876,
+      "epoch": 0.49,
+      "learning_rate": 5.74e-06,
+      "loss": 4.2268,
       "step": 290
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.74e-06,
-      "loss": 4.3521,
+      "epoch": 0.49,
+      "learning_rate": 5.76e-06,
+      "loss": 4.4759,
       "step": 291
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.76e-06,
-      "loss": 4.6181,
+      "epoch": 0.49,
+      "learning_rate": 5.78e-06,
+      "loss": 4.1496,
       "step": 292
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.78e-06,
-      "loss": 4.1318,
+      "epoch": 0.49,
+      "learning_rate": 5.8e-06,
+      "loss": 4.3539,
       "step": 293
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.8e-06,
-      "loss": 4.4907,
+      "epoch": 0.49,
+      "learning_rate": 5.82e-06,
+      "loss": 4.3933,
       "step": 294
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.82e-06,
-      "loss": 4.4207,
+      "epoch": 0.5,
+      "learning_rate": 5.84e-06,
+      "loss": 4.3932,
       "step": 295
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.84e-06,
-      "loss": 4.9954,
+      "epoch": 0.5,
+      "learning_rate": 5.86e-06,
+      "loss": 4.2008,
       "step": 296
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.86e-06,
-      "loss": 4.3187,
+      "epoch": 0.5,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.2785,
       "step": 297
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.9162,
+      "epoch": 0.5,
+      "learning_rate": 5.9e-06,
+      "loss": 4.4644,
       "step": 298
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.9e-06,
-      "loss": 4.5678,
+      "epoch": 0.5,
+      "learning_rate": 5.92e-06,
+      "loss": 4.4931,
       "step": 299
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.92e-06,
-      "loss": 4.7455,
+      "epoch": 0.5,
+      "learning_rate": 5.94e-06,
+      "loss": 4.7331,
       "step": 300
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.94e-06,
-      "loss": 4.04,
+      "epoch": 0.51,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 4.3267,
       "step": 301
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.9600000000000005e-06,
-      "loss": 4.1705,
+      "epoch": 0.51,
+      "learning_rate": 5.98e-06,
+      "loss": 4.2696,
       "step": 302
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.98e-06,
-      "loss": 4.1995,
+      "epoch": 0.51,
+      "learning_rate": 6e-06,
+      "loss": 4.3485,
       "step": 303
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6e-06,
-      "loss": 3.8708,
+      "epoch": 0.51,
+      "learning_rate": 6.02e-06,
+      "loss": 4.1293,
       "step": 304
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.02e-06,
-      "loss": 4.051,
+      "epoch": 0.51,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 4.3256,
       "step": 305
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.040000000000001e-06,
-      "loss": 4.0563,
+      "epoch": 0.51,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.313,
       "step": 306
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 4.259,
+      "epoch": 0.52,
+      "learning_rate": 6.08e-06,
+      "loss": 4.1031,
       "step": 307
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.08e-06,
-      "loss": 3.974,
+      "epoch": 0.52,
+      "learning_rate": 6.1e-06,
+      "loss": 4.3191,
       "step": 308
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.1e-06,
-      "loss": 3.9649,
+      "epoch": 0.52,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 4.2111,
       "step": 309
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.120000000000001e-06,
-      "loss": 4.1752,
+      "epoch": 0.52,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 4.3018,
       "step": 310
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.1400000000000005e-06,
-      "loss": 4.1021,
+      "epoch": 0.52,
+      "learning_rate": 6.16e-06,
+      "loss": 4.1658,
       "step": 311
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.16e-06,
-      "loss": 4.1044,
+      "epoch": 0.52,
+      "learning_rate": 6.18e-06,
+      "loss": 4.2064,
       "step": 312
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.18e-06,
-      "loss": 4.0725,
+      "epoch": 0.53,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 4.2251,
       "step": 313
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.200000000000001e-06,
-      "loss": 4.1376,
+      "epoch": 0.53,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 4.2262,
       "step": 314
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.220000000000001e-06,
-      "loss": 4.0619,
+      "epoch": 0.53,
+      "learning_rate": 6.24e-06,
+      "loss": 4.2471,
       "step": 315
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.24e-06,
-      "loss": 4.413,
+      "epoch": 0.53,
+      "learning_rate": 6.26e-06,
+      "loss": 4.2381,
       "step": 316
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.26e-06,
-      "loss": 4.3625,
+      "epoch": 0.53,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 4.1972,
       "step": 317
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.280000000000001e-06,
-      "loss": 4.2863,
+      "epoch": 0.53,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 4.2808,
       "step": 318
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.300000000000001e-06,
-      "loss": 4.1068,
+      "epoch": 0.54,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 4.1035,
       "step": 319
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.3200000000000005e-06,
-      "loss": 4.2889,
+      "epoch": 0.54,
+      "learning_rate": 6.34e-06,
+      "loss": 4.265,
       "step": 320
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.34e-06,
-      "loss": 4.1188,
+      "epoch": 0.54,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 4.1205,
       "step": 321
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.360000000000001e-06,
-      "loss": 4.1358,
+      "epoch": 0.54,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 4.1692,
       "step": 322
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.380000000000001e-06,
-      "loss": 4.0802,
+      "epoch": 0.54,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.1871,
       "step": 323
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.2859,
+      "epoch": 0.54,
+      "learning_rate": 6.42e-06,
+      "loss": 4.2543,
       "step": 324
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.42e-06,
-      "loss": 4.2048,
+      "epoch": 0.55,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 4.2072,
       "step": 325
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.440000000000001e-06,
-      "loss": 4.3831,
+      "epoch": 0.55,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 4.1326,
       "step": 326
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.460000000000001e-06,
-      "loss": 4.3815,
+      "epoch": 0.55,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 4.4151,
       "step": 327
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.480000000000001e-06,
-      "loss": 4.2369,
+      "epoch": 0.55,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 4.1615,
       "step": 328
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 4.1473,
+      "epoch": 0.55,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 4.0782,
       "step": 329
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.520000000000001e-06,
-      "loss": 4.3793,
+      "epoch": 0.55,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 4.3004,
       "step": 330
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.540000000000001e-06,
-      "loss": 4.0783,
+      "epoch": 0.56,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 4.1869,
       "step": 331
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.560000000000001e-06,
-      "loss": 4.2146,
+      "epoch": 0.56,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 4.137,
       "step": 332
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.5800000000000005e-06,
-      "loss": 4.2755,
+      "epoch": 0.56,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 4.1984,
       "step": 333
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.600000000000001e-06,
-      "loss": 4.0445,
+      "epoch": 0.56,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 4.2389,
       "step": 334
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.620000000000001e-06,
-      "loss": 4.2928,
+      "epoch": 0.56,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 4.2559,
       "step": 335
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.640000000000001e-06,
-      "loss": 4.3136,
+      "epoch": 0.56,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.1806,
       "step": 336
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 4.2456,
+      "epoch": 0.57,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 4.1488,
       "step": 337
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.680000000000001e-06,
-      "loss": 4.4271,
+      "epoch": 0.57,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 4.2468,
       "step": 338
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.700000000000001e-06,
-      "loss": 4.4132,
+      "epoch": 0.57,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 4.2594,
       "step": 339
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.720000000000001e-06,
-      "loss": 4.2516,
+      "epoch": 0.57,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 4.287,
       "step": 340
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.740000000000001e-06,
-      "loss": 4.3111,
+      "epoch": 0.57,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 4.2912,
       "step": 341
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.760000000000001e-06,
-      "loss": 4.4282,
+      "epoch": 0.58,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 4.3932,
       "step": 342
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.780000000000001e-06,
-      "loss": 4.3373,
+      "epoch": 0.58,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.418,
       "step": 343
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.4456,
+      "epoch": 0.58,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 4.2976,
       "step": 344
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.820000000000001e-06,
-      "loss": 4.2956,
+      "epoch": 0.58,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 4.4477,
       "step": 345
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.8400000000000014e-06,
-      "loss": 4.2442,
+      "epoch": 0.58,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 4.1469,
       "step": 346
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.860000000000001e-06,
-      "loss": 4.4597,
+      "epoch": 0.58,
+      "learning_rate": 6.88e-06,
+      "loss": 4.2839,
       "step": 347
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.88e-06,
-      "loss": 4.6718,
+      "epoch": 0.59,
+      "learning_rate": 6.9e-06,
+      "loss": 4.6134,
       "step": 348
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.9e-06,
-      "loss": 4.8422,
+      "epoch": 0.59,
+      "learning_rate": 6.92e-06,
+      "loss": 4.4709,
       "step": 349
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.92e-06,
-      "loss": 4.4774,
+      "epoch": 0.59,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 4.7728,
       "step": 350
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 6.9400000000000005e-06,
-      "loss": 4.4637,
+      "epoch": 0.59,
+      "learning_rate": 6.96e-06,
+      "loss": 4.1291,
       "step": 351
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 6.96e-06,
-      "loss": 4.157,
+      "epoch": 0.59,
+      "learning_rate": 6.98e-06,
+      "loss": 4.237,
       "step": 352
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 6.98e-06,
-      "loss": 4.1457,
+      "epoch": 0.59,
+      "learning_rate": 7e-06,
+      "loss": 4.2614,
       "step": 353
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7e-06,
-      "loss": 4.3418,
+      "epoch": 0.6,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.1032,
       "step": 354
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 4.1486,
+      "epoch": 0.6,
+      "learning_rate": 7.04e-06,
+      "loss": 4.1884,
       "step": 355
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.04e-06,
-      "loss": 4.1662,
+      "epoch": 0.6,
+      "learning_rate": 7.06e-06,
+      "loss": 4.2002,
       "step": 356
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.06e-06,
-      "loss": 4.3384,
+      "epoch": 0.6,
+      "learning_rate": 7.08e-06,
+      "loss": 4.2122,
       "step": 357
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.08e-06,
-      "loss": 4.0699,
+      "epoch": 0.6,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 4.1464,
       "step": 358
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.100000000000001e-06,
-      "loss": 4.238,
+      "epoch": 0.6,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 4.1748,
       "step": 359
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.1200000000000004e-06,
-      "loss": 4.3706,
+      "epoch": 0.61,
+      "learning_rate": 7.14e-06,
+      "loss": 4.1205,
       "step": 360
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.14e-06,
-      "loss": 4.2257,
+      "epoch": 0.61,
+      "learning_rate": 7.16e-06,
+      "loss": 4.2075,
       "step": 361
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.16e-06,
-      "loss": 4.0899,
+      "epoch": 0.61,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 4.2108,
       "step": 362
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.180000000000001e-06,
-      "loss": 4.1032,
+      "epoch": 0.61,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 4.2577,
       "step": 363
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.2000000000000005e-06,
-      "loss": 4.2998,
+      "epoch": 0.61,
+      "learning_rate": 7.22e-06,
+      "loss": 4.1437,
       "step": 364
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.22e-06,
-      "loss": 4.2338,
+      "epoch": 0.61,
+      "learning_rate": 7.24e-06,
+      "loss": 4.2128,
       "step": 365
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.24e-06,
-      "loss": 4.1811,
+      "epoch": 0.62,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 4.1773,
       "step": 366
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.260000000000001e-06,
-      "loss": 4.0937,
+      "epoch": 0.62,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 4.179,
       "step": 367
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.280000000000001e-06,
-      "loss": 4.5385,
+      "epoch": 0.62,
+      "learning_rate": 7.3e-06,
+      "loss": 4.164,
       "step": 368
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.3e-06,
-      "loss": 4.3573,
-      "step": 369
-    },
-    {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 7.32e-06,
-      "loss": 4.2215,
-      "step": 370
+      "loss": 4.2615,
+      "step": 369
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 7.340000000000001e-06,
-      "loss": 4.1502,
-      "step": 371
+      "loss": 4.1449,
+      "step": 370
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 7.360000000000001e-06,
-      "loss": 4.0908,
-      "step": 372
+      "loss": 4.0527,
+      "step": 371
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 7.3800000000000005e-06,
-      "loss": 4.099,
-      "step": 373
+      "loss": 4.2189,
+      "step": 372
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 7.4e-06,
-      "loss": 4.1223,
-      "step": 374
+      "loss": 4.0695,
+      "step": 373
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 7.420000000000001e-06,
-      "loss": 4.173,
-      "step": 375
+      "loss": 4.0884,
+      "step": 374
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 7.440000000000001e-06,
-      "loss": 4.2195,
-      "step": 376
+      "loss": 4.2513,
+      "step": 375
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 7.4600000000000006e-06,
-      "loss": 4.1713,
-      "step": 377
+      "loss": 4.1336,
+      "step": 376
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.63,
       "learning_rate": 7.48e-06,
-      "loss": 4.1561,
-      "step": 378
+      "loss": 4.2286,
+      "step": 377
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 4.1195,
-      "step": 379
+      "loss": 4.2841,
+      "step": 378
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 7.520000000000001e-06,
-      "loss": 4.2634,
-      "step": 380
+      "loss": 4.1134,
+      "step": 379
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 7.540000000000001e-06,
-      "loss": 4.0967,
-      "step": 381
+      "loss": 4.337,
+      "step": 380
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 7.5600000000000005e-06,
-      "loss": 4.2816,
-      "step": 382
+      "loss": 4.2637,
+      "step": 381
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 7.58e-06,
-      "loss": 4.0563,
-      "step": 383
+      "loss": 4.1276,
+      "step": 382
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 7.600000000000001e-06,
-      "loss": 4.3362,
-      "step": 384
+      "loss": 4.0296,
+      "step": 383
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 7.620000000000001e-06,
-      "loss": 4.3688,
-      "step": 385
+      "loss": 4.2207,
+      "step": 384
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.65,
       "learning_rate": 7.640000000000001e-06,
-      "loss": 4.1055,
-      "step": 386
+      "loss": 4.4024,
+      "step": 385
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 7.660000000000001e-06,
-      "loss": 4.3171,
-      "step": 387
+      "loss": 4.1624,
+      "step": 386
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 7.680000000000001e-06,
-      "loss": 4.3495,
-      "step": 388
+      "loss": 4.3254,
+      "step": 387
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 7.7e-06,
-      "loss": 4.4034,
-      "step": 389
+      "loss": 4.2542,
+      "step": 388
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 7.72e-06,
-      "loss": 4.3698,
-      "step": 390
+      "loss": 4.1433,
+      "step": 389
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 7.74e-06,
-      "loss": 4.4185,
-      "step": 391
+      "loss": 4.1481,
+      "step": 390
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 7.76e-06,
-      "loss": 4.2773,
-      "step": 392
+      "loss": 4.3634,
+      "step": 391
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 7.78e-06,
-      "loss": 4.1896,
-      "step": 393
+      "loss": 4.2915,
+      "step": 392
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 7.800000000000002e-06,
-      "loss": 4.4255,
-      "step": 394
+      "loss": 4.5435,
+      "step": 393
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 7.820000000000001e-06,
-      "loss": 4.1546,
-      "step": 395
+      "loss": 4.3662,
+      "step": 394
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 7.840000000000001e-06,
-      "loss": 4.4318,
-      "step": 396
+      "loss": 4.2978,
+      "step": 395
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 7.860000000000001e-06,
-      "loss": 4.3534,
-      "step": 397
+      "loss": 4.2279,
+      "step": 396
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 7.88e-06,
-      "loss": 4.4576,
-      "step": 398
+      "loss": 4.4285,
+      "step": 397
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 7.9e-06,
-      "loss": 4.7461,
-      "step": 399
+      "loss": 4.6224,
+      "step": 398
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 7.92e-06,
-      "loss": 4.6628,
-      "step": 400
+      "loss": 4.4336,
+      "step": 399
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 7.94e-06,
-      "loss": 4.0486,
-      "step": 401
+      "loss": 4.5344,
+      "step": 400
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.67,
       "learning_rate": 7.960000000000002e-06,
-      "loss": 4.1799,
-      "step": 402
+      "loss": 4.2421,
+      "step": 401
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 7.980000000000002e-06,
-      "loss": 4.1841,
-      "step": 403
+      "loss": 4.1492,
+      "step": 402
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 4.2834,
-      "step": 404
+      "loss": 4.211,
+      "step": 403
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 8.020000000000001e-06,
-      "loss": 4.1536,
-      "step": 405
+      "loss": 4.1921,
+      "step": 404
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 8.040000000000001e-06,
-      "loss": 4.1527,
-      "step": 406
+      "loss": 4.1333,
+      "step": 405
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 8.06e-06,
-      "loss": 4.0915,
-      "step": 407
+      "loss": 4.1795,
+      "step": 406
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 8.08e-06,
-      "loss": 4.1758,
-      "step": 408
+      "loss": 4.1875,
+      "step": 407
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 8.1e-06,
-      "loss": 4.1054,
-      "step": 409
+      "loss": 4.1004,
+      "step": 408
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 8.120000000000002e-06,
-      "loss": 4.3544,
-      "step": 410
+      "loss": 4.1633,
+      "step": 409
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 8.14e-06,
-      "loss": 4.3043,
-      "step": 411
+      "loss": 4.1125,
+      "step": 410
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 8.16e-06,
-      "loss": 4.2693,
-      "step": 412
+      "loss": 4.1233,
+      "step": 411
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 8.18e-06,
-      "loss": 4.2359,
-      "step": 413
+      "loss": 4.2896,
+      "step": 412
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 8.2e-06,
-      "loss": 4.0849,
-      "step": 414
+      "loss": 4.1791,
+      "step": 413
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 8.220000000000001e-06,
-      "loss": 4.1823,
-      "step": 415
+      "loss": 4.2289,
+      "step": 414
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 8.24e-06,
-      "loss": 4.281,
-      "step": 416
+      "loss": 4.2061,
+      "step": 415
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 8.26e-06,
-      "loss": 3.9741,
-      "step": 417
+      "loss": 4.1171,
+      "step": 416
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 8.28e-06,
-      "loss": 4.0802,
-      "step": 418
+      "loss": 4.1973,
+      "step": 417
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 8.3e-06,
-      "loss": 4.2756,
-      "step": 419
+      "loss": 4.0318,
+      "step": 418
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 8.32e-06,
-      "loss": 4.2086,
-      "step": 420
+      "loss": 4.2931,
+      "step": 419
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 8.34e-06,
-      "loss": 4.1629,
-      "step": 421
+      "loss": 4.1999,
+      "step": 420
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.71,
       "learning_rate": 8.36e-06,
-      "loss": 4.2496,
-      "step": 422
+      "loss": 3.9695,
+      "step": 421
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 8.380000000000001e-06,
-      "loss": 4.2382,
-      "step": 423
+      "loss": 4.1124,
+      "step": 422
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 4.1612,
-      "step": 424
+      "loss": 4.2425,
+      "step": 423
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 8.42e-06,
-      "loss": 4.4546,
-      "step": 425
+      "loss": 4.1725,
+      "step": 424
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 8.44e-06,
-      "loss": 4.2347,
-      "step": 426
+      "loss": 4.166,
+      "step": 425
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 8.46e-06,
-      "loss": 4.1191,
-      "step": 427
+      "loss": 4.2335,
+      "step": 426
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 8.48e-06,
-      "loss": 4.1574,
-      "step": 428
+      "loss": 4.1832,
+      "step": 427
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 8.5e-06,
-      "loss": 4.4139,
-      "step": 429
+      "loss": 4.1219,
+      "step": 428
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 8.52e-06,
-      "loss": 4.3594,
-      "step": 430
+      "loss": 4.237,
+      "step": 429
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 8.540000000000001e-06,
-      "loss": 4.2223,
-      "step": 431
+      "loss": 4.1834,
+      "step": 430
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 8.560000000000001e-06,
-      "loss": 4.1815,
-      "step": 432
+      "loss": 4.2492,
+      "step": 431
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 8.580000000000001e-06,
-      "loss": 4.2792,
-      "step": 433
+      "loss": 4.0503,
+      "step": 432
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 8.6e-06,
-      "loss": 4.1954,
-      "step": 434
+      "loss": 4.2888,
+      "step": 433
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 8.62e-06,
-      "loss": 4.4341,
-      "step": 435
+      "loss": 4.2068,
+      "step": 434
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 8.64e-06,
-      "loss": 4.1195,
-      "step": 436
+      "loss": 4.2792,
+      "step": 435
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 8.66e-06,
-      "loss": 4.1726,
-      "step": 437
+      "loss": 4.1248,
+      "step": 436
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.73,
       "learning_rate": 8.68e-06,
-      "loss": 4.088,
-      "step": 438
+      "loss": 4.0674,
+      "step": 437
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 8.700000000000001e-06,
-      "loss": 4.3113,
-      "step": 439
+      "loss": 4.2741,
+      "step": 438
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 8.720000000000001e-06,
-      "loss": 4.2111,
-      "step": 440
+      "loss": 4.1833,
+      "step": 439
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 8.740000000000001e-06,
-      "loss": 4.1302,
-      "step": 441
+      "loss": 4.3683,
+      "step": 440
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 8.76e-06,
-      "loss": 4.2824,
-      "step": 442
+      "loss": 4.2469,
+      "step": 441
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 8.78e-06,
-      "loss": 4.2469,
-      "step": 443
+      "loss": 4.4203,
+      "step": 442
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 8.8e-06,
-      "loss": 4.3457,
-      "step": 444
+      "loss": 4.3338,
+      "step": 443
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 8.82e-06,
-      "loss": 4.2329,
-      "step": 445
+      "loss": 4.2684,
+      "step": 444
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.75,
       "learning_rate": 8.84e-06,
-      "loss": 4.5752,
-      "step": 446
+      "loss": 4.0929,
+      "step": 445
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 8.860000000000002e-06,
-      "loss": 4.5156,
-      "step": 447
+      "loss": 4.3938,
+      "step": 446
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 8.880000000000001e-06,
-      "loss": 4.4837,
-      "step": 448
+      "loss": 4.3174,
+      "step": 447
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 8.900000000000001e-06,
-      "loss": 4.5317,
-      "step": 449
+      "loss": 4.2561,
+      "step": 448
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 8.920000000000001e-06,
-      "loss": 4.591,
-      "step": 450
+      "loss": 4.444,
+      "step": 449
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 8.94e-06,
-      "loss": 4.1316,
-      "step": 451
+      "loss": 4.5666,
+      "step": 450
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 8.96e-06,
-      "loss": 4.0327,
-      "step": 452
+      "loss": 4.2167,
+      "step": 451
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 8.98e-06,
-      "loss": 4.1289,
-      "step": 453
+      "loss": 4.2266,
+      "step": 452
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 9e-06,
-      "loss": 4.2124,
-      "step": 454
+      "loss": 4.1614,
+      "step": 453
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 9.020000000000002e-06,
-      "loss": 4.0944,
-      "step": 455
+      "loss": 4.0323,
+      "step": 454
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 9.040000000000002e-06,
-      "loss": 4.1625,
-      "step": 456
+      "loss": 4.214,
+      "step": 455
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 9.060000000000001e-06,
-      "loss": 4.0958,
-      "step": 457
+      "loss": 4.152,
+      "step": 456
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 9.080000000000001e-06,
-      "loss": 4.3331,
-      "step": 458
+      "loss": 4.2426,
+      "step": 457
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 9.100000000000001e-06,
-      "loss": 4.0962,
-      "step": 459
+      "loss": 4.2681,
+      "step": 458
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 9.12e-06,
-      "loss": 4.2437,
-      "step": 460
+      "loss": 4.1891,
+      "step": 459
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.77,
       "learning_rate": 9.14e-06,
-      "loss": 4.3344,
-      "step": 461
+      "loss": 4.2,
+      "step": 460
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.16e-06,
-      "loss": 4.2095,
-      "step": 462
+      "loss": 4.1147,
+      "step": 461
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.180000000000002e-06,
-      "loss": 4.1975,
-      "step": 463
+      "loss": 4.1357,
+      "step": 462
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.200000000000002e-06,
-      "loss": 4.2624,
-      "step": 464
+      "loss": 4.1446,
+      "step": 463
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.220000000000002e-06,
-      "loss": 4.2855,
-      "step": 465
+      "loss": 4.2087,
+      "step": 464
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.240000000000001e-06,
-      "loss": 4.0756,
-      "step": 466
+      "loss": 4.2335,
+      "step": 465
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.260000000000001e-06,
-      "loss": 4.2049,
-      "step": 467
+      "loss": 4.1698,
+      "step": 466
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 9.280000000000001e-06,
-      "loss": 4.2952,
-      "step": 468
+      "loss": 4.1785,
+      "step": 467
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 9.3e-06,
-      "loss": 4.0674,
-      "step": 469
+      "loss": 4.1921,
+      "step": 468
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 9.32e-06,
-      "loss": 4.422,
-      "step": 470
+      "loss": 4.1838,
+      "step": 469
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 9.340000000000002e-06,
-      "loss": 4.2918,
-      "step": 471
+      "loss": 4.1536,
+      "step": 470
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 9.360000000000002e-06,
-      "loss": 4.311,
-      "step": 472
+      "loss": 4.081,
+      "step": 471
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 9.38e-06,
-      "loss": 3.9875,
-      "step": 473
+      "loss": 4.1874,
+      "step": 472
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.4e-06,
-      "loss": 4.3065,
-      "step": 474
+      "loss": 4.1796,
+      "step": 473
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.42e-06,
-      "loss": 4.3725,
-      "step": 475
+      "loss": 4.0716,
+      "step": 474
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.440000000000001e-06,
-      "loss": 4.2321,
-      "step": 476
+      "loss": 4.1163,
+      "step": 475
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.460000000000001e-06,
-      "loss": 4.0517,
-      "step": 477
+      "loss": 4.1431,
+      "step": 476
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.48e-06,
-      "loss": 4.1719,
-      "step": 478
+      "loss": 4.1631,
+      "step": 477
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.5e-06,
-      "loss": 4.2294,
-      "step": 479
+      "loss": 4.1923,
+      "step": 478
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 9.52e-06,
-      "loss": 4.2519,
-      "step": 480
+      "loss": 4.0699,
+      "step": 479
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 9.54e-06,
-      "loss": 4.1538,
-      "step": 481
+      "loss": 4.0712,
+      "step": 480
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 9.56e-06,
-      "loss": 4.2508,
-      "step": 482
+      "loss": 4.3093,
+      "step": 481
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 9.58e-06,
-      "loss": 4.2302,
-      "step": 483
+      "loss": 4.1374,
+      "step": 482
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2961,
-      "step": 484
+      "loss": 4.3051,
+      "step": 483
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 9.620000000000001e-06,
-      "loss": 4.2522,
-      "step": 485
+      "loss": 4.2504,
+      "step": 484
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.640000000000001e-06,
-      "loss": 4.2322,
-      "step": 486
+      "loss": 4.2339,
+      "step": 485
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.66e-06,
-      "loss": 4.0833,
-      "step": 487
+      "loss": 4.1411,
+      "step": 486
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.68e-06,
-      "loss": 4.1891,
-      "step": 488
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 9.7e-06,
-      "loss": 3.9655,
-      "step": 489
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 9.72e-06,
-      "loss": 4.2419,
-      "step": 490
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 9.74e-06,
-      "loss": 4.2586,
-      "step": 491
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 9.760000000000001e-06,
-      "loss": 4.1422,
-      "step": 492
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 4.3416,
-      "step": 493
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.3329,
-      "step": 494
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.820000000000001e-06,
-      "loss": 4.5124,
-      "step": 495
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.84e-06,
-      "loss": 4.3125,
-      "step": 496
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.86e-06,
-      "loss": 4.3506,
-      "step": 497
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.88e-06,
-      "loss": 4.5408,
-      "step": 498
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.9e-06,
-      "loss": 4.4512,
-      "step": 499
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.920000000000002e-06,
-      "loss": 4.4503,
-      "step": 500
-    },
-    {
-      "epoch": 0.42,
-      "eval_loss": 4.234867572784424,
-      "eval_runtime": 791.3294,
-      "eval_samples_per_second": 3.339,
-      "eval_steps_per_second": 0.279,
-      "eval_wer": 1.9679492264974217,
-      "step": 500
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.940000000000001e-06,
-      "loss": 4.1788,
-      "step": 501
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 4.3094,
-      "step": 502
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.980000000000001e-06,
-      "loss": 4.1493,
-      "step": 503
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 1e-05,
-      "loss": 3.9753,
-      "step": 504
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.985486211901307e-06,
-      "loss": 4.2008,
-      "step": 505
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.970972423802613e-06,
-      "loss": 4.3298,
-      "step": 506
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.95645863570392e-06,
-      "loss": 4.0534,
-      "step": 507
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.941944847605226e-06,
-      "loss": 4.1157,
-      "step": 508
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.927431059506532e-06,
-      "loss": 4.2461,
-      "step": 509
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.912917271407838e-06,
-      "loss": 4.4141,
-      "step": 510
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.898403483309144e-06,
-      "loss": 4.2255,
-      "step": 511
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.883889695210452e-06,
-      "loss": 4.0412,
-      "step": 512
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.869375907111757e-06,
-      "loss": 4.238,
-      "step": 513
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.854862119013063e-06,
-      "loss": 4.0405,
-      "step": 514
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.84034833091437e-06,
-      "loss": 4.0428,
-      "step": 515
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.825834542815675e-06,
-      "loss": 4.1141,
-      "step": 516
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.811320754716981e-06,
-      "loss": 4.1896,
-      "step": 517
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.79680696661829e-06,
-      "loss": 4.0757,
-      "step": 518
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.782293178519594e-06,
-      "loss": 4.5685,
-      "step": 519
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.7677793904209e-06,
-      "loss": 4.1416,
-      "step": 520
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.753265602322208e-06,
-      "loss": 4.2306,
-      "step": 521
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.738751814223512e-06,
-      "loss": 4.2293,
-      "step": 522
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.724238026124818e-06,
-      "loss": 4.2091,
-      "step": 523
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.709724238026126e-06,
-      "loss": 4.1854,
-      "step": 524
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.69521044992743e-06,
-      "loss": 4.224,
-      "step": 525
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.680696661828739e-06,
-      "loss": 4.0392,
-      "step": 526
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.666182873730045e-06,
-      "loss": 4.188,
-      "step": 527
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.651669085631351e-06,
-      "loss": 4.1156,
-      "step": 528
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.637155297532657e-06,
-      "loss": 4.3848,
-      "step": 529
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.622641509433963e-06,
-      "loss": 4.38,
-      "step": 530
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.60812772133527e-06,
-      "loss": 4.37,
-      "step": 531
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.593613933236576e-06,
-      "loss": 4.3093,
-      "step": 532
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.579100145137882e-06,
-      "loss": 4.2271,
-      "step": 533
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.564586357039188e-06,
-      "loss": 4.2586,
-      "step": 534
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.550072568940494e-06,
-      "loss": 4.1681,
-      "step": 535
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.5355587808418e-06,
-      "loss": 4.1042,
-      "step": 536
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.521044992743107e-06,
-      "loss": 4.2432,
-      "step": 537
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.506531204644413e-06,
-      "loss": 3.8428,
-      "step": 538
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.49201741654572e-06,
-      "loss": 4.3257,
-      "step": 539
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.477503628447025e-06,
-      "loss": 4.2554,
-      "step": 540
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.462989840348332e-06,
-      "loss": 4.38,
-      "step": 541
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.448476052249638e-06,
-      "loss": 4.2164,
-      "step": 542
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.433962264150944e-06,
-      "loss": 4.1753,
-      "step": 543
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.41944847605225e-06,
-      "loss": 4.3098,
-      "step": 544
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.404934687953556e-06,
-      "loss": 4.4283,
-      "step": 545
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.390420899854863e-06,
-      "loss": 4.3051,
-      "step": 546
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.375907111756169e-06,
-      "loss": 4.1567,
-      "step": 547
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.361393323657475e-06,
-      "loss": 4.4045,
-      "step": 548
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.346879535558781e-06,
-      "loss": 4.6211,
-      "step": 549
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.332365747460089e-06,
-      "loss": 4.568,
-      "step": 550
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.317851959361393e-06,
-      "loss": 4.3265,
-      "step": 551
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.3033381712627e-06,
-      "loss": 4.1286,
-      "step": 552
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.288824383164008e-06,
-      "loss": 4.1772,
-      "step": 553
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.274310595065312e-06,
-      "loss": 4.242,
-      "step": 554
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.259796806966618e-06,
-      "loss": 4.2894,
-      "step": 555
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.245283018867926e-06,
-      "loss": 4.1459,
-      "step": 556
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.230769230769232e-06,
-      "loss": 4.1136,
-      "step": 557
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.216255442670537e-06,
-      "loss": 4.1929,
-      "step": 558
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.201741654571845e-06,
-      "loss": 4.3735,
-      "step": 559
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.187227866473151e-06,
-      "loss": 4.0632,
-      "step": 560
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.172714078374455e-06,
-      "loss": 4.1948,
-      "step": 561
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.158200290275763e-06,
-      "loss": 4.0541,
-      "step": 562
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.14368650217707e-06,
-      "loss": 4.2505,
-      "step": 563
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.129172714078376e-06,
-      "loss": 3.9946,
-      "step": 564
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.114658925979682e-06,
-      "loss": 4.0665,
-      "step": 565
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.100145137880988e-06,
-      "loss": 4.2316,
-      "step": 566
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.085631349782294e-06,
-      "loss": 4.1467,
-      "step": 567
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.0711175616836e-06,
-      "loss": 4.173,
-      "step": 568
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.056603773584907e-06,
-      "loss": 4.0513,
-      "step": 569
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.042089985486213e-06,
-      "loss": 4.1592,
-      "step": 570
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.027576197387519e-06,
-      "loss": 4.1818,
-      "step": 571
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.013062409288825e-06,
-      "loss": 4.3169,
-      "step": 572
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.998548621190131e-06,
-      "loss": 4.1925,
-      "step": 573
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.984034833091438e-06,
-      "loss": 4.1541,
-      "step": 574
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.969521044992744e-06,
-      "loss": 4.272,
-      "step": 575
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.95500725689405e-06,
-      "loss": 4.329,
-      "step": 576
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.940493468795356e-06,
-      "loss": 4.3739,
-      "step": 577
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.925979680696662e-06,
-      "loss": 4.1541,
-      "step": 578
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.911465892597968e-06,
-      "loss": 4.2054,
-      "step": 579
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.896952104499275e-06,
-      "loss": 4.2794,
-      "step": 580
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.882438316400581e-06,
-      "loss": 4.427,
-      "step": 581
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.867924528301887e-06,
-      "loss": 3.9855,
-      "step": 582
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.853410740203193e-06,
-      "loss": 4.1427,
-      "step": 583
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.8388969521045e-06,
-      "loss": 4.1454,
-      "step": 584
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.824383164005806e-06,
-      "loss": 4.3482,
-      "step": 585
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.809869375907114e-06,
-      "loss": 4.3232,
-      "step": 586
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.795355587808418e-06,
-      "loss": 4.3642,
-      "step": 587
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.780841799709724e-06,
-      "loss": 4.1983,
-      "step": 588
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.766328011611032e-06,
-      "loss": 4.1152,
-      "step": 589
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.751814223512337e-06,
-      "loss": 4.1843,
-      "step": 590
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.737300435413644e-06,
-      "loss": 4.3222,
-      "step": 591
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.72278664731495e-06,
-      "loss": 4.1081,
-      "step": 592
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.708272859216255e-06,
-      "loss": 4.5443,
-      "step": 593
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.693759071117563e-06,
-      "loss": 4.2205,
-      "step": 594
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.67924528301887e-06,
-      "loss": 4.3217,
-      "step": 595
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.664731494920174e-06,
-      "loss": 4.0726,
-      "step": 596
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.650217706821482e-06,
-      "loss": 4.3369,
-      "step": 597
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.635703918722788e-06,
-      "loss": 4.4008,
-      "step": 598
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.621190130624092e-06,
-      "loss": 4.3133,
-      "step": 599
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.6066763425254e-06,
-      "loss": 4.8078,
-      "step": 600
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.592162554426706e-06,
-      "loss": 4.1254,
-      "step": 601
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.577648766328013e-06,
-      "loss": 4.2669,
-      "step": 602
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.563134978229319e-06,
-      "loss": 4.4227,
-      "step": 603
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.548621190130625e-06,
-      "loss": 3.9875,
-      "step": 604
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.534107402031931e-06,
-      "loss": 4.2073,
-      "step": 605
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.519593613933237e-06,
-      "loss": 4.294,
-      "step": 606
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.505079825834544e-06,
-      "loss": 4.2053,
-      "step": 607
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.49056603773585e-06,
-      "loss": 4.2851,
-      "step": 608
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.476052249637156e-06,
-      "loss": 4.2182,
-      "step": 609
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.461538461538462e-06,
-      "loss": 4.2754,
-      "step": 610
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.447024673439768e-06,
-      "loss": 4.2814,
-      "step": 611
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.432510885341074e-06,
-      "loss": 4.0234,
-      "step": 612
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.41799709724238e-06,
-      "loss": 4.1668,
-      "step": 613
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.403483309143687e-06,
-      "loss": 4.078,
-      "step": 614
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.388969521044995e-06,
-      "loss": 4.2407,
-      "step": 615
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.3744557329463e-06,
-      "loss": 4.1871,
-      "step": 616
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.359941944847605e-06,
-      "loss": 4.1157,
-      "step": 617
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.345428156748913e-06,
-      "loss": 4.1482,
-      "step": 618
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.330914368650218e-06,
-      "loss": 4.1792,
-      "step": 619
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.316400580551524e-06,
-      "loss": 4.2051,
-      "step": 620
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.301886792452832e-06,
-      "loss": 4.0035,
-      "step": 621
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.287373004354136e-06,
-      "loss": 3.9219,
-      "step": 622
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.272859216255443e-06,
-      "loss": 4.1855,
-      "step": 623
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.25834542815675e-06,
-      "loss": 4.248,
-      "step": 624
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.243831640058055e-06,
-      "loss": 4.2014,
-      "step": 625
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.229317851959361e-06,
-      "loss": 4.2745,
-      "step": 626
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.214804063860669e-06,
-      "loss": 4.3367,
-      "step": 627
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.200290275761974e-06,
-      "loss": 3.9968,
-      "step": 628
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.185776487663281e-06,
-      "loss": 3.9928,
-      "step": 629
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.171262699564588e-06,
-      "loss": 4.2435,
-      "step": 630
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.156748911465894e-06,
-      "loss": 4.2881,
-      "step": 631
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.1422351233672e-06,
-      "loss": 4.144,
-      "step": 632
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.127721335268506e-06,
-      "loss": 4.1987,
-      "step": 633
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.113207547169812e-06,
-      "loss": 4.1523,
-      "step": 634
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.098693759071119e-06,
-      "loss": 4.3182,
-      "step": 635
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.084179970972425e-06,
-      "loss": 4.1759,
-      "step": 636
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.069666182873731e-06,
-      "loss": 3.9399,
-      "step": 637
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.055152394775037e-06,
-      "loss": 4.1084,
-      "step": 638
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.040638606676343e-06,
-      "loss": 4.1451,
-      "step": 639
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.02612481857765e-06,
-      "loss": 4.2404,
-      "step": 640
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.011611030478956e-06,
-      "loss": 4.1882,
-      "step": 641
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.997097242380262e-06,
-      "loss": 4.1495,
-      "step": 642
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.982583454281568e-06,
-      "loss": 4.4123,
-      "step": 643
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.968069666182874e-06,
-      "loss": 4.2135,
-      "step": 644
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.95355587808418e-06,
-      "loss": 4.2395,
-      "step": 645
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.939042089985487e-06,
-      "loss": 4.256,
-      "step": 646
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.924528301886793e-06,
-      "loss": 4.1155,
-      "step": 647
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.910014513788099e-06,
-      "loss": 4.6098,
-      "step": 648
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.895500725689405e-06,
-      "loss": 4.7887,
-      "step": 649
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.880986937590711e-06,
-      "loss": 4.2265,
-      "step": 650
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.866473149492018e-06,
-      "loss": 4.3677,
-      "step": 651
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.851959361393324e-06,
-      "loss": 4.1386,
-      "step": 652
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.837445573294632e-06,
-      "loss": 4.3104,
-      "step": 653
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.822931785195936e-06,
-      "loss": 4.2239,
-      "step": 654
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.808417997097242e-06,
-      "loss": 4.2248,
-      "step": 655
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.79390420899855e-06,
-      "loss": 4.1931,
-      "step": 656
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.779390420899855e-06,
-      "loss": 4.099,
-      "step": 657
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.764876632801161e-06,
-      "loss": 4.1726,
-      "step": 658
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.750362844702469e-06,
-      "loss": 4.1437,
-      "step": 659
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.735849056603775e-06,
-      "loss": 4.1334,
-      "step": 660
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.72133526850508e-06,
-      "loss": 4.1649,
-      "step": 661
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.706821480406387e-06,
-      "loss": 4.0725,
-      "step": 662
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.692307692307694e-06,
-      "loss": 4.2534,
-      "step": 663
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.677793904208998e-06,
-      "loss": 4.179,
-      "step": 664
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.663280116110306e-06,
-      "loss": 4.3741,
-      "step": 665
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.648766328011612e-06,
-      "loss": 4.22,
-      "step": 666
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.634252539912917e-06,
-      "loss": 4.2909,
-      "step": 667
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.619738751814224e-06,
-      "loss": 4.1742,
-      "step": 668
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.605224963715531e-06,
-      "loss": 4.0888,
-      "step": 669
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.590711175616836e-06,
-      "loss": 4.0804,
-      "step": 670
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.576197387518143e-06,
-      "loss": 4.1934,
-      "step": 671
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.561683599419449e-06,
-      "loss": 4.128,
-      "step": 672
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.5471698113207555e-06,
-      "loss": 4.3061,
-      "step": 673
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.532656023222062e-06,
-      "loss": 4.2366,
-      "step": 674
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.518142235123368e-06,
-      "loss": 4.142,
-      "step": 675
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.503628447024675e-06,
-      "loss": 3.8797,
-      "step": 676
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.48911465892598e-06,
-      "loss": 4.353,
-      "step": 677
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.4746008708272864e-06,
-      "loss": 4.191,
-      "step": 678
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.4600870827285935e-06,
-      "loss": 4.0922,
-      "step": 679
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.445573294629899e-06,
-      "loss": 4.3988,
-      "step": 680
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.431059506531205e-06,
-      "loss": 3.8998,
-      "step": 681
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.416545718432512e-06,
-      "loss": 4.0448,
-      "step": 682
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.402031930333817e-06,
-      "loss": 4.3168,
-      "step": 683
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.387518142235124e-06,
-      "loss": 4.0092,
-      "step": 684
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.373004354136431e-06,
-      "loss": 4.1862,
-      "step": 685
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.358490566037736e-06,
-      "loss": 4.221,
-      "step": 686
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.343976777939042e-06,
-      "loss": 4.1682,
-      "step": 687
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.329462989840349e-06,
-      "loss": 4.2095,
-      "step": 688
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.314949201741655e-06,
-      "loss": 4.375,
-      "step": 689
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.300435413642961e-06,
-      "loss": 4.2336,
-      "step": 690
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.285921625544268e-06,
-      "loss": 4.4989,
-      "step": 691
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.271407837445574e-06,
-      "loss": 4.4095,
-      "step": 692
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.25689404934688e-06,
-      "loss": 4.4392,
-      "step": 693
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.242380261248186e-06,
-      "loss": 4.3964,
-      "step": 694
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.227866473149493e-06,
-      "loss": 4.2091,
-      "step": 695
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.213352685050799e-06,
-      "loss": 4.0992,
-      "step": 696
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.198838896952105e-06,
-      "loss": 4.3356,
-      "step": 697
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.184325108853412e-06,
-      "loss": 4.5925,
-      "step": 698
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.169811320754717e-06,
-      "loss": 4.1561,
-      "step": 699
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.1552975326560235e-06,
-      "loss": 4.952,
-      "step": 700
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.1407837445573305e-06,
-      "loss": 4.2578,
-      "step": 701
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.126269956458637e-06,
-      "loss": 4.1095,
-      "step": 702
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.111756168359942e-06,
-      "loss": 4.1689,
-      "step": 703
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.097242380261249e-06,
-      "loss": 3.8803,
-      "step": 704
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.082728592162555e-06,
-      "loss": 4.12,
-      "step": 705
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.068214804063861e-06,
-      "loss": 4.1665,
-      "step": 706
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.053701015965168e-06,
-      "loss": 4.187,
-      "step": 707
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.039187227866474e-06,
-      "loss": 4.09,
-      "step": 708
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.024673439767779e-06,
-      "loss": 4.1873,
-      "step": 709
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.010159651669086e-06,
-      "loss": 4.1409,
-      "step": 710
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.9956458635703924e-06,
-      "loss": 4.0787,
-      "step": 711
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.981132075471699e-06,
-      "loss": 4.2809,
-      "step": 712
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.966618287373005e-06,
-      "loss": 4.0982,
-      "step": 713
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.952104499274311e-06,
-      "loss": 4.0377,
-      "step": 714
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.937590711175617e-06,
-      "loss": 4.1616,
-      "step": 715
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.923076923076923e-06,
-      "loss": 4.0324,
-      "step": 716
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.90856313497823e-06,
-      "loss": 4.1099,
-      "step": 717
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.894049346879537e-06,
-      "loss": 4.3045,
-      "step": 718
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.879535558780842e-06,
-      "loss": 4.1416,
-      "step": 719
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.865021770682149e-06,
-      "loss": 4.0256,
-      "step": 720
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.850507982583455e-06,
-      "loss": 4.1352,
-      "step": 721
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.8359941944847605e-06,
-      "loss": 4.0131,
-      "step": 722
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.8214804063860676e-06,
-      "loss": 4.0822,
-      "step": 723
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.806966618287374e-06,
-      "loss": 4.0693,
-      "step": 724
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.792452830188679e-06,
-      "loss": 4.2622,
-      "step": 725
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.777939042089986e-06,
-      "loss": 4.2479,
-      "step": 726
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.763425253991292e-06,
-      "loss": 4.1437,
-      "step": 727
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.748911465892598e-06,
-      "loss": 4.3095,
-      "step": 728
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.734397677793905e-06,
-      "loss": 4.0002,
-      "step": 729
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.719883889695211e-06,
-      "loss": 4.3412,
-      "step": 730
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.705370101596517e-06,
-      "loss": 4.2805,
-      "step": 731
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.690856313497823e-06,
-      "loss": 4.2931,
-      "step": 732
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.6763425253991295e-06,
-      "loss": 3.9637,
-      "step": 733
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.6618287373004365e-06,
-      "loss": 4.2733,
-      "step": 734
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.647314949201742e-06,
-      "loss": 4.1742,
-      "step": 735
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.632801161103049e-06,
-      "loss": 4.1126,
-      "step": 736
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.618287373004355e-06,
-      "loss": 4.201,
-      "step": 737
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.60377358490566e-06,
-      "loss": 4.3611,
-      "step": 738
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.5892597968069675e-06,
-      "loss": 4.3106,
-      "step": 739
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.574746008708274e-06,
-      "loss": 4.0926,
-      "step": 740
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.560232220609579e-06,
-      "loss": 4.4143,
-      "step": 741
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.545718432510886e-06,
-      "loss": 4.1246,
-      "step": 742
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.531204644412192e-06,
-      "loss": 4.243,
-      "step": 743
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.5166908563134976e-06,
-      "loss": 4.3983,
-      "step": 744
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.502177068214805e-06,
-      "loss": 4.3649,
-      "step": 745
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.487663280116111e-06,
-      "loss": 4.3901,
-      "step": 746
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.473149492017418e-06,
-      "loss": 4.5666,
-      "step": 747
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.458635703918723e-06,
-      "loss": 4.724,
-      "step": 748
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.444121915820029e-06,
-      "loss": 4.3138,
-      "step": 749
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.429608127721336e-06,
-      "loss": 4.5664,
-      "step": 750
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.415094339622642e-06,
-      "loss": 4.0266,
-      "step": 751
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.400580551523948e-06,
-      "loss": 4.2239,
-      "step": 752
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.386066763425255e-06,
-      "loss": 4.2583,
-      "step": 753
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.37155297532656e-06,
-      "loss": 4.1042,
-      "step": 754
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.357039187227867e-06,
-      "loss": 4.2795,
-      "step": 755
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.3425253991291735e-06,
-      "loss": 4.1196,
-      "step": 756
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.328011611030479e-06,
-      "loss": 4.1597,
-      "step": 757
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.313497822931786e-06,
-      "loss": 4.029,
-      "step": 758
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.298984034833092e-06,
-      "loss": 4.103,
-      "step": 759
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.2844702467343975e-06,
-      "loss": 4.3008,
-      "step": 760
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.2699564586357045e-06,
-      "loss": 4.0217,
-      "step": 761
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.255442670537011e-06,
-      "loss": 4.1494,
-      "step": 762
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.240928882438318e-06,
-      "loss": 4.1743,
-      "step": 763
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.226415094339623e-06,
-      "loss": 4.0613,
-      "step": 764
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.211901306240929e-06,
-      "loss": 4.306,
-      "step": 765
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.197387518142236e-06,
-      "loss": 4.1558,
-      "step": 766
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.182873730043542e-06,
-      "loss": 4.1751,
-      "step": 767
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.168359941944848e-06,
-      "loss": 3.9652,
-      "step": 768
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.153846153846155e-06,
-      "loss": 4.1907,
-      "step": 769
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.13933236574746e-06,
-      "loss": 4.335,
-      "step": 770
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.124818577648766e-06,
-      "loss": 3.8491,
-      "step": 771
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.1103047895500734e-06,
-      "loss": 4.2559,
-      "step": 772
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.095791001451379e-06,
-      "loss": 4.04,
-      "step": 773
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.081277213352685e-06,
-      "loss": 4.029,
-      "step": 774
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.066763425253992e-06,
-      "loss": 4.0631,
-      "step": 775
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.052249637155298e-06,
-      "loss": 4.1248,
-      "step": 776
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.037735849056604e-06,
-      "loss": 4.1689,
-      "step": 777
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.023222060957911e-06,
-      "loss": 4.1186,
-      "step": 778
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.008708272859218e-06,
-      "loss": 4.0585,
-      "step": 779
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.994194484760523e-06,
-      "loss": 4.2947,
-      "step": 780
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.979680696661829e-06,
-      "loss": 4.0833,
-      "step": 781
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.965166908563136e-06,
-      "loss": 4.0812,
-      "step": 782
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.9506531204644415e-06,
-      "loss": 3.8706,
-      "step": 783
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.936139332365748e-06,
-      "loss": 4.3378,
-      "step": 784
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.921625544267055e-06,
-      "loss": 4.3929,
-      "step": 785
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.90711175616836e-06,
-      "loss": 4.0669,
-      "step": 786
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.892597968069666e-06,
-      "loss": 4.3642,
-      "step": 787
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.878084179970973e-06,
-      "loss": 4.0127,
-      "step": 788
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.863570391872279e-06,
-      "loss": 4.0658,
-      "step": 789
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.849056603773585e-06,
-      "loss": 3.9261,
-      "step": 790
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.834542815674892e-06,
-      "loss": 4.439,
-      "step": 791
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.820029027576198e-06,
-      "loss": 4.4266,
-      "step": 792
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.8055152394775034e-06,
-      "loss": 4.6576,
-      "step": 793
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.7910014513788105e-06,
-      "loss": 4.1698,
-      "step": 794
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.776487663280117e-06,
-      "loss": 4.2262,
-      "step": 795
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.761973875181423e-06,
-      "loss": 3.9713,
-      "step": 796
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.747460087082729e-06,
-      "loss": 4.2141,
-      "step": 797
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.732946298984035e-06,
-      "loss": 4.3976,
-      "step": 798
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.7184325108853414e-06,
-      "loss": 4.4076,
-      "step": 799
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.703918722786648e-06,
-      "loss": 4.3116,
-      "step": 800
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.689404934687955e-06,
-      "loss": 4.1031,
-      "step": 801
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.67489114658926e-06,
-      "loss": 4.094,
-      "step": 802
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.660377358490566e-06,
-      "loss": 4.1113,
-      "step": 803
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.645863570391873e-06,
-      "loss": 4.1253,
-      "step": 804
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.6313497822931794e-06,
-      "loss": 4.0477,
-      "step": 805
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.616835994194485e-06,
-      "loss": 4.1518,
-      "step": 806
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.602322206095792e-06,
-      "loss": 4.2,
-      "step": 807
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.587808417997098e-06,
-      "loss": 4.0623,
-      "step": 808
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.573294629898403e-06,
-      "loss": 4.0619,
-      "step": 809
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.55878084179971e-06,
-      "loss": 4.0337,
-      "step": 810
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.5442670537010166e-06,
-      "loss": 4.1682,
-      "step": 811
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.529753265602322e-06,
-      "loss": 4.1935,
-      "step": 812
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.515239477503629e-06,
-      "loss": 4.1785,
-      "step": 813
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.500725689404935e-06,
-      "loss": 4.2914,
-      "step": 814
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.486211901306241e-06,
-      "loss": 4.122,
-      "step": 815
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.4716981132075475e-06,
-      "loss": 4.1946,
-      "step": 816
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.457184325108854e-06,
-      "loss": 4.0397,
-      "step": 817
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.44267053701016e-06,
-      "loss": 4.0283,
-      "step": 818
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.428156748911466e-06,
-      "loss": 4.065,
-      "step": 819
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.413642960812773e-06,
-      "loss": 4.2216,
-      "step": 820
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.399129172714079e-06,
-      "loss": 4.2041,
-      "step": 821
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.384615384615385e-06,
-      "loss": 4.0025,
-      "step": 822
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.370101596516692e-06,
-      "loss": 4.1591,
-      "step": 823
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.355587808417998e-06,
-      "loss": 4.3515,
-      "step": 824
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.341074020319303e-06,
-      "loss": 4.0397,
-      "step": 825
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.32656023222061e-06,
-      "loss": 4.4288,
-      "step": 826
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.3120464441219165e-06,
-      "loss": 4.1617,
-      "step": 827
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.297532656023222e-06,
-      "loss": 4.2937,
-      "step": 828
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.283018867924529e-06,
-      "loss": 4.0729,
-      "step": 829
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.268505079825835e-06,
-      "loss": 4.1548,
-      "step": 830
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.25399129172714e-06,
-      "loss": 4.1487,
-      "step": 831
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.239477503628447e-06,
-      "loss": 4.0895,
-      "step": 832
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.224963715529754e-06,
-      "loss": 4.0042,
-      "step": 833
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.210449927431061e-06,
-      "loss": 4.2332,
-      "step": 834
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.195936139332366e-06,
-      "loss": 4.2404,
-      "step": 835
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.181422351233672e-06,
-      "loss": 4.2952,
-      "step": 836
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.166908563134979e-06,
-      "loss": 4.0646,
-      "step": 837
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.1523947750362846e-06,
-      "loss": 3.8413,
-      "step": 838
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.137880986937592e-06,
-      "loss": 4.2514,
-      "step": 839
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.123367198838898e-06,
-      "loss": 3.9775,
-      "step": 840
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.108853410740203e-06,
-      "loss": 4.2609,
-      "step": 841
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.09433962264151e-06,
-      "loss": 4.2512,
-      "step": 842
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.079825834542816e-06,
-      "loss": 4.505,
-      "step": 843
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.065312046444122e-06,
-      "loss": 4.2803,
-      "step": 844
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.050798258345429e-06,
-      "loss": 4.1739,
-      "step": 845
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.036284470246735e-06,
-      "loss": 4.1231,
-      "step": 846
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.02177068214804e-06,
-      "loss": 4.4431,
-      "step": 847
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.007256894049347e-06,
-      "loss": 3.9688,
-      "step": 848
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.9927431059506535e-06,
-      "loss": 4.5994,
-      "step": 849
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.97822931785196e-06,
-      "loss": 4.6661,
-      "step": 850
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.963715529753266e-06,
-      "loss": 4.2855,
-      "step": 851
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.949201741654572e-06,
-      "loss": 4.1274,
-      "step": 852
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.934687953555878e-06,
-      "loss": 4.2258,
-      "step": 853
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.920174165457185e-06,
-      "loss": 4.1522,
-      "step": 854
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.905660377358491e-06,
-      "loss": 4.0969,
-      "step": 855
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.891146589259797e-06,
-      "loss": 4.0993,
-      "step": 856
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.876632801161104e-06,
-      "loss": 4.1198,
-      "step": 857
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.862119013062409e-06,
-      "loss": 4.0446,
-      "step": 858
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.847605224963715e-06,
-      "loss": 4.1519,
-      "step": 859
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.8330914368650224e-06,
-      "loss": 4.1818,
-      "step": 860
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.818577648766329e-06,
-      "loss": 3.9429,
-      "step": 861
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.804063860667635e-06,
-      "loss": 4.2424,
-      "step": 862
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.789550072568941e-06,
-      "loss": 4.2388,
-      "step": 863
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.775036284470247e-06,
-      "loss": 4.064,
-      "step": 864
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.760522496371553e-06,
-      "loss": 4.1204,
-      "step": 865
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.74600870827286e-06,
-      "loss": 4.1002,
-      "step": 866
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.731494920174166e-06,
-      "loss": 4.0604,
-      "step": 867
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.716981132075472e-06,
-      "loss": 4.2068,
-      "step": 868
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.702467343976778e-06,
-      "loss": 4.28,
-      "step": 869
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.687953555878084e-06,
-      "loss": 3.8426,
-      "step": 870
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.6734397677793905e-06,
-      "loss": 3.8959,
-      "step": 871
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.658925979680697e-06,
-      "loss": 4.0439,
-      "step": 872
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.644412191582004e-06,
-      "loss": 4.1897,
-      "step": 873
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.629898403483309e-06,
-      "loss": 3.9992,
-      "step": 874
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.615384615384616e-06,
-      "loss": 4.1393,
-      "step": 875
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.600870827285922e-06,
-      "loss": 4.0275,
-      "step": 876
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.586357039187228e-06,
-      "loss": 3.9959,
-      "step": 877
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.571843251088535e-06,
-      "loss": 4.2283,
-      "step": 878
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.557329462989841e-06,
-      "loss": 4.1318,
-      "step": 879
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.542815674891147e-06,
-      "loss": 4.311,
-      "step": 880
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.528301886792453e-06,
-      "loss": 4.0047,
-      "step": 881
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.5137880986937595e-06,
-      "loss": 4.2101,
-      "step": 882
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.499274310595066e-06,
-      "loss": 4.12,
-      "step": 883
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.484760522496372e-06,
-      "loss": 4.2002,
-      "step": 884
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.470246734397678e-06,
-      "loss": 4.0107,
-      "step": 885
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.455732946298984e-06,
-      "loss": 4.1442,
-      "step": 886
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.4412191582002904e-06,
-      "loss": 4.3743,
-      "step": 887
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.426705370101597e-06,
-      "loss": 4.285,
-      "step": 888
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.412191582002903e-06,
-      "loss": 4.2912,
-      "step": 889
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.397677793904209e-06,
-      "loss": 4.2921,
-      "step": 890
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.383164005805516e-06,
-      "loss": 4.2769,
-      "step": 891
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.368650217706822e-06,
-      "loss": 4.222,
-      "step": 892
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.354136429608128e-06,
-      "loss": 4.1861,
-      "step": 893
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.339622641509435e-06,
-      "loss": 4.0274,
-      "step": 894
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.325108853410741e-06,
-      "loss": 4.2386,
-      "step": 895
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.310595065312046e-06,
-      "loss": 4.2893,
-      "step": 896
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.296081277213353e-06,
-      "loss": 4.3261,
-      "step": 897
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.281567489114659e-06,
-      "loss": 4.1391,
-      "step": 898
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.2670537010159656e-06,
-      "loss": 4.1483,
-      "step": 899
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.252539912917272e-06,
-      "loss": 4.4841,
-      "step": 900
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.238026124818578e-06,
-      "loss": 4.1739,
-      "step": 901
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.223512336719884e-06,
-      "loss": 4.24,
-      "step": 902
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.20899854862119e-06,
-      "loss": 4.1859,
-      "step": 903
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.194484760522497e-06,
-      "loss": 4.1927,
-      "step": 904
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.179970972423803e-06,
-      "loss": 4.0778,
-      "step": 905
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.165457184325109e-06,
-      "loss": 4.2416,
-      "step": 906
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.150943396226416e-06,
-      "loss": 4.1455,
-      "step": 907
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.136429608127721e-06,
-      "loss": 4.4262,
-      "step": 908
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.1219158200290275e-06,
-      "loss": 4.2104,
-      "step": 909
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.1074020319303345e-06,
-      "loss": 4.2648,
-      "step": 910
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.092888243831641e-06,
-      "loss": 4.0618,
-      "step": 911
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.078374455732947e-06,
-      "loss": 4.0866,
-      "step": 912
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.063860667634253e-06,
-      "loss": 4.1445,
-      "step": 913
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.049346879535559e-06,
-      "loss": 4.0978,
-      "step": 914
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.0348330914368655e-06,
-      "loss": 4.1832,
-      "step": 915
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.020319303338172e-06,
-      "loss": 4.1319,
-      "step": 916
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.005805515239478e-06,
-      "loss": 4.1499,
-      "step": 917
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.991291727140784e-06,
-      "loss": 4.0823,
-      "step": 918
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.97677793904209e-06,
-      "loss": 4.3119,
-      "step": 919
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.962264150943396e-06,
-      "loss": 4.1966,
-      "step": 920
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.947750362844703e-06,
-      "loss": 4.1175,
-      "step": 921
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.933236574746009e-06,
-      "loss": 4.2104,
-      "step": 922
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.918722786647316e-06,
-      "loss": 4.0601,
-      "step": 923
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.904208998548621e-06,
-      "loss": 4.111,
-      "step": 924
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.889695210449927e-06,
-      "loss": 4.017,
-      "step": 925
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.875181422351234e-06,
-      "loss": 4.019,
-      "step": 926
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.86066763425254e-06,
-      "loss": 4.1868,
-      "step": 927
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.846153846153847e-06,
-      "loss": 4.2295,
-      "step": 928
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.831640058055153e-06,
-      "loss": 4.0424,
-      "step": 929
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.817126269956458e-06,
-      "loss": 3.9899,
-      "step": 930
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.8026124818577654e-06,
-      "loss": 4.1918,
-      "step": 931
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.7880986937590716e-06,
-      "loss": 4.2305,
-      "step": 932
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.7735849056603777e-06,
-      "loss": 4.177,
-      "step": 933
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.759071117561684e-06,
-      "loss": 4.2157,
-      "step": 934
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.74455732946299e-06,
-      "loss": 4.2859,
-      "step": 935
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.7300435413642967e-06,
-      "loss": 4.2321,
-      "step": 936
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.7155297532656025e-06,
-      "loss": 4.187,
-      "step": 937
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.7010159651669087e-06,
-      "loss": 4.3289,
-      "step": 938
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.6865021770682153e-06,
-      "loss": 4.1749,
-      "step": 939
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.671988388969521e-06,
-      "loss": 4.1726,
-      "step": 940
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.6574746008708277e-06,
-      "loss": 4.0525,
-      "step": 941
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.642960812772134e-06,
-      "loss": 4.2322,
-      "step": 942
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.62844702467344e-06,
-      "loss": 4.0624,
-      "step": 943
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.6139332365747467e-06,
-      "loss": 4.2563,
-      "step": 944
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.5994194484760525e-06,
-      "loss": 4.2842,
-      "step": 945
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.5849056603773586e-06,
-      "loss": 4.0499,
-      "step": 946
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.5703918722786653e-06,
-      "loss": 4.288,
-      "step": 947
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.555878084179971e-06,
-      "loss": 4.258,
-      "step": 948
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.5413642960812776e-06,
-      "loss": 4.3442,
-      "step": 949
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.526850507982584e-06,
-      "loss": 4.1987,
-      "step": 950
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.5123367198838896e-06,
-      "loss": 4.0576,
-      "step": 951
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.4978229317851962e-06,
-      "loss": 4.216,
-      "step": 952
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.4833091436865024e-06,
-      "loss": 4.0031,
-      "step": 953
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.4687953555878086e-06,
-      "loss": 3.8672,
-      "step": 954
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.454281567489115e-06,
-      "loss": 4.093,
-      "step": 955
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.439767779390421e-06,
-      "loss": 4.2003,
-      "step": 956
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.4252539912917276e-06,
-      "loss": 4.1416,
-      "step": 957
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.4107402031930338e-06,
-      "loss": 4.0459,
-      "step": 958
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.3962264150943395e-06,
-      "loss": 4.0946,
-      "step": 959
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.381712626995646e-06,
-      "loss": 4.062,
-      "step": 960
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.3671988388969524e-06,
-      "loss": 4.006,
-      "step": 961
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.3526850507982585e-06,
-      "loss": 4.2121,
-      "step": 962
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.3381712626995647e-06,
-      "loss": 4.0015,
-      "step": 963
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.323657474600871e-06,
-      "loss": 4.1986,
-      "step": 964
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.3091436865021775e-06,
-      "loss": 4.3211,
-      "step": 965
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.2946298984034837e-06,
-      "loss": 3.971,
-      "step": 966
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.2801161103047895e-06,
-      "loss": 4.1907,
-      "step": 967
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.265602322206096e-06,
-      "loss": 4.2515,
-      "step": 968
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.2510885341074023e-06,
-      "loss": 3.8172,
-      "step": 969
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.236574746008709e-06,
-      "loss": 4.1306,
-      "step": 970
+      "loss": 4.276,
+      "step": 487
     },
     {
       "epoch": 0.82,
-      "learning_rate": 3.2220609579100147e-06,
-      "loss": 3.9755,
-      "step": 971
+      "learning_rate": 9.7e-06,
+      "loss": 4.2906,
+      "step": 488
     },
     {
       "epoch": 0.82,
-      "learning_rate": 3.207547169811321e-06,
-      "loss": 4.052,
-      "step": 972
+      "learning_rate": 9.72e-06,
+      "loss": 4.216,
+      "step": 489
     },
     {
       "epoch": 0.82,
-      "learning_rate": 3.1930333817126275e-06,
+      "learning_rate": 9.74e-06,
       "loss": 4.1313,
-      "step": 973
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.1785195936139337e-06,
-      "loss": 4.0439,
-      "step": 974
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.1640058055152394e-06,
-      "loss": 4.1472,
-      "step": 975
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.149492017416546e-06,
-      "loss": 4.1512,
-      "step": 976
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.1349782293178522e-06,
-      "loss": 3.9953,
-      "step": 977
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.120464441219159e-06,
-      "loss": 4.1559,
-      "step": 978
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.1059506531204646e-06,
-      "loss": 4.0212,
-      "step": 979
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.091436865021771e-06,
-      "loss": 4.1268,
-      "step": 980
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.0769230769230774e-06,
-      "loss": 4.2255,
-      "step": 981
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.062409288824383e-06,
-      "loss": 4.1876,
-      "step": 982
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.0478955007256894e-06,
-      "loss": 4.0415,
-      "step": 983
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.033381712626996e-06,
-      "loss": 4.1892,
-      "step": 984
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.018867924528302e-06,
-      "loss": 4.001,
-      "step": 985
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.004354136429609e-06,
-      "loss": 4.1283,
-      "step": 986
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.9898403483309146e-06,
-      "loss": 4.2559,
-      "step": 987
+      "step": 490
     },
     {
       "epoch": 0.83,
-      "learning_rate": 2.9753265602322208e-06,
-      "loss": 4.1043,
-      "step": 988
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 4.3359,
+      "step": 491
     },
     {
       "epoch": 0.83,
-      "learning_rate": 2.9608127721335274e-06,
-      "loss": 4.1658,
-      "step": 989
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.1545,
+      "step": 492
     },
     {
       "epoch": 0.83,
-      "learning_rate": 2.946298984034833e-06,
-      "loss": 4.1996,
-      "step": 990
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.036,
+      "step": 493
     },
     {
       "epoch": 0.83,
-      "learning_rate": 2.9317851959361393e-06,
-      "loss": 4.2113,
-      "step": 991
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 4.2527,
+      "step": 494
     },
     {
       "epoch": 0.83,
-      "learning_rate": 2.917271407837446e-06,
-      "loss": 4.1202,
-      "step": 992
+      "learning_rate": 9.84e-06,
+      "loss": 4.3611,
+      "step": 495
     },
     {
       "epoch": 0.83,
-      "learning_rate": 2.9027576197387517e-06,
-      "loss": 3.9579,
-      "step": 993
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8882438316400583e-06,
-      "loss": 4.1758,
-      "step": 994
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8737300435413645e-06,
-      "loss": 4.3022,
-      "step": 995
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8592162554426707e-06,
-      "loss": 4.6287,
-      "step": 996
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8447024673439773e-06,
-      "loss": 4.2997,
-      "step": 997
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.830188679245283e-06,
-      "loss": 4.404,
-      "step": 998
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8156748911465897e-06,
-      "loss": 4.385,
-      "step": 999
+      "learning_rate": 9.86e-06,
+      "loss": 4.3937,
+      "step": 496
     },
     {
       "epoch": 0.84,
-      "learning_rate": 2.801161103047896e-06,
-      "loss": 4.555,
-      "step": 1000
+      "learning_rate": 9.88e-06,
+      "loss": 4.2911,
+      "step": 497
     },
     {
       "epoch": 0.84,
-      "eval_loss": 4.16819953918457,
-      "eval_runtime": 794.4467,
-      "eval_samples_per_second": 3.326,
-      "eval_steps_per_second": 0.278,
-      "eval_wer": 1.9665807219357399,
-      "step": 1000
+      "learning_rate": 9.9e-06,
+      "loss": 4.4337,
+      "step": 498
     },
     {
       "epoch": 0.84,
-      "learning_rate": 2.7866473149492017e-06,
-      "loss": 4.2529,
-      "step": 1001
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 4.4938,
+      "step": 499
     },
     {
       "epoch": 0.84,
-      "learning_rate": 2.7721335268505083e-06,
-      "loss": 4.2023,
-      "step": 1002
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 4.4108,
+      "step": 500
     },
     {
       "epoch": 0.84,
-      "learning_rate": 2.7576197387518145e-06,
-      "loss": 4.0947,
-      "step": 1003
+      "eval_loss": 4.200125694274902,
+      "eval_runtime": 786.8447,
+      "eval_samples_per_second": 3.358,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 1.9598968663228877,
+      "step": 500
     },
     {
       "epoch": 0.84,
-      "learning_rate": 2.7431059506531207e-06,
-      "loss": 4.058,
-      "step": 1004
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.2338,
+      "step": 501
     },
     {
       "epoch": 0.84,
-      "learning_rate": 2.728592162554427e-06,
-      "loss": 4.2492,
-      "step": 1005
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.714078374455733e-06,
-      "loss": 4.088,
-      "step": 1006
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6995645863570397e-06,
-      "loss": 4.0825,
-      "step": 1007
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.685050798258346e-06,
-      "loss": 4.0822,
-      "step": 1008
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6705370101596516e-06,
-      "loss": 4.1022,
-      "step": 1009
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6560232220609582e-06,
-      "loss": 4.1269,
-      "step": 1010
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6415094339622644e-06,
-      "loss": 4.0737,
-      "step": 1011
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 4.1417,
+      "step": 502
     },
     {
       "epoch": 0.85,
-      "learning_rate": 2.62699564586357e-06,
-      "loss": 4.2393,
-      "step": 1012
+      "learning_rate": 1e-05,
+      "loss": 4.1053,
+      "step": 503
     },
     {
       "epoch": 0.85,
-      "learning_rate": 2.612481857764877e-06,
-      "loss": 4.0142,
-      "step": 1013
+      "learning_rate": 9.893617021276596e-06,
+      "loss": 4.1975,
+      "step": 504
     },
     {
       "epoch": 0.85,
-      "learning_rate": 2.597968069666183e-06,
-      "loss": 4.1347,
-      "step": 1014
+      "learning_rate": 9.787234042553192e-06,
+      "loss": 4.1617,
+      "step": 505
     },
     {
       "epoch": 0.85,
-      "learning_rate": 2.5834542815674896e-06,
-      "loss": 4.1347,
-      "step": 1015
+      "learning_rate": 9.680851063829787e-06,
+      "loss": 4.2979,
+      "step": 506
     },
     {
       "epoch": 0.85,
-      "learning_rate": 2.568940493468796e-06,
-      "loss": 4.2207,
-      "step": 1016
+      "learning_rate": 9.574468085106385e-06,
+      "loss": 4.1643,
+      "step": 507
     },
     {
       "epoch": 0.85,
-      "learning_rate": 2.5544267053701016e-06,
-      "loss": 4.0783,
-      "step": 1017
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.539912917271408e-06,
-      "loss": 4.1704,
-      "step": 1018
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.5253991291727144e-06,
-      "loss": 4.2673,
-      "step": 1019
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.51088534107402e-06,
-      "loss": 3.9835,
-      "step": 1020
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.4963715529753268e-06,
-      "loss": 4.0653,
-      "step": 1021
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.481857764876633e-06,
-      "loss": 4.2703,
-      "step": 1022
+      "learning_rate": 9.46808510638298e-06,
+      "loss": 4.158,
+      "step": 508
     },
     {
       "epoch": 0.86,
-      "learning_rate": 2.467343976777939e-06,
-      "loss": 4.2143,
-      "step": 1023
+      "learning_rate": 9.361702127659576e-06,
+      "loss": 4.0941,
+      "step": 509
     },
     {
       "epoch": 0.86,
-      "learning_rate": 2.4528301886792453e-06,
-      "loss": 3.9575,
-      "step": 1024
+      "learning_rate": 9.255319148936171e-06,
+      "loss": 4.17,
+      "step": 510
     },
     {
       "epoch": 0.86,
-      "learning_rate": 2.438316400580552e-06,
-      "loss": 4.129,
-      "step": 1025
+      "learning_rate": 9.148936170212767e-06,
+      "loss": 4.2423,
+      "step": 511
     },
     {
       "epoch": 0.86,
-      "learning_rate": 2.4238026124818577e-06,
-      "loss": 4.0597,
-      "step": 1026
+      "learning_rate": 9.042553191489362e-06,
+      "loss": 4.1385,
+      "step": 512
     },
     {
       "epoch": 0.86,
-      "learning_rate": 2.4092888243831643e-06,
-      "loss": 4.1458,
-      "step": 1027
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 4.1332,
+      "step": 513
     },
     {
       "epoch": 0.86,
-      "learning_rate": 2.3947750362844705e-06,
-      "loss": 4.0542,
-      "step": 1028
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.3802612481857767e-06,
-      "loss": 3.9673,
-      "step": 1029
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.365747460087083e-06,
-      "loss": 3.9322,
-      "step": 1030
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.351233671988389e-06,
-      "loss": 4.1389,
-      "step": 1031
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.3367198838896953e-06,
-      "loss": 4.0368,
-      "step": 1032
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.322206095791002e-06,
-      "loss": 4.1147,
-      "step": 1033
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.307692307692308e-06,
-      "loss": 4.0622,
-      "step": 1034
+      "learning_rate": 8.829787234042555e-06,
+      "loss": 4.0833,
+      "step": 514
     },
     {
       "epoch": 0.87,
-      "learning_rate": 2.293178519593614e-06,
-      "loss": 3.9735,
-      "step": 1035
+      "learning_rate": 8.72340425531915e-06,
+      "loss": 4.2107,
+      "step": 515
     },
     {
       "epoch": 0.87,
-      "learning_rate": 2.2786647314949205e-06,
-      "loss": 3.9822,
-      "step": 1036
+      "learning_rate": 8.617021276595746e-06,
+      "loss": 4.1385,
+      "step": 516
     },
     {
       "epoch": 0.87,
-      "learning_rate": 2.2641509433962266e-06,
-      "loss": 3.9537,
-      "step": 1037
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 4.1791,
+      "step": 517
     },
     {
       "epoch": 0.87,
-      "learning_rate": 2.249637155297533e-06,
-      "loss": 4.3565,
-      "step": 1038
+      "learning_rate": 8.404255319148937e-06,
+      "loss": 4.2385,
+      "step": 518
     },
     {
       "epoch": 0.87,
-      "learning_rate": 2.235123367198839e-06,
-      "loss": 4.1527,
-      "step": 1039
+      "learning_rate": 8.297872340425532e-06,
+      "loss": 4.0488,
+      "step": 519
     },
     {
       "epoch": 0.87,
-      "learning_rate": 2.2206095791001452e-06,
-      "loss": 4.1156,
-      "step": 1040
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.2060957910014514e-06,
-      "loss": 4.2147,
-      "step": 1041
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.191582002902758e-06,
-      "loss": 4.07,
-      "step": 1042
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.177068214804064e-06,
-      "loss": 4.107,
-      "step": 1043
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.1625544267053704e-06,
-      "loss": 4.2446,
-      "step": 1044
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.1480406386066766e-06,
-      "loss": 4.246,
-      "step": 1045
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.1335268505079828e-06,
-      "loss": 4.1311,
-      "step": 1046
+      "learning_rate": 8.191489361702128e-06,
+      "loss": 4.1701,
+      "step": 520
     },
     {
       "epoch": 0.88,
-      "learning_rate": 2.119013062409289e-06,
-      "loss": 4.1425,
-      "step": 1047
+      "learning_rate": 8.085106382978723e-06,
+      "loss": 4.1984,
+      "step": 521
     },
     {
       "epoch": 0.88,
-      "learning_rate": 2.104499274310595e-06,
-      "loss": 4.4241,
-      "step": 1048
+      "learning_rate": 7.97872340425532e-06,
+      "loss": 4.2571,
+      "step": 522
     },
     {
       "epoch": 0.88,
-      "learning_rate": 2.0899854862119014e-06,
-      "loss": 4.6764,
-      "step": 1049
+      "learning_rate": 7.872340425531916e-06,
+      "loss": 4.0696,
+      "step": 523
     },
     {
       "epoch": 0.88,
-      "learning_rate": 2.075471698113208e-06,
-      "loss": 4.4022,
-      "step": 1050
+      "learning_rate": 7.765957446808511e-06,
+      "loss": 3.9969,
+      "step": 524
     },
     {
       "epoch": 0.88,
-      "learning_rate": 2.0609579100145137e-06,
-      "loss": 4.0961,
-      "step": 1051
+      "learning_rate": 7.659574468085107e-06,
+      "loss": 4.0892,
+      "step": 525
     },
     {
       "epoch": 0.88,
-      "learning_rate": 2.0464441219158204e-06,
-      "loss": 4.0602,
-      "step": 1052
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.0319303338171265e-06,
-      "loss": 4.0292,
-      "step": 1053
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.0174165457184327e-06,
-      "loss": 4.1502,
-      "step": 1054
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.002902757619739e-06,
-      "loss": 4.2151,
-      "step": 1055
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.988388969521045e-06,
-      "loss": 4.2744,
-      "step": 1056
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.9738751814223513e-06,
-      "loss": 4.2287,
-      "step": 1057
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.959361393323658e-06,
-      "loss": 3.9549,
-      "step": 1058
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.9448476052249637e-06,
-      "loss": 4.2096,
-      "step": 1059
+      "learning_rate": 7.553191489361703e-06,
+      "loss": 4.0812,
+      "step": 526
     },
     {
       "epoch": 0.89,
-      "learning_rate": 1.93033381712627e-06,
-      "loss": 4.0981,
-      "step": 1060
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 4.2357,
+      "step": 527
     },
     {
       "epoch": 0.89,
-      "learning_rate": 1.9158200290275765e-06,
-      "loss": 4.2533,
-      "step": 1061
+      "learning_rate": 7.340425531914894e-06,
+      "loss": 4.1053,
+      "step": 528
     },
     {
       "epoch": 0.89,
-      "learning_rate": 1.9013062409288827e-06,
-      "loss": 4.0279,
-      "step": 1062
+      "learning_rate": 7.234042553191491e-06,
+      "loss": 4.056,
+      "step": 529
     },
     {
       "epoch": 0.89,
-      "learning_rate": 1.8867924528301889e-06,
-      "loss": 4.1658,
-      "step": 1063
+      "learning_rate": 7.127659574468085e-06,
+      "loss": 4.0824,
+      "step": 530
     },
     {
       "epoch": 0.89,
-      "learning_rate": 1.872278664731495e-06,
-      "loss": 4.0382,
-      "step": 1064
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.8577648766328013e-06,
-      "loss": 3.9612,
-      "step": 1065
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.8432510885341077e-06,
-      "loss": 4.0423,
-      "step": 1066
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.8287373004354138e-06,
-      "loss": 4.1094,
-      "step": 1067
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.81422351233672e-06,
-      "loss": 4.0842,
-      "step": 1068
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.7997097242380262e-06,
-      "loss": 4.1169,
-      "step": 1069
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.7851959361393326e-06,
-      "loss": 4.2625,
-      "step": 1070
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.7706821480406388e-06,
-      "loss": 4.1601,
-      "step": 1071
+      "learning_rate": 7.021276595744682e-06,
+      "loss": 4.0396,
+      "step": 531
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.7561683599419448e-06,
-      "loss": 4.1046,
-      "step": 1072
+      "epoch": 0.89,
+      "learning_rate": 6.914893617021278e-06,
+      "loss": 4.2931,
+      "step": 532
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.7416545718432512e-06,
-      "loss": 3.8769,
-      "step": 1073
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 4.1061,
+      "step": 533
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.7271407837445576e-06,
-      "loss": 4.016,
-      "step": 1074
+      "learning_rate": 6.702127659574469e-06,
+      "loss": 4.124,
+      "step": 534
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.7126269956458638e-06,
-      "loss": 3.961,
-      "step": 1075
+      "learning_rate": 6.595744680851064e-06,
+      "loss": 4.149,
+      "step": 535
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.6981132075471698e-06,
-      "loss": 4.0277,
-      "step": 1076
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.6835994194484762e-06,
-      "loss": 4.2872,
-      "step": 1077
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.6690856313497824e-06,
-      "loss": 4.1383,
-      "step": 1078
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.6545718432510888e-06,
-      "loss": 4.136,
-      "step": 1079
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.6400580551523947e-06,
-      "loss": 3.9465,
-      "step": 1080
+      "learning_rate": 6.48936170212766e-06,
+      "loss": 4.0031,
+      "step": 536
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.6255442670537012e-06,
-      "loss": 4.2499,
-      "step": 1081
+      "epoch": 0.9,
+      "learning_rate": 6.382978723404256e-06,
+      "loss": 4.2344,
+      "step": 537
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.6110304789550073e-06,
-      "loss": 4.2032,
-      "step": 1082
+      "epoch": 0.9,
+      "learning_rate": 6.276595744680851e-06,
+      "loss": 4.2397,
+      "step": 538
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5965166908563137e-06,
-      "loss": 4.0541,
-      "step": 1083
+      "learning_rate": 6.170212765957447e-06,
+      "loss": 4.0717,
+      "step": 539
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5820029027576197e-06,
-      "loss": 4.1248,
-      "step": 1084
+      "learning_rate": 6.063829787234044e-06,
+      "loss": 4.156,
+      "step": 540
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5674891146589261e-06,
-      "loss": 4.212,
-      "step": 1085
+      "learning_rate": 5.957446808510638e-06,
+      "loss": 4.151,
+      "step": 541
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5529753265602323e-06,
-      "loss": 4.0355,
-      "step": 1086
+      "learning_rate": 5.851063829787235e-06,
+      "loss": 4.1377,
+      "step": 542
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5384615384615387e-06,
-      "loss": 4.2155,
-      "step": 1087
+      "learning_rate": 5.744680851063831e-06,
+      "loss": 4.3309,
+      "step": 543
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5239477503628447e-06,
-      "loss": 3.9952,
-      "step": 1088
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.509433962264151e-06,
-      "loss": 4.0712,
-      "step": 1089
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4949201741654573e-06,
-      "loss": 4.115,
-      "step": 1090
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4804063860667637e-06,
-      "loss": 4.1191,
-      "step": 1091
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4658925979680697e-06,
-      "loss": 4.1495,
-      "step": 1092
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4513788098693759e-06,
-      "loss": 4.1395,
-      "step": 1093
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4368650217706823e-06,
-      "loss": 4.2486,
-      "step": 1094
+      "learning_rate": 5.638297872340426e-06,
+      "loss": 4.161,
+      "step": 544
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.4223512336719887e-06,
-      "loss": 4.2146,
-      "step": 1095
+      "learning_rate": 5.531914893617022e-06,
+      "loss": 4.2825,
+      "step": 545
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.4078374455732949e-06,
-      "loss": 4.6284,
-      "step": 1096
+      "learning_rate": 5.425531914893617e-06,
+      "loss": 4.4817,
+      "step": 546
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3933236574746008e-06,
-      "loss": 4.4161,
-      "step": 1097
+      "learning_rate": 5.319148936170213e-06,
+      "loss": 4.2491,
+      "step": 547
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3788098693759072e-06,
-      "loss": 4.428,
-      "step": 1098
+      "learning_rate": 5.212765957446809e-06,
+      "loss": 4.5033,
+      "step": 548
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3642960812772134e-06,
-      "loss": 4.6127,
-      "step": 1099
+      "learning_rate": 5.106382978723404e-06,
+      "loss": 4.6914,
+      "step": 549
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3497822931785198e-06,
-      "loss": 4.3368,
-      "step": 1100
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.3352685050798258e-06,
-      "loss": 4.306,
-      "step": 1101
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.3207547169811322e-06,
-      "loss": 4.1164,
-      "step": 1102
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.3062409288824384e-06,
-      "loss": 4.0594,
-      "step": 1103
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.2917271407837448e-06,
-      "loss": 3.9618,
-      "step": 1104
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.2772133526850508e-06,
-      "loss": 4.0961,
-      "step": 1105
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.2626995645863572e-06,
-      "loss": 3.938,
-      "step": 1106
+      "learning_rate": 5e-06,
+      "loss": 4.3848,
+      "step": 550
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.2481857764876634e-06,
-      "loss": 4.3054,
-      "step": 1107
+      "learning_rate": 4.893617021276596e-06,
+      "loss": 4.1878,
+      "step": 551
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.2336719883889696e-06,
-      "loss": 4.2445,
-      "step": 1108
+      "learning_rate": 4.787234042553192e-06,
+      "loss": 4.1017,
+      "step": 552
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.219158200290276e-06,
-      "loss": 4.1992,
-      "step": 1109
+      "learning_rate": 4.680851063829788e-06,
+      "loss": 4.091,
+      "step": 553
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.2046444121915822e-06,
-      "loss": 4.0786,
-      "step": 1110
+      "learning_rate": 4.574468085106383e-06,
+      "loss": 4.0357,
+      "step": 554
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.1901306240928884e-06,
-      "loss": 4.061,
-      "step": 1111
+      "learning_rate": 4.468085106382979e-06,
+      "loss": 4.0749,
+      "step": 555
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.1756168359941945e-06,
-      "loss": 4.4113,
-      "step": 1112
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.161103047895501e-06,
-      "loss": 4.1809,
-      "step": 1113
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.146589259796807e-06,
-      "loss": 3.8364,
-      "step": 1114
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.1320754716981133e-06,
-      "loss": 4.1568,
-      "step": 1115
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.1175616835994195e-06,
-      "loss": 4.2159,
-      "step": 1116
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.1030478955007257e-06,
-      "loss": 4.1667,
-      "step": 1117
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.088534107402032e-06,
-      "loss": 3.9982,
-      "step": 1118
+      "learning_rate": 4.361702127659575e-06,
+      "loss": 4.106,
+      "step": 556
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0740203193033383e-06,
-      "loss": 3.8337,
-      "step": 1119
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 4.1974,
+      "step": 557
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0595065312046445e-06,
-      "loss": 4.1695,
-      "step": 1120
+      "learning_rate": 4.148936170212766e-06,
+      "loss": 4.2221,
+      "step": 558
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0449927431059507e-06,
-      "loss": 4.1488,
-      "step": 1121
+      "learning_rate": 4.042553191489362e-06,
+      "loss": 4.0609,
+      "step": 559
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0304789550072569e-06,
-      "loss": 4.1877,
-      "step": 1122
+      "learning_rate": 3.936170212765958e-06,
+      "loss": 4.1074,
+      "step": 560
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0159651669085633e-06,
-      "loss": 4.0255,
-      "step": 1123
+      "learning_rate": 3.8297872340425535e-06,
+      "loss": 4.4338,
+      "step": 561
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0014513788098695e-06,
-      "loss": 4.1761,
-      "step": 1124
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.869375907111757e-07,
-      "loss": 4.3559,
-      "step": 1125
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.724238026124818e-07,
-      "loss": 4.1163,
-      "step": 1126
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.579100145137882e-07,
-      "loss": 4.2633,
-      "step": 1127
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.433962264150944e-07,
-      "loss": 4.0815,
-      "step": 1128
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.288824383164006e-07,
-      "loss": 4.1479,
-      "step": 1129
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.143686502177069e-07,
-      "loss": 4.2254,
-      "step": 1130
+      "learning_rate": 3.723404255319149e-06,
+      "loss": 4.041,
+      "step": 562
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.998548621190131e-07,
-      "loss": 3.9276,
-      "step": 1131
+      "learning_rate": 3.6170212765957453e-06,
+      "loss": 4.0901,
+      "step": 563
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.853410740203194e-07,
-      "loss": 3.9613,
-      "step": 1132
+      "learning_rate": 3.510638297872341e-06,
+      "loss": 4.1839,
+      "step": 564
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.708272859216256e-07,
-      "loss": 4.2874,
-      "step": 1133
+      "learning_rate": 3.4042553191489363e-06,
+      "loss": 4.1574,
+      "step": 565
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.563134978229319e-07,
-      "loss": 4.1779,
-      "step": 1134
+      "learning_rate": 3.297872340425532e-06,
+      "loss": 4.0428,
+      "step": 566
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.417997097242381e-07,
-      "loss": 4.3097,
-      "step": 1135
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.272859216255444e-07,
-      "loss": 4.0715,
-      "step": 1136
+      "learning_rate": 3.191489361702128e-06,
+      "loss": 4.1384,
+      "step": 567
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.127721335268506e-07,
+      "learning_rate": 3.0851063829787237e-06,
       "loss": 4.0906,
-      "step": 1137
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.982583454281569e-07,
-      "loss": 4.1087,
-      "step": 1138
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.837445573294631e-07,
-      "loss": 4.0879,
-      "step": 1139
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.692307692307694e-07,
-      "loss": 3.9786,
-      "step": 1140
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.547169811320755e-07,
-      "loss": 4.3693,
-      "step": 1141
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.402031930333818e-07,
-      "loss": 4.1353,
-      "step": 1142
+      "step": 568
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.256894049346879e-07,
-      "loss": 4.3544,
-      "step": 1143
+      "learning_rate": 2.978723404255319e-06,
+      "loss": 3.9581,
+      "step": 569
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.111756168359943e-07,
-      "loss": 4.3842,
-      "step": 1144
+      "learning_rate": 2.8723404255319155e-06,
+      "loss": 4.1344,
+      "step": 570
     },
     {
       "epoch": 0.96,
-      "learning_rate": 6.966618287373004e-07,
-      "loss": 4.128,
-      "step": 1145
+      "learning_rate": 2.765957446808511e-06,
+      "loss": 4.1249,
+      "step": 571
     },
     {
       "epoch": 0.96,
-      "learning_rate": 6.821480406386067e-07,
-      "loss": 4.4319,
-      "step": 1146
+      "learning_rate": 2.6595744680851065e-06,
+      "loss": 4.1715,
+      "step": 572
     },
     {
       "epoch": 0.96,
-      "learning_rate": 6.676342525399129e-07,
-      "loss": 4.3995,
-      "step": 1147
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.531204644412192e-07,
-      "loss": 4.277,
-      "step": 1148
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.386066763425254e-07,
-      "loss": 4.6563,
-      "step": 1149
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.240928882438317e-07,
-      "loss": 4.3997,
-      "step": 1150
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.09579100145138e-07,
-      "loss": 4.0219,
-      "step": 1151
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.950653120464442e-07,
-      "loss": 3.9859,
-      "step": 1152
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.805515239477505e-07,
-      "loss": 4.0447,
-      "step": 1153
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 4.2385,
+      "step": 573
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.660377358490567e-07,
-      "loss": 3.9644,
-      "step": 1154
+      "learning_rate": 2.446808510638298e-06,
+      "loss": 4.1462,
+      "step": 574
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.515239477503629e-07,
-      "loss": 4.1066,
-      "step": 1155
+      "learning_rate": 2.340425531914894e-06,
+      "loss": 4.258,
+      "step": 575
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.370101596516691e-07,
-      "loss": 4.1409,
-      "step": 1156
+      "learning_rate": 2.2340425531914894e-06,
+      "loss": 4.1278,
+      "step": 576
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.224963715529753e-07,
-      "loss": 3.9766,
-      "step": 1157
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 4.1169,
+      "step": 577
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.079825834542816e-07,
-      "loss": 4.1354,
-      "step": 1158
+      "learning_rate": 2.021276595744681e-06,
+      "loss": 4.0495,
+      "step": 578
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.934687953555878e-07,
-      "loss": 4.0027,
-      "step": 1159
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.789550072568941e-07,
-      "loss": 4.2344,
-      "step": 1160
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.644412191582003e-07,
-      "loss": 3.9541,
-      "step": 1161
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.4992743105950656e-07,
-      "loss": 4.2362,
-      "step": 1162
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.354136429608128e-07,
-      "loss": 4.1806,
-      "step": 1163
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.2089985486211904e-07,
-      "loss": 4.0122,
-      "step": 1164
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.063860667634253e-07,
-      "loss": 3.9897,
-      "step": 1165
+      "learning_rate": 1.9148936170212767e-06,
+      "loss": 4.004,
+      "step": 579
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.9187227866473153e-07,
-      "loss": 4.2314,
-      "step": 1166
+      "learning_rate": 1.8085106382978727e-06,
+      "loss": 4.2692,
+      "step": 580
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.773584905660378e-07,
-      "loss": 3.9178,
-      "step": 1167
+      "learning_rate": 1.7021276595744682e-06,
+      "loss": 4.3664,
+      "step": 581
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.6284470246734397e-07,
-      "loss": 4.0088,
-      "step": 1168
+      "learning_rate": 1.595744680851064e-06,
+      "loss": 4.2793,
+      "step": 582
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.483309143686502e-07,
-      "loss": 4.0474,
-      "step": 1169
+      "learning_rate": 1.4893617021276596e-06,
+      "loss": 4.0669,
+      "step": 583
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.3381712626995645e-07,
-      "loss": 4.1257,
-      "step": 1170
+      "learning_rate": 1.3829787234042555e-06,
+      "loss": 4.2035,
+      "step": 584
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.193033381712627e-07,
-      "loss": 4.1431,
-      "step": 1171
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.04789550072569e-07,
-      "loss": 4.0051,
-      "step": 1172
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.9027576197387524e-07,
-      "loss": 4.2092,
-      "step": 1173
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.757619738751814e-07,
-      "loss": 4.0239,
-      "step": 1174
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.6124818577648767e-07,
-      "loss": 3.9977,
-      "step": 1175
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.467343976777939e-07,
-      "loss": 4.0475,
-      "step": 1176
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.3222060957910016e-07,
-      "loss": 4.2143,
-      "step": 1177
+      "learning_rate": 1.276595744680851e-06,
+      "loss": 4.1755,
+      "step": 585
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.177068214804064e-07,
-      "loss": 4.4889,
-      "step": 1178
+      "learning_rate": 1.170212765957447e-06,
+      "loss": 4.0676,
+      "step": 586
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.0319303338171264e-07,
-      "loss": 4.0366,
-      "step": 1179
+      "learning_rate": 1.0638297872340427e-06,
+      "loss": 4.2543,
+      "step": 587
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.886792452830189e-07,
-      "loss": 4.2305,
-      "step": 1180
+      "learning_rate": 9.574468085106384e-07,
+      "loss": 4.2331,
+      "step": 588
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.741654571843251e-07,
-      "loss": 4.1438,
-      "step": 1181
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 4.3553,
+      "step": 589
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.5965166908563135e-07,
-      "loss": 4.1192,
-      "step": 1182
+      "learning_rate": 7.446808510638298e-07,
+      "loss": 4.2031,
+      "step": 590
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.4513788098693762e-07,
-      "loss": 4.1408,
-      "step": 1183
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.3062409288824383e-07,
-      "loss": 4.1181,
-      "step": 1184
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.1611030478955008e-07,
-      "loss": 4.1789,
-      "step": 1185
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.0159651669085632e-07,
-      "loss": 4.3983,
-      "step": 1186
+      "learning_rate": 6.382978723404255e-07,
+      "loss": 4.4084,
+      "step": 591
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.708272859216255e-08,
-      "loss": 4.2023,
-      "step": 1187
+      "learning_rate": 5.319148936170213e-07,
+      "loss": 4.2994,
+      "step": 592
     },
     {
       "epoch": 1.0,
-      "learning_rate": 7.256894049346881e-08,
-      "loss": 4.3103,
-      "step": 1188
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 4.5066,
+      "step": 593
     },
     {
       "epoch": 1.0,
-      "learning_rate": 5.805515239477504e-08,
-      "loss": 4.4211,
-      "step": 1189
+      "learning_rate": 3.1914893617021275e-07,
+      "loss": 4.4907,
+      "step": 594
     },
     {
       "epoch": 1.0,
-      "step": 1189,
+      "step": 594,
       "total_flos": 0.0,
-      "train_loss": 4.268135019066556,
-      "train_runtime": 6757.0177,
-      "train_samples_per_second": 4.223,
-      "train_steps_per_second": 0.176
+      "train_loss": 4.336673439953865,
+      "train_runtime": 5718.4377,
+      "train_samples_per_second": 4.991,
+      "train_steps_per_second": 0.104
     }
   ],
-  "max_steps": 1189,
+  "max_steps": 594,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,