diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,1805 +1,10754 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9987389659520807,
-  "global_step": 297,
+  "epoch": 1.0,
+  "global_step": 1784,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 2e-08,
-      "loss": 4.7862,
+      "learning_rate": 0.0,
+      "loss": 4.915,
       "step": 1
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2e-08,
-      "loss": 4.8712,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.0319,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4e-08,
-      "loss": 4.9393,
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 5.1506,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 4.8916,
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.6824,
       "step": 4
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8e-08,
-      "loss": 4.7806,
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 4.7852,
       "step": 5
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 4.9123,
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.8191,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.8899,
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.6997,
       "step": 7
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.4e-07,
-      "loss": 4.9651,
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 4.8571,
       "step": 8
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.6e-07,
-      "loss": 4.8742,
+      "epoch": 0.01,
+      "learning_rate": 4.2e-07,
+      "loss": 5.0588,
       "step": 9
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.8e-07,
-      "loss": 4.749,
+      "epoch": 0.01,
+      "learning_rate": 4.2e-07,
+      "loss": 4.7518,
       "step": 10
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 4.8902,
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.8048,
       "step": 11
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.2e-07,
-      "loss": 4.9384,
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 4.7803,
       "step": 12
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.8481,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.9879,
       "step": 13
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.6e-07,
-      "loss": 4.8025,
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 4.6087,
       "step": 14
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.8e-07,
-      "loss": 4.8327,
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 4.7458,
       "step": 15
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.7626,
+      "epoch": 0.01,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 4.8735,
       "step": 16
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.2e-07,
-      "loss": 4.8658,
+      "epoch": 0.01,
+      "learning_rate": 8.4e-07,
+      "loss": 4.8776,
       "step": 17
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 4.8883,
+      "epoch": 0.01,
+      "learning_rate": 9e-07,
+      "loss": 4.7647,
       "step": 18
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.6e-07,
-      "loss": 4.8363,
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 4.7182,
       "step": 19
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.8e-07,
-      "loss": 4.9294,
+      "epoch": 0.01,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 5.0357,
       "step": 20
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.9403,
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 4.7483,
       "step": 21
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 4.8762,
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 5.063,
       "step": 22
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.4e-07,
-      "loss": 4.8012,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.9108,
       "step": 23
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.6000000000000004e-07,
-      "loss": 4.9476,
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 4.6387,
       "step": 24
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 4.7731,
+      "epoch": 0.01,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 4.8228,
       "step": 25
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 5.086,
+      "epoch": 0.01,
+      "learning_rate": 1.38e-06,
+      "loss": 4.9225,
       "step": 26
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 5.2e-07,
-      "loss": 4.8115,
+      "epoch": 0.02,
+      "learning_rate": 1.44e-06,
+      "loss": 4.7585,
       "step": 27
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 5.4e-07,
-      "loss": 4.8335,
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 4.4815,
       "step": 28
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.6e-07,
-      "loss": 4.7309,
+      "epoch": 0.02,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 4.8255,
       "step": 29
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.800000000000001e-07,
-      "loss": 4.8824,
+      "epoch": 0.02,
+      "learning_rate": 1.62e-06,
+      "loss": 4.6586,
       "step": 30
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.875,
+      "epoch": 0.02,
+      "learning_rate": 1.68e-06,
+      "loss": 4.5517,
       "step": 31
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 6.200000000000001e-07,
-      "loss": 4.8,
+      "epoch": 0.02,
+      "learning_rate": 1.74e-06,
+      "loss": 4.9114,
       "step": 32
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 6.4e-07,
-      "loss": 4.8361,
+      "epoch": 0.02,
+      "learning_rate": 1.8e-06,
+      "loss": 4.6263,
       "step": 33
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 6.6e-07,
-      "loss": 4.7767,
+      "epoch": 0.02,
+      "learning_rate": 1.86e-06,
+      "loss": 4.5165,
       "step": 34
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.800000000000001e-07,
-      "loss": 4.8105,
+      "epoch": 0.02,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 4.7293,
       "step": 35
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 7.000000000000001e-07,
-      "loss": 4.7742,
+      "epoch": 0.02,
+      "learning_rate": 1.98e-06,
+      "loss": 4.9974,
       "step": 36
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 7.2e-07,
-      "loss": 4.9031,
+      "epoch": 0.02,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 4.6251,
       "step": 37
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 7.4e-07,
-      "loss": 4.8678,
+      "epoch": 0.02,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.7255,
       "step": 38
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 7.6e-07,
-      "loss": 4.9272,
+      "epoch": 0.02,
+      "learning_rate": 2.16e-06,
+      "loss": 4.8292,
       "step": 39
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 7.8e-07,
-      "loss": 4.764,
+      "epoch": 0.02,
+      "learning_rate": 2.22e-06,
+      "loss": 4.81,
       "step": 40
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.8599,
+      "epoch": 0.02,
+      "learning_rate": 2.28e-06,
+      "loss": 4.622,
       "step": 41
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 8.200000000000001e-07,
-      "loss": 4.9507,
+      "epoch": 0.02,
+      "learning_rate": 2.34e-06,
+      "loss": 4.4237,
       "step": 42
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 8.400000000000001e-07,
-      "loss": 4.7209,
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.7165,
       "step": 43
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 8.6e-07,
-      "loss": 5.0393,
+      "epoch": 0.02,
+      "learning_rate": 2.46e-06,
+      "loss": 4.5473,
       "step": 44
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 8.8e-07,
-      "loss": 5.0139,
+      "epoch": 0.03,
+      "learning_rate": 2.52e-06,
+      "loss": 4.8144,
       "step": 45
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 9.000000000000001e-07,
-      "loss": 5.0851,
+      "epoch": 0.03,
+      "learning_rate": 2.58e-06,
+      "loss": 4.7817,
       "step": 46
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.1161,
+      "epoch": 0.03,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 4.8919,
       "step": 47
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.400000000000001e-07,
-      "loss": 5.2155,
+      "epoch": 0.03,
+      "learning_rate": 2.7e-06,
+      "loss": 5.3074,
       "step": 48
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 5.2669,
+      "epoch": 0.03,
+      "learning_rate": 2.76e-06,
+      "loss": 5.303,
       "step": 49
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.800000000000001e-07,
-      "loss": 5.1875,
+      "epoch": 0.03,
+      "learning_rate": 2.82e-06,
+      "loss": 5.3018,
       "step": 50
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.673,
+      "epoch": 0.03,
+      "learning_rate": 2.88e-06,
+      "loss": 4.4273,
       "step": 51
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.02e-06,
-      "loss": 4.5594,
+      "epoch": 0.03,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.569,
       "step": 52
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.04e-06,
-      "loss": 4.5074,
+      "epoch": 0.03,
+      "learning_rate": 3e-06,
+      "loss": 4.5515,
       "step": 53
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.06e-06,
-      "loss": 4.5297,
+      "epoch": 0.03,
+      "learning_rate": 3.06e-06,
+      "loss": 4.4204,
       "step": 54
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.08e-06,
-      "loss": 4.4645,
+      "epoch": 0.03,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 4.4619,
       "step": 55
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.1e-06,
-      "loss": 4.4743,
+      "epoch": 0.03,
+      "learning_rate": 3.18e-06,
+      "loss": 4.582,
       "step": 56
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.12e-06,
-      "loss": 4.6112,
+      "epoch": 0.03,
+      "learning_rate": 3.24e-06,
+      "loss": 4.4675,
       "step": 57
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.14e-06,
-      "loss": 4.6401,
+      "epoch": 0.03,
+      "learning_rate": 3.3e-06,
+      "loss": 4.4125,
       "step": 58
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.1600000000000001e-06,
-      "loss": 4.6055,
+      "epoch": 0.03,
+      "learning_rate": 3.36e-06,
+      "loss": 4.4569,
       "step": 59
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.1800000000000001e-06,
-      "loss": 4.5181,
+      "epoch": 0.03,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 4.4054,
       "step": 60
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.5917,
+      "epoch": 0.03,
+      "learning_rate": 3.48e-06,
+      "loss": 4.4286,
       "step": 61
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.2200000000000002e-06,
-      "loss": 4.4533,
+      "epoch": 0.03,
+      "learning_rate": 3.54e-06,
+      "loss": 4.3546,
       "step": 62
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.2400000000000002e-06,
-      "loss": 4.5889,
+      "epoch": 0.04,
+      "learning_rate": 3.6e-06,
+      "loss": 4.711,
       "step": 63
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.26e-06,
-      "loss": 4.5351,
+      "epoch": 0.04,
+      "learning_rate": 3.66e-06,
+      "loss": 4.2235,
       "step": 64
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.28e-06,
-      "loss": 4.5995,
+      "epoch": 0.04,
+      "learning_rate": 3.72e-06,
+      "loss": 4.4544,
       "step": 65
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.3e-06,
-      "loss": 4.5842,
+      "epoch": 0.04,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.532,
       "step": 66
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.32e-06,
-      "loss": 4.5496,
+      "epoch": 0.04,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.5484,
       "step": 67
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.34e-06,
-      "loss": 4.5796,
+      "epoch": 0.04,
+      "learning_rate": 3.9e-06,
+      "loss": 4.5452,
       "step": 68
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.3600000000000001e-06,
-      "loss": 4.5204,
+      "epoch": 0.04,
+      "learning_rate": 3.96e-06,
+      "loss": 4.4212,
       "step": 69
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.3800000000000001e-06,
-      "loss": 4.4595,
+      "epoch": 0.04,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.3737,
       "step": 70
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.4492,
+      "epoch": 0.04,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 4.4685,
       "step": 71
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.42e-06,
-      "loss": 4.5821,
+      "epoch": 0.04,
+      "learning_rate": 4.14e-06,
+      "loss": 4.222,
       "step": 72
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.44e-06,
-      "loss": 4.4998,
+      "epoch": 0.04,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.4547,
       "step": 73
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.46e-06,
-      "loss": 4.4812,
+      "epoch": 0.04,
+      "learning_rate": 4.26e-06,
+      "loss": 4.2145,
       "step": 74
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.48e-06,
-      "loss": 4.5778,
+      "epoch": 0.04,
+      "learning_rate": 4.32e-06,
+      "loss": 4.7721,
       "step": 75
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.5e-06,
-      "loss": 4.4696,
+      "epoch": 0.04,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 4.5625,
       "step": 76
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.52e-06,
-      "loss": 4.4308,
+      "epoch": 0.04,
+      "learning_rate": 4.44e-06,
+      "loss": 4.3557,
       "step": 77
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.54e-06,
-      "loss": 4.4694,
+      "epoch": 0.04,
+      "learning_rate": 4.5e-06,
+      "loss": 4.2864,
       "step": 78
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.56e-06,
-      "loss": 4.3769,
+      "epoch": 0.04,
+      "learning_rate": 4.56e-06,
+      "loss": 4.279,
       "step": 79
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.5800000000000001e-06,
-      "loss": 4.5486,
+      "epoch": 0.04,
+      "learning_rate": 4.62e-06,
+      "loss": 4.6199,
       "step": 80
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.4755,
+      "epoch": 0.05,
+      "learning_rate": 4.68e-06,
+      "loss": 4.2361,
       "step": 81
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6200000000000002e-06,
-      "loss": 4.4662,
+      "epoch": 0.05,
+      "learning_rate": 4.74e-06,
+      "loss": 4.3772,
       "step": 82
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6400000000000002e-06,
-      "loss": 4.4412,
+      "epoch": 0.05,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.3099,
       "step": 83
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6600000000000002e-06,
-      "loss": 4.4749,
+      "epoch": 0.05,
+      "learning_rate": 4.86e-06,
+      "loss": 4.3713,
       "step": 84
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6800000000000002e-06,
-      "loss": 4.5525,
+      "epoch": 0.05,
+      "learning_rate": 4.92e-06,
+      "loss": 4.2608,
       "step": 85
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.7000000000000002e-06,
-      "loss": 4.4525,
+      "epoch": 0.05,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.2858,
       "step": 86
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.72e-06,
-      "loss": 4.5006,
+      "epoch": 0.05,
+      "learning_rate": 5.04e-06,
+      "loss": 4.6258,
       "step": 87
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.74e-06,
-      "loss": 4.5137,
+      "epoch": 0.05,
+      "learning_rate": 5.1e-06,
+      "loss": 4.2484,
       "step": 88
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.76e-06,
-      "loss": 4.549,
+      "epoch": 0.05,
+      "learning_rate": 5.16e-06,
+      "loss": 4.606,
       "step": 89
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.7800000000000001e-06,
-      "loss": 4.5675,
+      "epoch": 0.05,
+      "learning_rate": 5.22e-06,
+      "loss": 4.1498,
       "step": 90
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.8000000000000001e-06,
-      "loss": 4.595,
+      "epoch": 0.05,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 4.5393,
       "step": 91
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.8200000000000002e-06,
-      "loss": 4.6441,
+      "epoch": 0.05,
+      "learning_rate": 5.34e-06,
+      "loss": 4.209,
       "step": 92
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.8400000000000002e-06,
-      "loss": 4.5773,
+      "epoch": 0.05,
+      "learning_rate": 5.4e-06,
+      "loss": 4.0755,
       "step": 93
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.8600000000000002e-06,
-      "loss": 4.522,
+      "epoch": 0.05,
+      "learning_rate": 5.46e-06,
+      "loss": 4.5134,
       "step": 94
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 4.4908,
+      "epoch": 0.05,
+      "learning_rate": 5.52e-06,
+      "loss": 4.7309,
       "step": 95
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9000000000000002e-06,
-      "loss": 4.6946,
+      "epoch": 0.05,
+      "learning_rate": 5.58e-06,
+      "loss": 4.2903,
       "step": 96
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.8747,
+      "epoch": 0.05,
+      "learning_rate": 5.64e-06,
+      "loss": 4.4812,
       "step": 97
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.94e-06,
-      "loss": 4.8247,
+      "epoch": 0.05,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 4.5923,
       "step": 98
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9600000000000003e-06,
-      "loss": 4.786,
+      "epoch": 0.06,
+      "learning_rate": 5.76e-06,
+      "loss": 4.9472,
       "step": 99
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.98e-06,
-      "loss": 4.8753,
+      "epoch": 0.06,
+      "learning_rate": 5.76e-06,
+      "loss": 4.7627,
       "step": 100
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.332,
+      "epoch": 0.06,
+      "learning_rate": 5.82e-06,
+      "loss": 4.372,
       "step": 101
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.02e-06,
-      "loss": 4.3795,
+      "epoch": 0.06,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.1729,
       "step": 102
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.04e-06,
-      "loss": 4.3807,
+      "epoch": 0.06,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 4.2992,
       "step": 103
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.06e-06,
-      "loss": 4.2868,
+      "epoch": 0.06,
+      "learning_rate": 6e-06,
+      "loss": 4.2621,
       "step": 104
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.08e-06,
-      "loss": 4.3938,
+      "epoch": 0.06,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.3061,
       "step": 105
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.3391,
+      "epoch": 0.06,
+      "learning_rate": 6.12e-06,
+      "loss": 4.287,
       "step": 106
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.12e-06,
-      "loss": 4.2794,
+      "epoch": 0.06,
+      "learning_rate": 6.18e-06,
+      "loss": 4.3187,
       "step": 107
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.1400000000000003e-06,
-      "loss": 4.3783,
+      "epoch": 0.06,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 4.395,
       "step": 108
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.16e-06,
-      "loss": 4.382,
+      "epoch": 0.06,
+      "learning_rate": 6.3e-06,
+      "loss": 4.2955,
       "step": 109
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.1800000000000003e-06,
-      "loss": 4.3473,
+      "epoch": 0.06,
+      "learning_rate": 6.36e-06,
+      "loss": 4.3274,
       "step": 110
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.2e-06,
-      "loss": 4.4273,
+      "epoch": 0.06,
+      "learning_rate": 6.42e-06,
+      "loss": 4.2659,
       "step": 111
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.2200000000000003e-06,
-      "loss": 4.3599,
+      "epoch": 0.06,
+      "learning_rate": 6.48e-06,
+      "loss": 4.1848,
       "step": 112
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.24e-06,
-      "loss": 4.2758,
+      "epoch": 0.06,
+      "learning_rate": 6.54e-06,
+      "loss": 4.3092,
       "step": 113
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.2600000000000004e-06,
-      "loss": 4.3514,
+      "epoch": 0.06,
+      "learning_rate": 6.6e-06,
+      "loss": 4.0276,
       "step": 114
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.28e-06,
-      "loss": 4.3079,
+      "epoch": 0.06,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.0794,
       "step": 115
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.3000000000000004e-06,
-      "loss": 4.3189,
+      "epoch": 0.07,
+      "learning_rate": 6.72e-06,
+      "loss": 4.2448,
       "step": 116
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.3200000000000002e-06,
-      "loss": 4.2784,
+      "epoch": 0.07,
+      "learning_rate": 6.78e-06,
+      "loss": 4.3882,
       "step": 117
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.3400000000000005e-06,
-      "loss": 4.3314,
+      "epoch": 0.07,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 4.239,
       "step": 118
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.3600000000000003e-06,
-      "loss": 4.3653,
+      "epoch": 0.07,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 4.3298,
       "step": 119
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.38e-06,
-      "loss": 4.36,
+      "epoch": 0.07,
+      "learning_rate": 6.96e-06,
+      "loss": 4.4041,
       "step": 120
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.3902,
+      "epoch": 0.07,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.4322,
       "step": 121
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.42e-06,
-      "loss": 4.4351,
+      "epoch": 0.07,
+      "learning_rate": 7.08e-06,
+      "loss": 4.2802,
       "step": 122
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4400000000000004e-06,
-      "loss": 4.2534,
+      "epoch": 0.07,
+      "learning_rate": 7.14e-06,
+      "loss": 4.4441,
       "step": 123
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.46e-06,
-      "loss": 4.3887,
+      "epoch": 0.07,
+      "learning_rate": 7.2e-06,
+      "loss": 4.344,
       "step": 124
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4800000000000004e-06,
-      "loss": 4.4345,
+      "epoch": 0.07,
+      "learning_rate": 7.26e-06,
+      "loss": 4.0195,
       "step": 125
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.5e-06,
-      "loss": 4.3609,
+      "epoch": 0.07,
+      "learning_rate": 7.32e-06,
+      "loss": 4.6502,
       "step": 126
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.52e-06,
-      "loss": 4.4055,
+      "epoch": 0.07,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.3387,
       "step": 127
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.5400000000000002e-06,
-      "loss": 4.2582,
+      "epoch": 0.07,
+      "learning_rate": 7.44e-06,
+      "loss": 4.3012,
       "step": 128
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.56e-06,
-      "loss": 4.4031,
+      "epoch": 0.07,
+      "learning_rate": 7.5e-06,
+      "loss": 4.1813,
       "step": 129
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.5800000000000003e-06,
-      "loss": 4.4855,
+      "epoch": 0.07,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.2588,
       "step": 130
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.6e-06,
-      "loss": 4.4102,
+      "epoch": 0.07,
+      "learning_rate": 7.62e-06,
+      "loss": 4.5306,
       "step": 131
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.6200000000000003e-06,
-      "loss": 4.2256,
+      "epoch": 0.07,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.2289,
       "step": 132
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 2.64e-06,
-      "loss": 4.4007,
+      "epoch": 0.07,
+      "learning_rate": 7.74e-06,
+      "loss": 4.3241,
       "step": 133
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 2.6600000000000004e-06,
-      "loss": 4.2806,
+      "epoch": 0.08,
+      "learning_rate": 7.8e-06,
+      "loss": 4.2998,
       "step": 134
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 2.68e-06,
-      "loss": 4.4708,
+      "epoch": 0.08,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.2204,
       "step": 135
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 2.7000000000000004e-06,
-      "loss": 4.3876,
+      "epoch": 0.08,
+      "learning_rate": 7.92e-06,
+      "loss": 4.3853,
       "step": 136
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 2.7200000000000002e-06,
-      "loss": 4.3317,
+      "epoch": 0.08,
+      "learning_rate": 7.98e-06,
+      "loss": 4.1586,
       "step": 137
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 2.7400000000000004e-06,
-      "loss": 4.2371,
+      "epoch": 0.08,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 4.3613,
       "step": 138
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.7600000000000003e-06,
-      "loss": 4.2725,
+      "epoch": 0.08,
+      "learning_rate": 8.1e-06,
+      "loss": 4.486,
       "step": 139
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.7800000000000005e-06,
-      "loss": 4.3226,
+      "epoch": 0.08,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 4.1409,
       "step": 140
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.5303,
+      "epoch": 0.08,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 4.3727,
       "step": 141
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 2.82e-06,
-      "loss": 4.2812,
+      "epoch": 0.08,
+      "learning_rate": 8.28e-06,
+      "loss": 4.6136,
       "step": 142
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 2.84e-06,
-      "loss": 4.3971,
+      "epoch": 0.08,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 4.2883,
       "step": 143
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 2.86e-06,
-      "loss": 4.426,
+      "epoch": 0.08,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.3931,
       "step": 144
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.88e-06,
-      "loss": 4.4879,
+      "epoch": 0.08,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 4.632,
       "step": 145
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.9e-06,
-      "loss": 4.3671,
+      "epoch": 0.08,
+      "learning_rate": 8.52e-06,
+      "loss": 4.4625,
       "step": 146
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.92e-06,
-      "loss": 4.4797,
+      "epoch": 0.08,
+      "learning_rate": 8.58e-06,
+      "loss": 4.4465,
       "step": 147
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.6502,
+      "epoch": 0.08,
+      "learning_rate": 8.64e-06,
+      "loss": 4.5748,
       "step": 148
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.96e-06,
-      "loss": 4.6188,
+      "epoch": 0.08,
+      "learning_rate": 8.7e-06,
+      "loss": 4.7479,
       "step": 149
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.9800000000000003e-06,
-      "loss": 4.7733,
+      "epoch": 0.08,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 4.1989,
       "step": 150
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 3e-06,
-      "loss": 4.3361,
+      "epoch": 0.08,
+      "learning_rate": 8.82e-06,
+      "loss": 4.0333,
       "step": 151
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 3.0200000000000003e-06,
-      "loss": 4.3216,
+      "epoch": 0.09,
+      "learning_rate": 8.88e-06,
+      "loss": 4.375,
       "step": 152
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 3.04e-06,
-      "loss": 4.3243,
+      "epoch": 0.09,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 4.0024,
       "step": 153
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 3.0600000000000003e-06,
-      "loss": 4.2681,
+      "epoch": 0.09,
+      "learning_rate": 9e-06,
+      "loss": 4.4603,
       "step": 154
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 3.08e-06,
-      "loss": 4.2981,
+      "epoch": 0.09,
+      "learning_rate": 9.06e-06,
+      "loss": 4.3041,
       "step": 155
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 3.1000000000000004e-06,
-      "loss": 4.3336,
+      "epoch": 0.09,
+      "learning_rate": 9.12e-06,
+      "loss": 4.5384,
       "step": 156
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 3.12e-06,
-      "loss": 4.2307,
+      "epoch": 0.09,
+      "learning_rate": 9.18e-06,
+      "loss": 4.3262,
       "step": 157
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 3.1400000000000004e-06,
-      "loss": 4.3043,
+      "epoch": 0.09,
+      "learning_rate": 9.24e-06,
+      "loss": 4.4315,
       "step": 158
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 3.1600000000000002e-06,
-      "loss": 4.28,
+      "epoch": 0.09,
+      "learning_rate": 9.3e-06,
+      "loss": 4.305,
       "step": 159
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 3.1800000000000005e-06,
-      "loss": 4.2736,
+      "epoch": 0.09,
+      "learning_rate": 9.36e-06,
+      "loss": 4.257,
       "step": 160
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.2756,
+      "epoch": 0.09,
+      "learning_rate": 9.42e-06,
+      "loss": 4.3088,
       "step": 161
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 3.2200000000000005e-06,
-      "loss": 4.2625,
+      "epoch": 0.09,
+      "learning_rate": 9.48e-06,
+      "loss": 4.4273,
       "step": 162
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 3.2400000000000003e-06,
-      "loss": 4.3129,
+      "epoch": 0.09,
+      "learning_rate": 9.54e-06,
+      "loss": 4.2524,
       "step": 163
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 3.2600000000000006e-06,
-      "loss": 4.2453,
+      "epoch": 0.09,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.4965,
       "step": 164
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 3.2800000000000004e-06,
-      "loss": 4.3079,
+      "epoch": 0.09,
+      "learning_rate": 9.66e-06,
+      "loss": 4.2816,
       "step": 165
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 3.3000000000000006e-06,
-      "loss": 4.2754,
+      "epoch": 0.09,
+      "learning_rate": 9.72e-06,
+      "loss": 4.4945,
       "step": 166
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 3.3200000000000004e-06,
-      "loss": 4.3026,
+      "epoch": 0.09,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.3226,
       "step": 167
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 3.3400000000000006e-06,
-      "loss": 4.2798,
+      "epoch": 0.09,
+      "learning_rate": 9.84e-06,
+      "loss": 4.0308,
       "step": 168
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 3.3600000000000004e-06,
-      "loss": 4.2466,
+      "epoch": 0.09,
+      "learning_rate": 9.9e-06,
+      "loss": 4.138,
       "step": 169
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 3.3800000000000007e-06,
-      "loss": 4.1995,
+      "epoch": 0.1,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.271,
       "step": 170
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.1824,
+      "epoch": 0.1,
+      "learning_rate": 1.002e-05,
+      "loss": 4.3458,
       "step": 171
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 3.4200000000000007e-06,
-      "loss": 4.2936,
+      "epoch": 0.1,
+      "learning_rate": 1.008e-05,
+      "loss": 4.2662,
       "step": 172
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 3.44e-06,
-      "loss": 4.1523,
+      "epoch": 0.1,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 4.2941,
       "step": 173
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 3.46e-06,
-      "loss": 4.2063,
+      "epoch": 0.1,
+      "learning_rate": 1.02e-05,
+      "loss": 4.1816,
       "step": 174
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 3.48e-06,
-      "loss": 4.3818,
+      "epoch": 0.1,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 4.421,
       "step": 175
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 3.5e-06,
-      "loss": 4.2121,
+      "epoch": 0.1,
+      "learning_rate": 1.032e-05,
+      "loss": 4.1504,
       "step": 176
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 3.52e-06,
-      "loss": 4.35,
+      "epoch": 0.1,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 4.2012,
       "step": 177
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 3.54e-06,
-      "loss": 4.2109,
+      "epoch": 0.1,
+      "learning_rate": 1.044e-05,
+      "loss": 4.1953,
       "step": 178
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 3.5600000000000002e-06,
-      "loss": 4.2574,
+      "epoch": 0.1,
+      "learning_rate": 1.05e-05,
+      "loss": 3.972,
       "step": 179
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.58e-06,
-      "loss": 4.3989,
+      "epoch": 0.1,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 4.4177,
       "step": 180
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.6000000000000003e-06,
-      "loss": 4.2638,
+      "epoch": 0.1,
+      "learning_rate": 1.062e-05,
+      "loss": 4.2797,
       "step": 181
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.62e-06,
-      "loss": 4.2118,
+      "epoch": 0.1,
+      "learning_rate": 1.068e-05,
+      "loss": 4.2346,
       "step": 182
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.6400000000000003e-06,
-      "loss": 4.1813,
+      "epoch": 0.1,
+      "learning_rate": 1.074e-05,
+      "loss": 4.3591,
       "step": 183
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.66e-06,
-      "loss": 4.2778,
+      "epoch": 0.1,
+      "learning_rate": 1.08e-05,
+      "loss": 4.3874,
       "step": 184
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.6800000000000003e-06,
-      "loss": 4.3297,
+      "epoch": 0.1,
+      "learning_rate": 1.086e-05,
+      "loss": 4.3402,
       "step": 185
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.7e-06,
-      "loss": 4.272,
+      "epoch": 0.1,
+      "learning_rate": 1.092e-05,
+      "loss": 4.1991,
       "step": 186
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.7200000000000004e-06,
-      "loss": 4.3077,
+      "epoch": 0.1,
+      "learning_rate": 1.098e-05,
+      "loss": 4.2469,
       "step": 187
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.74e-06,
-      "loss": 4.2812,
+      "epoch": 0.11,
+      "learning_rate": 1.104e-05,
+      "loss": 4.4661,
       "step": 188
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.7600000000000004e-06,
-      "loss": 4.2457,
+      "epoch": 0.11,
+      "learning_rate": 1.11e-05,
+      "loss": 4.1801,
       "step": 189
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.3328,
+      "epoch": 0.11,
+      "learning_rate": 1.116e-05,
+      "loss": 4.1973,
       "step": 190
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.8000000000000005e-06,
-      "loss": 4.3663,
+      "epoch": 0.11,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 4.4443,
       "step": 191
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.820000000000001e-06,
-      "loss": 4.4388,
+      "epoch": 0.11,
+      "learning_rate": 1.128e-05,
+      "loss": 4.1546,
       "step": 192
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.5001,
+      "epoch": 0.11,
+      "learning_rate": 1.134e-05,
+      "loss": 4.1076,
       "step": 193
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.86e-06,
-      "loss": 4.4342,
+      "epoch": 0.11,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.2657,
       "step": 194
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.88e-06,
-      "loss": 4.4113,
+      "epoch": 0.11,
+      "learning_rate": 1.146e-05,
+      "loss": 4.5543,
       "step": 195
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.900000000000001e-06,
-      "loss": 4.3612,
+      "epoch": 0.11,
+      "learning_rate": 1.152e-05,
+      "loss": 4.388,
       "step": 196
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.920000000000001e-06,
-      "loss": 4.2716,
+      "epoch": 0.11,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 4.3508,
       "step": 197
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.94e-06,
-      "loss": 4.7501,
+      "epoch": 0.11,
+      "learning_rate": 1.164e-05,
+      "loss": 4.3336,
       "step": 198
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.96e-06,
-      "loss": 4.5016,
+      "epoch": 0.11,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 4.4416,
       "step": 199
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.980000000000001e-06,
-      "loss": 4.7116,
+      "epoch": 0.11,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 4.2942,
       "step": 200
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.3106,
+      "epoch": 0.11,
+      "learning_rate": 1.182e-05,
+      "loss": 4.2803,
       "step": 201
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.2625,
+      "epoch": 0.11,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 4.2008,
       "step": 202
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.04e-06,
-      "loss": 4.2393,
+      "epoch": 0.11,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 4.3831,
       "step": 203
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.060000000000001e-06,
-      "loss": 4.1657,
+      "epoch": 0.11,
+      "learning_rate": 1.2e-05,
+      "loss": 4.0557,
       "step": 204
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.08e-06,
-      "loss": 4.2338,
+      "epoch": 0.11,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 4.2071,
       "step": 205
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.1e-06,
-      "loss": 4.229,
+      "epoch": 0.12,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 4.306,
       "step": 206
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.12e-06,
-      "loss": 4.2768,
+      "epoch": 0.12,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 4.3491,
       "step": 207
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.14e-06,
-      "loss": 4.245,
+      "epoch": 0.12,
+      "learning_rate": 1.224e-05,
+      "loss": 4.0568,
       "step": 208
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.16e-06,
-      "loss": 4.2213,
+      "epoch": 0.12,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 4.0992,
       "step": 209
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.18e-06,
-      "loss": 4.3171,
+      "epoch": 0.12,
+      "learning_rate": 1.236e-05,
+      "loss": 4.2725,
       "step": 210
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.1127,
+      "epoch": 0.12,
+      "learning_rate": 1.242e-05,
+      "loss": 4.2728,
       "step": 211
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.22e-06,
-      "loss": 4.2402,
+      "epoch": 0.12,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 4.2163,
       "step": 212
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.24e-06,
-      "loss": 4.265,
+      "epoch": 0.12,
+      "learning_rate": 1.254e-05,
+      "loss": 4.212,
       "step": 213
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.26e-06,
-      "loss": 4.2684,
+      "epoch": 0.12,
+      "learning_rate": 1.26e-05,
+      "loss": 4.0068,
       "step": 214
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.2800000000000005e-06,
-      "loss": 4.2735,
+      "epoch": 0.12,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 4.4041,
       "step": 215
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.3e-06,
-      "loss": 4.2555,
+      "epoch": 0.12,
+      "learning_rate": 1.272e-05,
+      "loss": 3.88,
       "step": 216
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.32e-06,
-      "loss": 4.2211,
+      "epoch": 0.12,
+      "learning_rate": 1.278e-05,
+      "loss": 4.1302,
       "step": 217
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.34e-06,
-      "loss": 4.1925,
+      "epoch": 0.12,
+      "learning_rate": 1.284e-05,
+      "loss": 4.2383,
       "step": 218
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.360000000000001e-06,
-      "loss": 4.1934,
+      "epoch": 0.12,
+      "learning_rate": 1.29e-05,
+      "loss": 4.1737,
       "step": 219
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.38e-06,
-      "loss": 4.2549,
+      "epoch": 0.12,
+      "learning_rate": 1.296e-05,
+      "loss": 4.4725,
       "step": 220
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.4e-06,
-      "loss": 4.1733,
+      "epoch": 0.12,
+      "learning_rate": 1.302e-05,
+      "loss": 4.3053,
       "step": 221
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.42e-06,
-      "loss": 4.1752,
+      "epoch": 0.12,
+      "learning_rate": 1.308e-05,
+      "loss": 4.3469,
       "step": 222
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.440000000000001e-06,
-      "loss": 4.1934,
+      "epoch": 0.12,
+      "learning_rate": 1.314e-05,
+      "loss": 4.136,
       "step": 223
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.4600000000000005e-06,
-      "loss": 4.2419,
+      "epoch": 0.13,
+      "learning_rate": 1.32e-05,
+      "loss": 4.2982,
       "step": 224
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 4.48e-06,
-      "loss": 4.2159,
+      "epoch": 0.13,
+      "learning_rate": 1.326e-05,
+      "loss": 4.3243,
       "step": 225
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 4.5e-06,
-      "loss": 4.226,
+      "epoch": 0.13,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 4.4277,
       "step": 226
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 4.520000000000001e-06,
-      "loss": 4.2576,
+      "epoch": 0.13,
+      "learning_rate": 1.338e-05,
+      "loss": 4.3491,
       "step": 227
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 4.540000000000001e-06,
-      "loss": 4.1938,
+      "epoch": 0.13,
+      "learning_rate": 1.344e-05,
+      "loss": 4.0476,
       "step": 228
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 4.56e-06,
-      "loss": 4.2063,
+      "epoch": 0.13,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 4.2359,
       "step": 229
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 4.58e-06,
-      "loss": 4.2194,
+      "epoch": 0.13,
+      "learning_rate": 1.356e-05,
+      "loss": 4.2449,
       "step": 230
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.600000000000001e-06,
-      "loss": 4.3171,
+      "epoch": 0.13,
+      "learning_rate": 1.362e-05,
+      "loss": 4.2231,
       "step": 231
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.620000000000001e-06,
-      "loss": 4.1888,
+      "epoch": 0.13,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 4.2565,
       "step": 232
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.6400000000000005e-06,
-      "loss": 4.2994,
+      "epoch": 0.13,
+      "learning_rate": 1.374e-05,
+      "loss": 4.401,
       "step": 233
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.66e-06,
-      "loss": 4.2743,
+      "epoch": 0.13,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.2996,
       "step": 234
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.680000000000001e-06,
-      "loss": 4.3265,
+      "epoch": 0.13,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 4.1248,
       "step": 235
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.7e-06,
-      "loss": 4.146,
+      "epoch": 0.13,
+      "learning_rate": 1.392e-05,
+      "loss": 4.6215,
       "step": 236
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 4.7200000000000005e-06,
-      "loss": 4.2531,
+      "epoch": 0.13,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 4.3122,
       "step": 237
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 4.74e-06,
-      "loss": 4.3796,
+      "epoch": 0.13,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 4.2869,
       "step": 238
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 4.76e-06,
-      "loss": 4.2271,
+      "epoch": 0.13,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 4.2742,
       "step": 239
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.78e-06,
-      "loss": 4.2738,
+      "epoch": 0.13,
+      "learning_rate": 1.416e-05,
+      "loss": 4.2167,
       "step": 240
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.3837,
+      "epoch": 0.14,
+      "learning_rate": 1.422e-05,
+      "loss": 4.4837,
       "step": 241
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.8200000000000004e-06,
-      "loss": 4.3178,
+      "epoch": 0.14,
+      "learning_rate": 1.428e-05,
+      "loss": 4.2326,
       "step": 242
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 4.84e-06,
-      "loss": 4.2609,
+      "epoch": 0.14,
+      "learning_rate": 1.434e-05,
+      "loss": 4.3337,
       "step": 243
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 4.86e-06,
-      "loss": 4.2858,
+      "epoch": 0.14,
+      "learning_rate": 1.44e-05,
+      "loss": 4.469,
       "step": 244
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 4.2618,
+      "epoch": 0.14,
+      "learning_rate": 1.446e-05,
+      "loss": 4.6109,
       "step": 245
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 4.4445,
+      "epoch": 0.14,
+      "learning_rate": 1.452e-05,
+      "loss": 4.5276,
       "step": 246
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.92e-06,
-      "loss": 4.3901,
+      "epoch": 0.14,
+      "learning_rate": 1.458e-05,
+      "loss": 4.3512,
       "step": 247
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.94e-06,
-      "loss": 4.3951,
+      "epoch": 0.14,
+      "learning_rate": 1.464e-05,
+      "loss": 4.42,
       "step": 248
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 4.960000000000001e-06,
-      "loss": 4.5076,
+      "epoch": 0.14,
+      "learning_rate": 1.47e-05,
+      "loss": 4.2021,
       "step": 249
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.5527,
+      "epoch": 0.14,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 4.2523,
       "step": 250
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 5e-06,
-      "loss": 4.2833,
+      "epoch": 0.14,
+      "learning_rate": 1.482e-05,
+      "loss": 4.1961,
       "step": 251
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 5.02e-06,
-      "loss": 4.206,
+      "epoch": 0.14,
+      "learning_rate": 1.488e-05,
+      "loss": 4.144,
       "step": 252
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 5.04e-06,
-      "loss": 4.1542,
+      "epoch": 0.14,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 4.1655,
       "step": 253
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 5.060000000000001e-06,
-      "loss": 4.1541,
+      "epoch": 0.14,
+      "learning_rate": 1.5e-05,
+      "loss": 4.3351,
       "step": 254
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 5.0800000000000005e-06,
-      "loss": 4.1914,
+      "epoch": 0.14,
+      "learning_rate": 1.506e-05,
+      "loss": 4.235,
       "step": 255
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 5.1e-06,
-      "loss": 4.1824,
+      "epoch": 0.14,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 4.1797,
       "step": 256
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 5.12e-06,
-      "loss": 4.3117,
+      "epoch": 0.14,
+      "learning_rate": 1.518e-05,
+      "loss": 4.2129,
       "step": 257
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 5.140000000000001e-06,
-      "loss": 4.192,
+      "epoch": 0.14,
+      "learning_rate": 1.524e-05,
+      "loss": 4.3488,
       "step": 258
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 5.1600000000000006e-06,
-      "loss": 4.2298,
+      "epoch": 0.15,
+      "learning_rate": 1.53e-05,
+      "loss": 4.4473,
       "step": 259
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 5.18e-06,
-      "loss": 4.1561,
+      "epoch": 0.15,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 3.9477,
       "step": 260
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 5.2e-06,
-      "loss": 4.2479,
+      "epoch": 0.15,
+      "learning_rate": 1.542e-05,
+      "loss": 4.0845,
       "step": 261
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 5.220000000000001e-06,
-      "loss": 4.2713,
+      "epoch": 0.15,
+      "learning_rate": 1.548e-05,
+      "loss": 4.3156,
       "step": 262
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 5.240000000000001e-06,
-      "loss": 4.1502,
+      "epoch": 0.15,
+      "learning_rate": 1.554e-05,
+      "loss": 4.2628,
       "step": 263
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.2600000000000005e-06,
-      "loss": 4.1789,
+      "epoch": 0.15,
+      "learning_rate": 1.56e-05,
+      "loss": 4.0813,
       "step": 264
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.28e-06,
-      "loss": 4.2032,
+      "epoch": 0.15,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 4.1697,
       "step": 265
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.300000000000001e-06,
-      "loss": 4.2824,
+      "epoch": 0.15,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 4.0005,
       "step": 266
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.320000000000001e-06,
-      "loss": 4.2033,
+      "epoch": 0.15,
+      "learning_rate": 1.578e-05,
+      "loss": 4.3022,
       "step": 267
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.3400000000000005e-06,
-      "loss": 4.1714,
+      "epoch": 0.15,
+      "learning_rate": 1.584e-05,
+      "loss": 4.2533,
       "step": 268
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.36e-06,
-      "loss": 4.1364,
+      "epoch": 0.15,
+      "learning_rate": 1.59e-05,
+      "loss": 4.4185,
       "step": 269
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 5.380000000000001e-06,
-      "loss": 4.261,
+      "epoch": 0.15,
+      "learning_rate": 1.596e-05,
+      "loss": 4.0305,
       "step": 270
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 5.400000000000001e-06,
-      "loss": 4.1305,
+      "epoch": 0.15,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 4.323,
       "step": 271
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 5.420000000000001e-06,
-      "loss": 4.1106,
+      "epoch": 0.15,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 4.1396,
       "step": 272
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 5.4400000000000004e-06,
-      "loss": 4.1723,
+      "epoch": 0.15,
+      "learning_rate": 1.614e-05,
+      "loss": 4.0473,
       "step": 273
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 5.460000000000001e-06,
-      "loss": 4.1495,
+      "epoch": 0.15,
+      "learning_rate": 1.62e-05,
+      "loss": 4.31,
       "step": 274
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 5.480000000000001e-06,
-      "loss": 4.2855,
+      "epoch": 0.15,
+      "learning_rate": 1.626e-05,
+      "loss": 4.2983,
       "step": 275
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 4.199,
+      "epoch": 0.15,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 4.2654,
       "step": 276
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 5.5200000000000005e-06,
-      "loss": 4.2175,
+      "epoch": 0.16,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 4.2673,
       "step": 277
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 5.540000000000001e-06,
-      "loss": 4.152,
+      "epoch": 0.16,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 4.3477,
       "step": 278
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 5.560000000000001e-06,
-      "loss": 4.1857,
+      "epoch": 0.16,
+      "learning_rate": 1.65e-05,
+      "loss": 4.1842,
       "step": 279
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 5.580000000000001e-06,
-      "loss": 4.1679,
+      "epoch": 0.16,
+      "learning_rate": 1.656e-05,
+      "loss": 4.1999,
       "step": 280
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.2036,
+      "epoch": 0.16,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 4.3175,
       "step": 281
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 5.620000000000001e-06,
-      "loss": 4.2049,
+      "epoch": 0.16,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 4.3657,
       "step": 282
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 5.64e-06,
-      "loss": 4.3554,
+      "epoch": 0.16,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 4.3419,
       "step": 283
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 5.66e-06,
-      "loss": 4.1291,
+      "epoch": 0.16,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.1069,
       "step": 284
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 5.68e-06,
-      "loss": 4.2472,
+      "epoch": 0.16,
+      "learning_rate": 1.686e-05,
+      "loss": 4.2524,
       "step": 285
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 5.7e-06,
-      "loss": 4.2736,
+      "epoch": 0.16,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 4.0175,
       "step": 286
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.72e-06,
-      "loss": 4.1614,
+      "epoch": 0.16,
+      "learning_rate": 1.698e-05,
+      "loss": 4.2729,
       "step": 287
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.74e-06,
-      "loss": 4.2049,
+      "epoch": 0.16,
+      "learning_rate": 1.704e-05,
+      "loss": 4.467,
       "step": 288
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.76e-06,
-      "loss": 4.206,
+      "epoch": 0.16,
+      "learning_rate": 1.71e-05,
+      "loss": 4.1999,
       "step": 289
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 5.78e-06,
-      "loss": 4.2975,
+      "epoch": 0.16,
+      "learning_rate": 1.716e-05,
+      "loss": 4.2919,
       "step": 290
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 5.8e-06,
-      "loss": 4.2788,
+      "epoch": 0.16,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 4.2256,
       "step": 291
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 5.82e-06,
-      "loss": 4.3599,
+      "epoch": 0.16,
+      "learning_rate": 1.728e-05,
+      "loss": 4.611,
       "step": 292
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 5.84e-06,
-      "loss": 4.3441,
+      "epoch": 0.16,
+      "learning_rate": 1.734e-05,
+      "loss": 4.3976,
       "step": 293
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 5.86e-06,
-      "loss": 4.5265,
+      "epoch": 0.16,
+      "learning_rate": 1.74e-05,
+      "loss": 4.4916,
       "step": 294
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.3277,
+      "epoch": 0.17,
+      "learning_rate": 1.746e-05,
+      "loss": 4.3946,
       "step": 295
     },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 4.2977,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.758e-05,
+      "loss": 4.2863,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.764e-05,
+      "loss": 4.4486,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.77e-05,
+      "loss": 4.2922,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.776e-05,
+      "loss": 4.5119,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.782e-05,
+      "loss": 4.2882,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 4.1278,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.794e-05,
+      "loss": 4.3202,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8e-05,
+      "loss": 4.1762,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.806e-05,
+      "loss": 4.3179,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.812e-05,
+      "loss": 4.1969,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.818e-05,
+      "loss": 4.3899,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.824e-05,
+      "loss": 4.4459,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.83e-05,
+      "loss": 4.4296,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.836e-05,
+      "loss": 4.2879,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.842e-05,
+      "loss": 4.2889,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.848e-05,
+      "loss": 4.1826,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854e-05,
+      "loss": 4.2387,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.86e-05,
+      "loss": 4.3569,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.866e-05,
+      "loss": 4.3756,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.872e-05,
+      "loss": 4.1568,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878e-05,
+      "loss": 4.1962,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884e-05,
+      "loss": 3.9272,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 4.1472,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.896e-05,
+      "loss": 4.2929,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.902e-05,
+      "loss": 4.4923,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.908e-05,
+      "loss": 4.3105,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.914e-05,
+      "loss": 4.3273,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.2032,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 4.1822,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.932e-05,
+      "loss": 4.1634,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.938e-05,
+      "loss": 4.0854,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.944e-05,
+      "loss": 4.2017,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.95e-05,
+      "loss": 4.1804,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 4.4926,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 4.2843,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.968e-05,
+      "loss": 4.3779,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.974e-05,
+      "loss": 4.3823,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.98e-05,
+      "loss": 4.348,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 4.3798,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 4.1543,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 4.2705,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.004e-05,
+      "loss": 4.1086,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.01e-05,
+      "loss": 4.2179,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.016e-05,
+      "loss": 4.2029,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 4.1593,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 4.3541,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 4.5021,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.04e-05,
+      "loss": 4.0483,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.046e-05,
+      "loss": 4.3249,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 4.1787,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 4.6715,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.064e-05,
+      "loss": 4.57,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.07e-05,
+      "loss": 4.2913,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 4.2299,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.082e-05,
+      "loss": 4.3529,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.088e-05,
+      "loss": 4.4146,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.094e-05,
+      "loss": 4.0571,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1e-05,
+      "loss": 4.1935,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 3.8818,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 4.1422,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.118e-05,
+      "loss": 4.3262,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.124e-05,
+      "loss": 4.3255,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.13e-05,
+      "loss": 4.4807,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.136e-05,
+      "loss": 4.1969,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 4.227,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.148e-05,
+      "loss": 4.0115,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.154e-05,
+      "loss": 4.2946,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.16e-05,
+      "loss": 4.2937,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.166e-05,
+      "loss": 4.2544,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.172e-05,
+      "loss": 4.3866,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.178e-05,
+      "loss": 4.2233,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.184e-05,
+      "loss": 4.1498,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.19e-05,
+      "loss": 4.0466,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.196e-05,
+      "loss": 4.1159,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.202e-05,
+      "loss": 4.199,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.208e-05,
+      "loss": 4.6349,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.214e-05,
+      "loss": 4.2175,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.22e-05,
+      "loss": 4.0352,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.226e-05,
+      "loss": 4.0145,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.232e-05,
+      "loss": 4.0913,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.238e-05,
+      "loss": 4.1255,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2440000000000002e-05,
+      "loss": 4.0008,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.25e-05,
+      "loss": 4.1971,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.256e-05,
+      "loss": 4.3701,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.262e-05,
+      "loss": 3.9528,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.268e-05,
+      "loss": 4.1182,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.274e-05,
+      "loss": 4.2439,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.1558,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.286e-05,
+      "loss": 4.3503,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.292e-05,
+      "loss": 4.3124,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.298e-05,
+      "loss": 4.3033,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.304e-05,
+      "loss": 4.0629,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 4.3325,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 4.5781,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.322e-05,
+      "loss": 4.1335,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.328e-05,
+      "loss": 4.5669,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.334e-05,
+      "loss": 4.5815,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.3085,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 4.4148,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 4.3114,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.358e-05,
+      "loss": 4.3277,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.364e-05,
+      "loss": 4.7393,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.37e-05,
+      "loss": 5.237,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 4.8684,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 4.3808,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 4.1168,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.394e-05,
+      "loss": 4.1313,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4e-05,
+      "loss": 4.325,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 4.0398,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4120000000000003e-05,
+      "loss": 4.2195,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 4.1106,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 4.2559,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.43e-05,
+      "loss": 4.3395,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 4.2696,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.442e-05,
+      "loss": 4.2173,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.448e-05,
+      "loss": 4.2487,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.454e-05,
+      "loss": 4.263,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 4.2865,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.4659999999999998e-05,
+      "loss": 4.2806,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.472e-05,
+      "loss": 4.1959,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.478e-05,
+      "loss": 4.2755,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.484e-05,
+      "loss": 4.2586,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.49e-05,
+      "loss": 4.2952,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4959999999999998e-05,
+      "loss": 4.2922,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.502e-05,
+      "loss": 4.1313,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.508e-05,
+      "loss": 4.4236,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.514e-05,
+      "loss": 4.057,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.52e-05,
+      "loss": 4.5362,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.526e-05,
+      "loss": 4.4132,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 4.2817,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.538e-05,
+      "loss": 4.153,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.544e-05,
+      "loss": 4.1823,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.55e-05,
+      "loss": 4.1813,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.556e-05,
+      "loss": 4.2932,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.562e-05,
+      "loss": 4.1991,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.568e-05,
+      "loss": 4.2998,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.574e-05,
+      "loss": 4.2479,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.58e-05,
+      "loss": 4.3735,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.586e-05,
+      "loss": 4.3559,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.592e-05,
+      "loss": 4.004,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 4.1885,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.604e-05,
+      "loss": 4.4007,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.61e-05,
+      "loss": 4.1327,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.616e-05,
+      "loss": 4.3869,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.622e-05,
+      "loss": 4.3519,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.628e-05,
+      "loss": 4.3361,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 4.2986,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.64e-05,
+      "loss": 4.511,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.646e-05,
+      "loss": 4.3132,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.652e-05,
+      "loss": 5.0121,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.658e-05,
+      "loss": 4.7076,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 5.1611,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 4.6371,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.676e-05,
+      "loss": 4.4926,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.682e-05,
+      "loss": 3.9792,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.688e-05,
+      "loss": 4.1466,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6940000000000003e-05,
+      "loss": 4.2595,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 3.977,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 4.0583,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.712e-05,
+      "loss": 4.1932,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.718e-05,
+      "loss": 3.9978,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.724e-05,
+      "loss": 4.016,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 3.95,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 4.1647,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 4.1004,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.748e-05,
+      "loss": 4.0818,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.754e-05,
+      "loss": 4.12,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.0955,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 4.0958,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 4.2873,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.778e-05,
+      "loss": 4.1271,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.784e-05,
+      "loss": 4.2386,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.79e-05,
+      "loss": 4.0488,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 4.1552,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 4.3574,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 4.1464,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 4.2077,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.2895,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.826e-05,
+      "loss": 4.2703,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.832e-05,
+      "loss": 4.369,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.838e-05,
+      "loss": 4.3544,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.844e-05,
+      "loss": 4.2284,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 4.3457,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.856e-05,
+      "loss": 4.331,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.862e-05,
+      "loss": 4.0001,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.868e-05,
+      "loss": 4.2137,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.874e-05,
+      "loss": 4.4671,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.88e-05,
+      "loss": 4.1296,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8859999999999998e-05,
+      "loss": 4.4587,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.892e-05,
+      "loss": 4.4997,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.898e-05,
+      "loss": 4.3843,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.904e-05,
+      "loss": 4.5038,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.91e-05,
+      "loss": 4.4373,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916e-05,
+      "loss": 4.3053,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.922e-05,
+      "loss": 4.1663,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.928e-05,
+      "loss": 4.5699,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.934e-05,
+      "loss": 4.3797,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.94e-05,
+      "loss": 4.419,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.946e-05,
+      "loss": 4.3915,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 4.3229,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.958e-05,
+      "loss": 4.1784,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.964e-05,
+      "loss": 4.6169,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.97e-05,
+      "loss": 4.7747,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.976e-05,
+      "loss": 4.1429,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 4.325467586517334,
+      "eval_runtime": 911.2036,
+      "eval_samples_per_second": 2.899,
+      "eval_steps_per_second": 0.363,
+      "eval_wer": 1.959579531931773,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.982e-05,
+      "loss": 4.402,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 4.098,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.994e-05,
+      "loss": 4.1644,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3e-05,
+      "loss": 4.2779,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9976635514018694e-05,
+      "loss": 4.0974,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9953271028037384e-05,
+      "loss": 4.3191,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9929906542056074e-05,
+      "loss": 4.1537,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9906542056074768e-05,
+      "loss": 4.1245,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.988317757009346e-05,
+      "loss": 4.3043,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.985981308411215e-05,
+      "loss": 4.4321,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983644859813084e-05,
+      "loss": 4.2622,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9813084112149534e-05,
+      "loss": 4.1302,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9789719626168224e-05,
+      "loss": 4.2863,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9766355140186918e-05,
+      "loss": 4.0416,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9742990654205608e-05,
+      "loss": 4.1466,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.97196261682243e-05,
+      "loss": 4.5779,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.969626168224299e-05,
+      "loss": 4.2311,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9672897196261685e-05,
+      "loss": 4.5464,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9649532710280375e-05,
+      "loss": 4.3729,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9626168224299065e-05,
+      "loss": 4.0685,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9602803738317758e-05,
+      "loss": 3.9723,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.957943925233645e-05,
+      "loss": 4.1532,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.955607476635514e-05,
+      "loss": 4.062,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.953271028037383e-05,
+      "loss": 4.3025,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9509345794392525e-05,
+      "loss": 4.4767,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9485981308411218e-05,
+      "loss": 4.325,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9462616822429905e-05,
+      "loss": 4.4228,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9439252336448598e-05,
+      "loss": 3.9998,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.941588785046729e-05,
+      "loss": 4.3276,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9392523364485985e-05,
+      "loss": 4.4454,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.936915887850467e-05,
+      "loss": 4.1381,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9345794392523365e-05,
+      "loss": 3.9719,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.932242990654206e-05,
+      "loss": 4.0366,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.929906542056075e-05,
+      "loss": 3.9813,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.927570093457944e-05,
+      "loss": 4.3925,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9252336448598132e-05,
+      "loss": 4.251,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9228971962616825e-05,
+      "loss": 4.1116,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9205607476635515e-05,
+      "loss": 4.3855,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9182242990654205e-05,
+      "loss": 4.3704,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.91588785046729e-05,
+      "loss": 4.3425,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.913551401869159e-05,
+      "loss": 4.5238,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9112149532710282e-05,
+      "loss": 4.4286,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9088785046728972e-05,
+      "loss": 4.5212,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9065420560747665e-05,
+      "loss": 4.3171,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9042056074766355e-05,
+      "loss": 4.1578,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.901869158878505e-05,
+      "loss": 4.4432,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.899532710280374e-05,
+      "loss": 4.6005,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.897196261682243e-05,
+      "loss": 4.5918,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8948598130841122e-05,
+      "loss": 4.7832,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8925233644859816e-05,
+      "loss": 4.6226,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8901869158878506e-05,
+      "loss": 4.6,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8878504672897196e-05,
+      "loss": 4.3506,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.885514018691589e-05,
+      "loss": 4.4258,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8831775700934582e-05,
+      "loss": 4.1916,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8808411214953272e-05,
+      "loss": 4.1482,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8785046728971962e-05,
+      "loss": 4.2643,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8761682242990656e-05,
+      "loss": 4.3101,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.873831775700935e-05,
+      "loss": 4.1195,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8714953271028036e-05,
+      "loss": 4.056,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.869158878504673e-05,
+      "loss": 4.0648,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8668224299065423e-05,
+      "loss": 4.3317,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8644859813084116e-05,
+      "loss": 4.2129,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8621495327102803e-05,
+      "loss": 3.9612,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8598130841121496e-05,
+      "loss": 4.4715,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.857476635514019e-05,
+      "loss": 4.4355,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.855140186915888e-05,
+      "loss": 4.1844,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.852803738317757e-05,
+      "loss": 4.044,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8504672897196263e-05,
+      "loss": 4.3697,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8481308411214956e-05,
+      "loss": 4.2911,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8457943925233646e-05,
+      "loss": 4.4537,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8434579439252336e-05,
+      "loss": 4.1536,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.841121495327103e-05,
+      "loss": 4.4095,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.838785046728972e-05,
+      "loss": 4.1888,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8364485981308413e-05,
+      "loss": 4.0191,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8341121495327103e-05,
+      "loss": 4.0913,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8317757009345797e-05,
+      "loss": 4.1657,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8294392523364487e-05,
+      "loss": 4.2038,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.827102803738318e-05,
+      "loss": 4.0571,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.824766355140187e-05,
+      "loss": 4.1484,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.822429906542056e-05,
+      "loss": 4.4237,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8200934579439253e-05,
+      "loss": 4.2888,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8177570093457947e-05,
+      "loss": 4.2798,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8154205607476637e-05,
+      "loss": 4.0213,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8130841121495327e-05,
+      "loss": 4.4724,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.810747663551402e-05,
+      "loss": 4.4205,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8084112149532714e-05,
+      "loss": 4.1293,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.80607476635514e-05,
+      "loss": 4.3651,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8037383177570094e-05,
+      "loss": 4.3099,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8014018691588787e-05,
+      "loss": 4.4224,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.799065420560748e-05,
+      "loss": 4.4837,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7967289719626167e-05,
+      "loss": 4.2061,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.794392523364486e-05,
+      "loss": 4.456,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7920560747663554e-05,
+      "loss": 4.3043,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7897196261682244e-05,
+      "loss": 4.4633,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7873831775700934e-05,
+      "loss": 4.4562,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7850467289719627e-05,
+      "loss": 4.4316,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.782710280373832e-05,
+      "loss": 4.6998,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.780373831775701e-05,
+      "loss": 4.5698,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.77803738317757e-05,
+      "loss": 4.7437,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7757009345794394e-05,
+      "loss": 4.8452,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7733644859813084e-05,
+      "loss": 3.9349,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7710280373831777e-05,
+      "loss": 4.3194,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7686915887850467e-05,
+      "loss": 4.1411,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.766355140186916e-05,
+      "loss": 4.2453,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.764018691588785e-05,
+      "loss": 4.071,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7616822429906544e-05,
+      "loss": 4.0501,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7593457943925234e-05,
+      "loss": 4.16,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7570093457943924e-05,
+      "loss": 4.1742,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7546728971962618e-05,
+      "loss": 4.6202,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.752336448598131e-05,
+      "loss": 4.1865,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.75e-05,
+      "loss": 4.4477,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.747663551401869e-05,
+      "loss": 4.1238,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7453271028037384e-05,
+      "loss": 4.2782,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7429906542056078e-05,
+      "loss": 4.171,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7406542056074764e-05,
+      "loss": 4.28,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7383177570093458e-05,
+      "loss": 4.0922,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.735981308411215e-05,
+      "loss": 4.1164,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7336448598130845e-05,
+      "loss": 4.1105,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.731308411214953e-05,
+      "loss": 4.0947,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7289719626168225e-05,
+      "loss": 4.231,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7266355140186918e-05,
+      "loss": 4.3028,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7242990654205608e-05,
+      "loss": 4.3094,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7219626168224298e-05,
+      "loss": 4.2783,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.719626168224299e-05,
+      "loss": 4.4667,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7172897196261685e-05,
+      "loss": 4.3927,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7149532710280375e-05,
+      "loss": 4.1948,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7126168224299065e-05,
+      "loss": 4.1488,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7102803738317758e-05,
+      "loss": 4.5734,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7079439252336448e-05,
+      "loss": 4.7672,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.705607476635514e-05,
+      "loss": 4.1367,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.703271028037383e-05,
+      "loss": 4.2877,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7009345794392525e-05,
+      "loss": 4.2702,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6985981308411215e-05,
+      "loss": 4.092,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.696261682242991e-05,
+      "loss": 4.2568,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.69392523364486e-05,
+      "loss": 4.2572,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.691588785046729e-05,
+      "loss": 4.095,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6892523364485982e-05,
+      "loss": 4.258,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6869158878504675e-05,
+      "loss": 3.9506,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6845794392523365e-05,
+      "loss": 4.3752,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6822429906542055e-05,
+      "loss": 4.6322,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.679906542056075e-05,
+      "loss": 4.2883,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6775700934579442e-05,
+      "loss": 4.4331,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.675233644859813e-05,
+      "loss": 4.1822,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6728971962616822e-05,
+      "loss": 4.3922,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6705607476635515e-05,
+      "loss": 4.3146,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.668224299065421e-05,
+      "loss": 4.4116,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6658878504672895e-05,
+      "loss": 4.958,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.663551401869159e-05,
+      "loss": 5.0488,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6612149532710282e-05,
+      "loss": 4.6143,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6588785046728972e-05,
+      "loss": 4.6788,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6565420560747662e-05,
+      "loss": 4.2092,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6542056074766356e-05,
+      "loss": 4.1492,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.651869158878505e-05,
+      "loss": 4.1696,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.649532710280374e-05,
+      "loss": 4.1479,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.647196261682243e-05,
+      "loss": 4.1533,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6448598130841122e-05,
+      "loss": 4.1071,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6425233644859816e-05,
+      "loss": 4.1628,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6401869158878506e-05,
+      "loss": 4.4812,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6378504672897196e-05,
+      "loss": 4.4136,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.635514018691589e-05,
+      "loss": 4.2936,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.633177570093458e-05,
+      "loss": 4.287,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6308411214953273e-05,
+      "loss": 4.2534,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6285046728971963e-05,
+      "loss": 4.2742,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6261682242990656e-05,
+      "loss": 4.3767,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6238317757009346e-05,
+      "loss": 4.0789,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.621495327102804e-05,
+      "loss": 4.3652,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.619158878504673e-05,
+      "loss": 4.1736,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.616822429906542e-05,
+      "loss": 4.0471,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6144859813084113e-05,
+      "loss": 4.0858,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6121495327102806e-05,
+      "loss": 4.3764,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6098130841121496e-05,
+      "loss": 4.3355,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6074766355140186e-05,
+      "loss": 4.5081,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.605140186915888e-05,
+      "loss": 4.1611,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6028037383177573e-05,
+      "loss": 4.1265,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.600467289719626e-05,
+      "loss": 4.027,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.5981308411214953e-05,
+      "loss": 4.2169,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.5957943925233647e-05,
+      "loss": 4.1648,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.593457943925234e-05,
+      "loss": 4.1783,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.5911214953271027e-05,
+      "loss": 4.2289,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.588785046728972e-05,
+      "loss": 4.2056,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.5864485981308413e-05,
+      "loss": 4.1371,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.5841121495327103e-05,
+      "loss": 4.2906,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.5817757009345793e-05,
+      "loss": 4.3842,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.5794392523364487e-05,
+      "loss": 4.3779,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.577102803738318e-05,
+      "loss": 4.4326,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.574766355140187e-05,
+      "loss": 4.3783,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.572429906542056e-05,
+      "loss": 3.9719,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5700934579439254e-05,
+      "loss": 4.0846,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5677570093457944e-05,
+      "loss": 4.2709,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5654205607476637e-05,
+      "loss": 4.3541,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5630841121495327e-05,
+      "loss": 4.0952,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.560747663551402e-05,
+      "loss": 4.3175,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.558411214953271e-05,
+      "loss": 4.2835,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5560747663551404e-05,
+      "loss": 4.2477,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5537383177570094e-05,
+      "loss": 4.423,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5514018691588784e-05,
+      "loss": 4.4991,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5490654205607477e-05,
+      "loss": 3.9253,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.546728971962617e-05,
+      "loss": 4.2595,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.544392523364486e-05,
+      "loss": 4.2097,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.542056074766355e-05,
+      "loss": 4.266,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5397196261682244e-05,
+      "loss": 4.0802,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5373831775700937e-05,
+      "loss": 4.1576,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5350467289719624e-05,
+      "loss": 4.1498,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.5327102803738317e-05,
+      "loss": 4.4774,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.530373831775701e-05,
+      "loss": 4.3782,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5280373831775704e-05,
+      "loss": 4.2295,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.525700934579439e-05,
+      "loss": 4.1143,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5233644859813084e-05,
+      "loss": 4.1706,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5210280373831778e-05,
+      "loss": 3.9805,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5186915887850468e-05,
+      "loss": 4.2145,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5163551401869158e-05,
+      "loss": 4.362,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.514018691588785e-05,
+      "loss": 4.1171,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5116822429906544e-05,
+      "loss": 4.093,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5093457943925234e-05,
+      "loss": 4.5891,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5070093457943924e-05,
+      "loss": 4.3616,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5046728971962618e-05,
+      "loss": 4.0134,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5023364485981308e-05,
+      "loss": 4.1927,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5e-05,
+      "loss": 4.3493,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.497663551401869e-05,
+      "loss": 4.0589,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4953271028037385e-05,
+      "loss": 4.1947,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4929906542056075e-05,
+      "loss": 4.4066,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4906542056074768e-05,
+      "loss": 4.3165,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4883177570093458e-05,
+      "loss": 4.1629,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4859813084112148e-05,
+      "loss": 4.4456,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.483644859813084e-05,
+      "loss": 4.2816,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4813084112149535e-05,
+      "loss": 4.3095,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4789719626168225e-05,
+      "loss": 3.8964,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4766355140186915e-05,
+      "loss": 4.2793,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4742990654205608e-05,
+      "loss": 4.2645,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.47196261682243e-05,
+      "loss": 4.1055,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4696261682242988e-05,
+      "loss": 4.1787,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.467289719626168e-05,
+      "loss": 4.2148,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4649532710280375e-05,
+      "loss": 4.6198,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.462616822429907e-05,
+      "loss": 4.2668,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4602803738317755e-05,
+      "loss": 4.268,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.457943925233645e-05,
+      "loss": 4.1952,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4556074766355142e-05,
+      "loss": 4.0979,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4532710280373832e-05,
+      "loss": 3.922,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4509345794392522e-05,
+      "loss": 4.2186,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.4485981308411215e-05,
+      "loss": 4.1441,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.446261682242991e-05,
+      "loss": 4.4435,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.44392523364486e-05,
+      "loss": 4.0728,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.441588785046729e-05,
+      "loss": 4.2247,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4392523364485982e-05,
+      "loss": 4.1626,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4369158878504672e-05,
+      "loss": 4.8281,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4345794392523365e-05,
+      "loss": 4.2355,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4322429906542055e-05,
+      "loss": 4.4198,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.429906542056075e-05,
+      "loss": 4.5735,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.427570093457944e-05,
+      "loss": 4.1968,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4252336448598132e-05,
+      "loss": 4.5801,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4228971962616822e-05,
+      "loss": 4.2539,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4205607476635516e-05,
+      "loss": 4.2025,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4182242990654206e-05,
+      "loss": 4.2608,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.41588785046729e-05,
+      "loss": 3.9838,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.413551401869159e-05,
+      "loss": 4.1223,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.411214953271028e-05,
+      "loss": 4.296,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4088785046728972e-05,
+      "loss": 4.1593,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.4065420560747666e-05,
+      "loss": 4.1912,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4042056074766356e-05,
+      "loss": 4.3467,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4018691588785046e-05,
+      "loss": 4.1586,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.399532710280374e-05,
+      "loss": 4.5449,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3971962616822433e-05,
+      "loss": 4.0916,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.394859813084112e-05,
+      "loss": 4.1968,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3925233644859813e-05,
+      "loss": 3.9868,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3901869158878506e-05,
+      "loss": 3.9261,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.38785046728972e-05,
+      "loss": 4.1364,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3855140186915886e-05,
+      "loss": 4.3283,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.383177570093458e-05,
+      "loss": 3.9927,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3808411214953273e-05,
+      "loss": 4.5144,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3785046728971963e-05,
+      "loss": 4.2735,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3761682242990653e-05,
+      "loss": 4.1041,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3738317757009346e-05,
+      "loss": 4.1381,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.371495327102804e-05,
+      "loss": 4.1218,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.369158878504673e-05,
+      "loss": 4.2006,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.366822429906542e-05,
+      "loss": 4.0368,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3644859813084113e-05,
+      "loss": 4.1325,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3621495327102803e-05,
+      "loss": 4.0072,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3598130841121497e-05,
+      "loss": 4.4205,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3574766355140187e-05,
+      "loss": 4.3263,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.355140186915888e-05,
+      "loss": 4.4133,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.352803738317757e-05,
+      "loss": 4.3504,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3504672897196263e-05,
+      "loss": 4.3287,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3481308411214953e-05,
+      "loss": 4.2639,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3457943925233643e-05,
+      "loss": 4.151,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3434579439252337e-05,
+      "loss": 4.0087,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.341121495327103e-05,
+      "loss": 4.4572,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.338785046728972e-05,
+      "loss": 3.8878,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.336448598130841e-05,
+      "loss": 4.362,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3341121495327104e-05,
+      "loss": 4.2274,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3317757009345797e-05,
+      "loss": 4.2581,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3294392523364484e-05,
+      "loss": 4.4928,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3271028037383177e-05,
+      "loss": 4.0819,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.324766355140187e-05,
+      "loss": 4.2332,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3224299065420564e-05,
+      "loss": 4.1497,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.320093457943925e-05,
+      "loss": 4.5181,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3177570093457944e-05,
+      "loss": 4.253,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3154205607476637e-05,
+      "loss": 4.0871,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3130841121495327e-05,
+      "loss": 4.583,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3107476635514017e-05,
+      "loss": 4.3745,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.308411214953271e-05,
+      "loss": 4.4024,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3060747663551404e-05,
+      "loss": 4.374,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3037383177570094e-05,
+      "loss": 4.1955,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3014018691588784e-05,
+      "loss": 4.2214,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2990654205607477e-05,
+      "loss": 4.1361,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2967289719626167e-05,
+      "loss": 4.2242,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.294392523364486e-05,
+      "loss": 4.2808,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.292056074766355e-05,
+      "loss": 4.1916,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2897196261682244e-05,
+      "loss": 3.9273,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2873831775700934e-05,
+      "loss": 4.3042,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2850467289719628e-05,
+      "loss": 4.3704,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2827102803738318e-05,
+      "loss": 4.2526,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2803738317757008e-05,
+      "loss": 4.2533,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.27803738317757e-05,
+      "loss": 4.2388,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2757009345794394e-05,
+      "loss": 4.0274,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2733644859813084e-05,
+      "loss": 4.1831,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2710280373831774e-05,
+      "loss": 4.1491,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2686915887850468e-05,
+      "loss": 4.1519,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.266355140186916e-05,
+      "loss": 4.2317,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2640186915887848e-05,
+      "loss": 4.1775,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.261682242990654e-05,
+      "loss": 4.1761,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2593457943925235e-05,
+      "loss": 4.3685,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2570093457943928e-05,
+      "loss": 4.084,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2546728971962615e-05,
+      "loss": 4.527,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2523364485981308e-05,
+      "loss": 4.3357,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.25e-05,
+      "loss": 4.1422,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.247663551401869e-05,
+      "loss": 4.3068,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2453271028037385e-05,
+      "loss": 4.3026,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2429906542056075e-05,
+      "loss": 4.1574,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2406542056074768e-05,
+      "loss": 4.1831,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2383177570093458e-05,
+      "loss": 4.1569,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.235981308411215e-05,
+      "loss": 4.5027,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.233644859813084e-05,
+      "loss": 4.4855,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.231308411214953e-05,
+      "loss": 4.3037,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2289719626168225e-05,
+      "loss": 3.9236,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.226635514018692e-05,
+      "loss": 3.9739,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.224299065420561e-05,
+      "loss": 4.2009,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.22196261682243e-05,
+      "loss": 4.3131,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2196261682242992e-05,
+      "loss": 4.2057,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2172897196261685e-05,
+      "loss": 4.0355,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2149532710280372e-05,
+      "loss": 4.3406,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2126168224299065e-05,
+      "loss": 4.1452,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.210280373831776e-05,
+      "loss": 4.1109,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2079439252336452e-05,
+      "loss": 4.1571,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.205607476635514e-05,
+      "loss": 4.3227,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2032710280373832e-05,
+      "loss": 4.017,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2009345794392525e-05,
+      "loss": 4.1389,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.1985981308411215e-05,
+      "loss": 3.9873,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1962616822429905e-05,
+      "loss": 4.4358,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.19392523364486e-05,
+      "loss": 4.6147,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1915887850467292e-05,
+      "loss": 5.0169,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1892523364485982e-05,
+      "loss": 4.2392,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1869158878504672e-05,
+      "loss": 4.1059,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1845794392523366e-05,
+      "loss": 4.4478,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.182242990654206e-05,
+      "loss": 4.2442,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.179906542056075e-05,
+      "loss": 4.2118,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.177570093457944e-05,
+      "loss": 4.1839,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1752336448598132e-05,
+      "loss": 4.1408,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1728971962616822e-05,
+      "loss": 4.4276,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1705607476635516e-05,
+      "loss": 4.2676,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1682242990654206e-05,
+      "loss": 4.4974,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.16588785046729e-05,
+      "loss": 3.8399,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.163551401869159e-05,
+      "loss": 4.192,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1612149532710283e-05,
+      "loss": 4.2465,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1588785046728973e-05,
+      "loss": 4.0557,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1565420560747663e-05,
+      "loss": 4.11,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1542056074766356e-05,
+      "loss": 4.2677,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.151869158878505e-05,
+      "loss": 4.2174,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.149532710280374e-05,
+      "loss": 4.3056,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.147196261682243e-05,
+      "loss": 4.2743,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1448598130841123e-05,
+      "loss": 4.1441,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1425233644859816e-05,
+      "loss": 4.1886,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1401869158878503e-05,
+      "loss": 4.3016,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1378504672897196e-05,
+      "loss": 4.6068,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.135514018691589e-05,
+      "loss": 4.1668,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1331775700934583e-05,
+      "loss": 4.3656,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.130841121495327e-05,
+      "loss": 4.2676,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1285046728971963e-05,
+      "loss": 4.5382,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1261682242990657e-05,
+      "loss": 4.2992,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1238317757009347e-05,
+      "loss": 4.1328,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1214953271028037e-05,
+      "loss": 4.3023,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.119158878504673e-05,
+      "loss": 4.466,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1168224299065423e-05,
+      "loss": 3.9738,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1144859813084113e-05,
+      "loss": 4.2629,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1121495327102803e-05,
+      "loss": 4.4319,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1098130841121497e-05,
+      "loss": 4.3974,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1074766355140187e-05,
+      "loss": 4.4638,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.105140186915888e-05,
+      "loss": 4.6437,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.102803738317757e-05,
+      "loss": 4.0386,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1004672897196264e-05,
+      "loss": 3.9846,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0981308411214954e-05,
+      "loss": 4.4017,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0957943925233647e-05,
+      "loss": 4.2956,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0934579439252337e-05,
+      "loss": 4.3848,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0911214953271027e-05,
+      "loss": 4.6985,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.088785046728972e-05,
+      "loss": 4.31,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0864485981308414e-05,
+      "loss": 4.2907,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0841121495327104e-05,
+      "loss": 4.0717,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0817757009345794e-05,
+      "loss": 4.2242,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0794392523364487e-05,
+      "loss": 4.7169,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.077102803738318e-05,
+      "loss": 4.4339,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0747663551401867e-05,
+      "loss": 4.859,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.072429906542056e-05,
+      "loss": 4.2538,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0700934579439254e-05,
+      "loss": 4.1148,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0677570093457947e-05,
+      "loss": 4.581,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0654205607476634e-05,
+      "loss": 4.1642,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0630841121495327e-05,
+      "loss": 4.2891,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.060747663551402e-05,
+      "loss": 4.4619,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.058411214953271e-05,
+      "loss": 4.2654,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.05607476635514e-05,
+      "loss": 4.1666,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0537383177570094e-05,
+      "loss": 4.3177,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0514018691588788e-05,
+      "loss": 4.1332,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0490654205607478e-05,
+      "loss": 4.2219,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0467289719626168e-05,
+      "loss": 4.0597,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.044392523364486e-05,
+      "loss": 4.2189,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.042056074766355e-05,
+      "loss": 4.1731,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0397196261682244e-05,
+      "loss": 4.461,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0373831775700934e-05,
+      "loss": 4.2649,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0350467289719628e-05,
+      "loss": 4.1877,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0327102803738318e-05,
+      "loss": 4.3274,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.030373831775701e-05,
+      "loss": 4.215,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.02803738317757e-05,
+      "loss": 3.926,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.025700934579439e-05,
+      "loss": 4.1644,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0233644859813085e-05,
+      "loss": 3.8874,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0210280373831778e-05,
+      "loss": 4.1391,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0186915887850468e-05,
+      "loss": 4.2274,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0163551401869158e-05,
+      "loss": 4.3098,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.014018691588785e-05,
+      "loss": 4.4181,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0116822429906545e-05,
+      "loss": 4.1241,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.009345794392523e-05,
+      "loss": 4.0162,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0070093457943925e-05,
+      "loss": 4.0678,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0046728971962618e-05,
+      "loss": 4.4035,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.002336448598131e-05,
+      "loss": 4.4717,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 4.1201,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997663551401869e-05,
+      "loss": 4.2463,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9953271028037385e-05,
+      "loss": 4.3228,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9929906542056075e-05,
+      "loss": 4.3491,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9906542056074765e-05,
+      "loss": 4.3187,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.988317757009346e-05,
+      "loss": 4.1501,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9859813084112152e-05,
+      "loss": 3.9318,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9836448598130842e-05,
+      "loss": 3.9416,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9813084112149532e-05,
+      "loss": 4.386,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9789719626168225e-05,
+      "loss": 4.2767,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9766355140186915e-05,
+      "loss": 4.2735,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.974299065420561e-05,
+      "loss": 4.3571,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97196261682243e-05,
+      "loss": 4.1921,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696261682242992e-05,
+      "loss": 4.36,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9672897196261682e-05,
+      "loss": 4.3099,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9649532710280375e-05,
+      "loss": 3.8311,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9626168224299065e-05,
+      "loss": 4.344,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.960280373831776e-05,
+      "loss": 5.0039,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.957943925233645e-05,
+      "loss": 3.9486,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9556074766355142e-05,
+      "loss": 4.1195,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9532710280373832e-05,
+      "loss": 4.3297,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9509345794392522e-05,
+      "loss": 4.2691,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9485981308411216e-05,
+      "loss": 3.943,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.946261682242991e-05,
+      "loss": 4.3363,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.94392523364486e-05,
+      "loss": 4.1259,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.941588785046729e-05,
+      "loss": 4.2742,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9392523364485982e-05,
+      "loss": 4.0067,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9369158878504676e-05,
+      "loss": 4.5069,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9345794392523363e-05,
+      "loss": 4.1197,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9322429906542056e-05,
+      "loss": 4.0948,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.929906542056075e-05,
+      "loss": 4.1632,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9275700934579443e-05,
+      "loss": 4.085,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.925233644859813e-05,
+      "loss": 4.2531,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9228971962616823e-05,
+      "loss": 4.1484,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9205607476635516e-05,
+      "loss": 4.038,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9182242990654206e-05,
+      "loss": 4.3566,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9158878504672896e-05,
+      "loss": 4.1035,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.913551401869159e-05,
+      "loss": 4.1524,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9112149532710283e-05,
+      "loss": 4.1356,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9088785046728973e-05,
+      "loss": 4.1174,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9065420560747663e-05,
+      "loss": 4.0761,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9042056074766356e-05,
+      "loss": 4.2321,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9018691588785046e-05,
+      "loss": 4.532,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.899532710280374e-05,
+      "loss": 4.3225,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.897196261682243e-05,
+      "loss": 4.2305,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8948598130841123e-05,
+      "loss": 4.525,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8925233644859813e-05,
+      "loss": 4.103,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8901869158878507e-05,
+      "loss": 4.0992,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8878504672897197e-05,
+      "loss": 4.0067,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8855140186915887e-05,
+      "loss": 3.8331,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.883177570093458e-05,
+      "loss": 3.8962,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808411214953273e-05,
+      "loss": 4.3257,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785046728971963e-05,
+      "loss": 4.288,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8761682242990653e-05,
+      "loss": 4.0812,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8738317757009347e-05,
+      "loss": 4.0414,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.871495327102804e-05,
+      "loss": 4.3498,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8691588785046727e-05,
+      "loss": 4.1599,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.866822429906542e-05,
+      "loss": 4.3491,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8644859813084114e-05,
+      "loss": 4.0596,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8621495327102807e-05,
+      "loss": 4.3239,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8598130841121494e-05,
+      "loss": 4.6033,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8574766355140187e-05,
+      "loss": 4.4095,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.855140186915888e-05,
+      "loss": 4.3342,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.852803738317757e-05,
+      "loss": 4.148,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.850467289719626e-05,
+      "loss": 3.7929,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8481308411214954e-05,
+      "loss": 4.1604,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8457943925233647e-05,
+      "loss": 4.8418,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8434579439252337e-05,
+      "loss": 5.1553,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8411214953271027e-05,
+      "loss": 4.4493,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 4.264136791229248,
+      "eval_runtime": 907.6252,
+      "eval_samples_per_second": 2.911,
+      "eval_steps_per_second": 0.365,
+      "eval_wer": 1.9047005156683856,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.838785046728972e-05,
+      "loss": 4.3957,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.836448598130841e-05,
+      "loss": 4.1981,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8341121495327104e-05,
+      "loss": 4.3142,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8317757009345794e-05,
+      "loss": 4.1964,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8294392523364487e-05,
+      "loss": 4.1522,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8271028037383177e-05,
+      "loss": 4.2069,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.824766355140187e-05,
+      "loss": 4.3269,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.822429906542056e-05,
+      "loss": 4.3425,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.820093457943925e-05,
+      "loss": 4.2841,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8177570093457944e-05,
+      "loss": 4.084,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8154205607476638e-05,
+      "loss": 4.2206,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8130841121495328e-05,
+      "loss": 4.1333,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8107476635514018e-05,
+      "loss": 4.2549,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.808411214953271e-05,
+      "loss": 4.3198,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8060747663551404e-05,
+      "loss": 4.6233,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.803738317757009e-05,
+      "loss": 4.1943,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8014018691588784e-05,
+      "loss": 4.2945,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7990654205607478e-05,
+      "loss": 4.1784,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.796728971962617e-05,
+      "loss": 4.2729,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7943925233644858e-05,
+      "loss": 4.2215,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.792056074766355e-05,
+      "loss": 4.2098,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7897196261682245e-05,
+      "loss": 4.2428,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7873831775700935e-05,
+      "loss": 4.2183,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7850467289719625e-05,
+      "loss": 4.2025,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7827102803738318e-05,
+      "loss": 4.1428,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.780373831775701e-05,
+      "loss": 3.846,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.77803738317757e-05,
+      "loss": 4.2519,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.775700934579439e-05,
+      "loss": 4.3163,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7733644859813085e-05,
+      "loss": 4.1913,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7710280373831775e-05,
+      "loss": 4.2665,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7686915887850468e-05,
+      "loss": 4.3115,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7663551401869158e-05,
+      "loss": 4.0578,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.764018691588785e-05,
+      "loss": 4.2709,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.761682242990654e-05,
+      "loss": 3.9123,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7593457943925235e-05,
+      "loss": 4.1373,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7570093457943925e-05,
+      "loss": 4.2593,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7546728971962615e-05,
+      "loss": 4.2973,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.752336448598131e-05,
+      "loss": 4.2134,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 4.2375,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7476635514018692e-05,
+      "loss": 4.4934,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7453271028037382e-05,
+      "loss": 4.4026,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7429906542056075e-05,
+      "loss": 4.4204,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.740654205607477e-05,
+      "loss": 4.4592,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.738317757009346e-05,
+      "loss": 4.1426,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.735981308411215e-05,
+      "loss": 4.4057,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7336448598130842e-05,
+      "loss": 4.2709,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7313084112149535e-05,
+      "loss": 4.2221,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7289719626168222e-05,
+      "loss": 4.1426,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7266355140186915e-05,
+      "loss": 4.3478,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.724299065420561e-05,
+      "loss": 5.1019,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.72196261682243e-05,
+      "loss": 4.6266,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.719626168224299e-05,
+      "loss": 4.102,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7172897196261682e-05,
+      "loss": 4.0698,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7149532710280376e-05,
+      "loss": 3.8926,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7126168224299066e-05,
+      "loss": 4.1157,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7102803738317756e-05,
+      "loss": 4.1241,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.707943925233645e-05,
+      "loss": 4.4176,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7056074766355142e-05,
+      "loss": 4.0995,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7032710280373833e-05,
+      "loss": 4.3256,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7009345794392523e-05,
+      "loss": 4.0139,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6985981308411216e-05,
+      "loss": 4.2708,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6962616822429906e-05,
+      "loss": 4.2298,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.69392523364486e-05,
+      "loss": 4.4264,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.691588785046729e-05,
+      "loss": 3.8271,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6892523364485983e-05,
+      "loss": 4.2122,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6869158878504673e-05,
+      "loss": 3.7996,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6845794392523366e-05,
+      "loss": 4.2465,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6822429906542056e-05,
+      "loss": 4.0495,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6799065420560746e-05,
+      "loss": 4.4378,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.677570093457944e-05,
+      "loss": 3.9142,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6752336448598133e-05,
+      "loss": 4.1469,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6728971962616823e-05,
+      "loss": 3.9169,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6705607476635513e-05,
+      "loss": 4.0311,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6682242990654206e-05,
+      "loss": 4.2565,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.66588785046729e-05,
+      "loss": 4.1096,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6635514018691586e-05,
+      "loss": 4.3814,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.661214953271028e-05,
+      "loss": 4.1697,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6588785046728973e-05,
+      "loss": 4.1184,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6565420560747667e-05,
+      "loss": 4.1549,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6542056074766353e-05,
+      "loss": 4.2968,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6518691588785047e-05,
+      "loss": 4.1357,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.649532710280374e-05,
+      "loss": 4.1343,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.647196261682243e-05,
+      "loss": 3.9211,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.644859813084112e-05,
+      "loss": 4.182,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6425233644859813e-05,
+      "loss": 4.0629,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6401869158878507e-05,
+      "loss": 4.3143,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6378504672897197e-05,
+      "loss": 4.4086,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6355140186915887e-05,
+      "loss": 4.2333,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633177570093458e-05,
+      "loss": 4.337,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.630841121495327e-05,
+      "loss": 4.0149,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6285046728971964e-05,
+      "loss": 4.0734,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6261682242990654e-05,
+      "loss": 4.2937,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6238317757009347e-05,
+      "loss": 4.3993,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6214953271028037e-05,
+      "loss": 4.4215,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.619158878504673e-05,
+      "loss": 4.3007,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.616822429906542e-05,
+      "loss": 4.7796,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.614485981308411e-05,
+      "loss": 4.3896,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6121495327102804e-05,
+      "loss": 4.937,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6098130841121497e-05,
+      "loss": 4.3732,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6074766355140187e-05,
+      "loss": 4.7071,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6051401869158877e-05,
+      "loss": 3.8142,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.602803738317757e-05,
+      "loss": 4.3234,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6004672897196264e-05,
+      "loss": 4.2441,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.598130841121495e-05,
+      "loss": 3.9535,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5957943925233644e-05,
+      "loss": 3.9524,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5934579439252337e-05,
+      "loss": 4.0232,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.591121495327103e-05,
+      "loss": 3.8744,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5887850467289717e-05,
+      "loss": 3.9939,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.586448598130841e-05,
+      "loss": 4.3436,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5841121495327104e-05,
+      "loss": 4.2237,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5817757009345794e-05,
+      "loss": 3.9764,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5794392523364484e-05,
+      "loss": 4.3748,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5771028037383178e-05,
+      "loss": 3.7894,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.574766355140187e-05,
+      "loss": 4.2248,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.572429906542056e-05,
+      "loss": 4.3594,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.570093457943925e-05,
+      "loss": 4.2596,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5677570093457944e-05,
+      "loss": 4.1759,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5654205607476634e-05,
+      "loss": 4.1523,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5630841121495328e-05,
+      "loss": 4.1149,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5607476635514018e-05,
+      "loss": 4.0031,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.558411214953271e-05,
+      "loss": 4.2304,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.55607476635514e-05,
+      "loss": 4.1218,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5537383177570095e-05,
+      "loss": 3.9554,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5514018691588785e-05,
+      "loss": 4.0651,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5490654205607475e-05,
+      "loss": 4.4828,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5467289719626168e-05,
+      "loss": 4.0348,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.544392523364486e-05,
+      "loss": 4.0473,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.542056074766355e-05,
+      "loss": 4.5651,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.539719626168224e-05,
+      "loss": 3.8391,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5373831775700935e-05,
+      "loss": 4.4066,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5350467289719628e-05,
+      "loss": 4.3283,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5327102803738315e-05,
+      "loss": 4.1313,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5303738317757008e-05,
+      "loss": 4.1537,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.52803738317757e-05,
+      "loss": 4.196,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5257009345794393e-05,
+      "loss": 4.4604,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5233644859813083e-05,
+      "loss": 4.1326,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5210280373831775e-05,
+      "loss": 4.1546,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5186915887850468e-05,
+      "loss": 4.3823,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.516355140186916e-05,
+      "loss": 4.0458,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5140186915887848e-05,
+      "loss": 4.1307,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5116822429906542e-05,
+      "loss": 4.15,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5093457943925234e-05,
+      "loss": 4.6058,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5070093457943927e-05,
+      "loss": 4.3072,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5046728971962615e-05,
+      "loss": 4.4706,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5023364485981309e-05,
+      "loss": 4.1741,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5e-05,
+      "loss": 4.3202,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4976635514018692e-05,
+      "loss": 4.2725,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4953271028037384e-05,
+      "loss": 4.1745,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4929906542056075e-05,
+      "loss": 4.422,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4906542056074767e-05,
+      "loss": 4.203,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4883177570093459e-05,
+      "loss": 4.1221,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.485981308411215e-05,
+      "loss": 4.2642,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4836448598130842e-05,
+      "loss": 4.259,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4813084112149532e-05,
+      "loss": 4.309,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4789719626168226e-05,
+      "loss": 4.3005,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4766355140186916e-05,
+      "loss": 4.3132,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4742990654205609e-05,
+      "loss": 4.1137,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4719626168224299e-05,
+      "loss": 4.1557,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4696261682242993e-05,
+      "loss": 4.0059,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4672897196261683e-05,
+      "loss": 3.9281,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4649532710280374e-05,
+      "loss": 4.2902,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4626168224299066e-05,
+      "loss": 4.1478,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4602803738317758e-05,
+      "loss": 4.1787,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.457943925233645e-05,
+      "loss": 4.014,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4556074766355141e-05,
+      "loss": 4.0444,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4532710280373833e-05,
+      "loss": 4.26,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4509345794392524e-05,
+      "loss": 4.1584,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4485981308411214e-05,
+      "loss": 3.8993,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4462616822429908e-05,
+      "loss": 4.2507,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4439252336448598e-05,
+      "loss": 4.3269,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4415887850467291e-05,
+      "loss": 3.9758,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4392523364485981e-05,
+      "loss": 4.0067,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4369158878504675e-05,
+      "loss": 4.3608,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4345794392523365e-05,
+      "loss": 3.9976,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4322429906542058e-05,
+      "loss": 4.0712,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4299065420560748e-05,
+      "loss": 4.3114,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.427570093457944e-05,
+      "loss": 4.2522,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4252336448598131e-05,
+      "loss": 4.2322,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4228971962616823e-05,
+      "loss": 3.798,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4205607476635515e-05,
+      "loss": 4.1185,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4182242990654207e-05,
+      "loss": 4.0063,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4158878504672898e-05,
+      "loss": 4.0903,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.413551401869159e-05,
+      "loss": 4.0785,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.411214953271028e-05,
+      "loss": 3.9893,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4088785046728973e-05,
+      "loss": 4.5213,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4065420560747663e-05,
+      "loss": 3.9469,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4042056074766357e-05,
+      "loss": 4.2824,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4018691588785047e-05,
+      "loss": 4.0892,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.399532710280374e-05,
+      "loss": 3.95,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.397196261682243e-05,
+      "loss": 3.8737,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3948598130841122e-05,
+      "loss": 4.2447,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3925233644859814e-05,
+      "loss": 4.2337,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3901869158878505e-05,
+      "loss": 4.6481,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3878504672897197e-05,
+      "loss": 4.1727,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3855140186915889e-05,
+      "loss": 4.5153,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.383177570093458e-05,
+      "loss": 4.0322,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3808411214953272e-05,
+      "loss": 4.1188,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3785046728971962e-05,
+      "loss": 4.45,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3761682242990655e-05,
+      "loss": 4.3541,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3738317757009345e-05,
+      "loss": 4.0267,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3714953271028039e-05,
+      "loss": 4.0902,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3691588785046729e-05,
+      "loss": 4.0704,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3668224299065422e-05,
+      "loss": 4.0326,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3644859813084112e-05,
+      "loss": 3.9149,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3621495327102804e-05,
+      "loss": 3.8928,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3598130841121496e-05,
+      "loss": 4.1782,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3574766355140187e-05,
+      "loss": 4.4078,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3551401869158879e-05,
+      "loss": 4.1271,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.352803738317757e-05,
+      "loss": 4.0725,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3504672897196263e-05,
+      "loss": 3.8818,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3481308411214954e-05,
+      "loss": 4.2811,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3457943925233644e-05,
+      "loss": 4.2066,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3434579439252338e-05,
+      "loss": 4.2165,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3411214953271028e-05,
+      "loss": 4.0162,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3387850467289721e-05,
+      "loss": 4.082,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3364485981308411e-05,
+      "loss": 4.159,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3341121495327104e-05,
+      "loss": 3.999,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3317757009345794e-05,
+      "loss": 3.9556,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3294392523364486e-05,
+      "loss": 3.9182,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3271028037383178e-05,
+      "loss": 4.3089,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.324766355140187e-05,
+      "loss": 4.0974,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3224299065420561e-05,
+      "loss": 3.851,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3200934579439253e-05,
+      "loss": 4.2691,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3177570093457945e-05,
+      "loss": 4.1825,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3154205607476636e-05,
+      "loss": 4.3331,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3130841121495328e-05,
+      "loss": 4.24,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.310747663551402e-05,
+      "loss": 4.055,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.308411214953271e-05,
+      "loss": 4.0521,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3060747663551403e-05,
+      "loss": 4.0261,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3037383177570093e-05,
+      "loss": 4.2366,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3014018691588787e-05,
+      "loss": 4.169,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2990654205607477e-05,
+      "loss": 4.1076,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.296728971962617e-05,
+      "loss": 4.072,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.294392523364486e-05,
+      "loss": 4.0198,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2920560747663552e-05,
+      "loss": 4.4133,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2897196261682243e-05,
+      "loss": 4.1118,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2873831775700935e-05,
+      "loss": 4.1442,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2850467289719627e-05,
+      "loss": 3.9661,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2827102803738318e-05,
+      "loss": 4.1963,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.280373831775701e-05,
+      "loss": 4.2408,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2780373831775702e-05,
+      "loss": 4.1973,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2757009345794392e-05,
+      "loss": 4.5804,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2733644859813085e-05,
+      "loss": 4.2627,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2710280373831775e-05,
+      "loss": 4.2332,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2686915887850469e-05,
+      "loss": 3.9109,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2663551401869159e-05,
+      "loss": 4.3764,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2640186915887852e-05,
+      "loss": 4.3619,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2616822429906542e-05,
+      "loss": 4.2002,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2593457943925234e-05,
+      "loss": 4.3737,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2570093457943925e-05,
+      "loss": 4.4729,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2546728971962617e-05,
+      "loss": 4.2608,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2523364485981309e-05,
+      "loss": 4.1423,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.25e-05,
+      "loss": 4.1753,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2476635514018692e-05,
+      "loss": 4.2292,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2453271028037384e-05,
+      "loss": 4.3562,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2429906542056074e-05,
+      "loss": 4.0769,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2406542056074767e-05,
+      "loss": 4.0141,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2383177570093457e-05,
+      "loss": 3.871,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.235981308411215e-05,
+      "loss": 4.127,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.233644859813084e-05,
+      "loss": 4.5144,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2313084112149534e-05,
+      "loss": 3.9517,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2289719626168224e-05,
+      "loss": 4.2184,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2266355140186916e-05,
+      "loss": 4.2403,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2242990654205608e-05,
+      "loss": 4.1123,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.22196261682243e-05,
+      "loss": 4.3894,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2196261682242991e-05,
+      "loss": 4.247,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2172897196261683e-05,
+      "loss": 4.0374,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2149532710280374e-05,
+      "loss": 4.029,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2126168224299066e-05,
+      "loss": 4.2308,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2102803738317758e-05,
+      "loss": 4.0277,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.207943925233645e-05,
+      "loss": 4.099,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.205607476635514e-05,
+      "loss": 4.1056,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2032710280373833e-05,
+      "loss": 4.3618,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2009345794392523e-05,
+      "loss": 4.0625,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1985981308411216e-05,
+      "loss": 4.1225,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1962616822429906e-05,
+      "loss": 4.4333,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.19392523364486e-05,
+      "loss": 4.2453,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.191588785046729e-05,
+      "loss": 4.1666,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1892523364485981e-05,
+      "loss": 4.3033,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1869158878504673e-05,
+      "loss": 4.2379,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1845794392523365e-05,
+      "loss": 4.1905,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1822429906542057e-05,
+      "loss": 4.0858,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1799065420560748e-05,
+      "loss": 4.2976,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.177570093457944e-05,
+      "loss": 4.1154,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1752336448598132e-05,
+      "loss": 4.1099,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1728971962616822e-05,
+      "loss": 4.207,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1705607476635515e-05,
+      "loss": 4.2139,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1682242990654205e-05,
+      "loss": 4.1284,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1658878504672898e-05,
+      "loss": 4.1511,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1635514018691588e-05,
+      "loss": 3.8309,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1612149532710282e-05,
+      "loss": 4.2548,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1588785046728972e-05,
+      "loss": 4.3287,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1565420560747664e-05,
+      "loss": 4.6614,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1542056074766355e-05,
+      "loss": 4.0532,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1518691588785047e-05,
+      "loss": 4.0075,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1495327102803739e-05,
+      "loss": 3.8986,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.147196261682243e-05,
+      "loss": 4.0358,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1448598130841122e-05,
+      "loss": 3.7676,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1425233644859814e-05,
+      "loss": 4.5986,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1401869158878504e-05,
+      "loss": 3.9657,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1378504672897197e-05,
+      "loss": 4.1888,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1355140186915887e-05,
+      "loss": 4.4359,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.133177570093458e-05,
+      "loss": 4.1384,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.130841121495327e-05,
+      "loss": 4.1835,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1285046728971964e-05,
+      "loss": 4.0106,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1261682242990654e-05,
+      "loss": 4.1476,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1238317757009346e-05,
+      "loss": 4.1366,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1214953271028037e-05,
+      "loss": 3.9199,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1191588785046729e-05,
+      "loss": 4.1007,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.116822429906542e-05,
+      "loss": 3.9246,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.1144859813084113e-05,
+      "loss": 4.0436,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.1121495327102804e-05,
+      "loss": 4.2192,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.1098130841121496e-05,
+      "loss": 4.1179,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.1074766355140186e-05,
+      "loss": 3.9699,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.105140186915888e-05,
+      "loss": 4.3127,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.102803738317757e-05,
+      "loss": 4.0864,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.1004672897196263e-05,
+      "loss": 4.0028,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0981308411214953e-05,
+      "loss": 4.5721,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0957943925233646e-05,
+      "loss": 4.0788,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0934579439252336e-05,
+      "loss": 3.8027,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.091121495327103e-05,
+      "loss": 3.8412,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.088785046728972e-05,
+      "loss": 4.0666,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0864485981308411e-05,
+      "loss": 4.0204,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0841121495327103e-05,
+      "loss": 4.0542,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0817757009345795e-05,
+      "loss": 4.1382,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0794392523364486e-05,
+      "loss": 3.9716,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0771028037383178e-05,
+      "loss": 3.8684,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.074766355140187e-05,
+      "loss": 4.0632,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0724299065420561e-05,
+      "loss": 4.2209,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0700934579439251e-05,
+      "loss": 4.2317,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0677570093457945e-05,
+      "loss": 3.92,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0654205607476635e-05,
+      "loss": 4.1275,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0630841121495328e-05,
+      "loss": 4.1231,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0607476635514018e-05,
+      "loss": 4.1757,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0584112149532712e-05,
+      "loss": 3.9458,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0560747663551402e-05,
+      "loss": 4.0713,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0537383177570093e-05,
+      "loss": 4.2744,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0514018691588785e-05,
+      "loss": 4.2818,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0490654205607477e-05,
+      "loss": 4.3961,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0467289719626168e-05,
+      "loss": 4.569,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.044392523364486e-05,
+      "loss": 4.387,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0420560747663552e-05,
+      "loss": 4.0494,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0397196261682244e-05,
+      "loss": 4.3602,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0373831775700934e-05,
+      "loss": 4.1212,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0350467289719627e-05,
+      "loss": 4.2386,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0327102803738317e-05,
+      "loss": 4.3046,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.030373831775701e-05,
+      "loss": 4.667,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.02803738317757e-05,
+      "loss": 4.113,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0257009345794394e-05,
+      "loss": 4.1922,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0233644859813084e-05,
+      "loss": 4.5367,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0210280373831776e-05,
+      "loss": 4.2978,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0186915887850467e-05,
+      "loss": 4.2534,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0163551401869159e-05,
+      "loss": 4.198,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.014018691588785e-05,
+      "loss": 4.2726,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0116822429906542e-05,
+      "loss": 4.0672,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0093457943925234e-05,
+      "loss": 4.4141,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0070093457943926e-05,
+      "loss": 3.8552,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0046728971962616e-05,
+      "loss": 4.3052,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0023364485981309e-05,
+      "loss": 4.3766,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 4.5794,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.976635514018693e-06,
+      "loss": 4.1219,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.953271028037383e-06,
+      "loss": 3.9744,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.929906542056076e-06,
+      "loss": 3.9778,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.906542056074766e-06,
+      "loss": 4.3931,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.883177570093458e-06,
+      "loss": 4.1596,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.85981308411215e-06,
+      "loss": 3.8741,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.836448598130841e-06,
+      "loss": 4.1559,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.813084112149533e-06,
+      "loss": 3.9958,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.789719626168224e-06,
+      "loss": 4.3489,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.766355140186916e-06,
+      "loss": 4.3537,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.742990654205608e-06,
+      "loss": 4.2607,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.7196261682243e-06,
+      "loss": 3.9353,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.696261682242991e-06,
+      "loss": 4.1306,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.672897196261681e-06,
+      "loss": 4.2435,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.649532710280375e-06,
+      "loss": 3.9262,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.626168224299065e-06,
+      "loss": 3.8826,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.602803738317758e-06,
+      "loss": 4.0797,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.579439252336448e-06,
+      "loss": 4.1199,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.556074766355141e-06,
+      "loss": 4.1839,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.532710280373831e-06,
+      "loss": 4.019,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.509345794392523e-06,
+      "loss": 3.7721,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.485981308411215e-06,
+      "loss": 4.3469,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.462616822429907e-06,
+      "loss": 3.997,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.439252336448598e-06,
+      "loss": 4.2968,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.41588785046729e-06,
+      "loss": 4.4505,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.392523364485982e-06,
+      "loss": 4.1766,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.369158878504673e-06,
+      "loss": 4.3172,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.345794392523363e-06,
+      "loss": 4.2677,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.322429906542057e-06,
+      "loss": 4.1343,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.299065420560747e-06,
+      "loss": 4.1269,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.27570093457944e-06,
+      "loss": 3.9261,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.25233644859813e-06,
+      "loss": 4.0419,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.228971962616824e-06,
+      "loss": 4.17,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.205607476635514e-06,
+      "loss": 4.3503,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.182242990654205e-06,
+      "loss": 4.3083,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.158878504672897e-06,
+      "loss": 4.1669,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.135514018691589e-06,
+      "loss": 4.3395,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.11214953271028e-06,
+      "loss": 4.1712,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.088785046728972e-06,
+      "loss": 4.3581,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.065420560747664e-06,
+      "loss": 4.4303,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.042056074766356e-06,
+      "loss": 3.9564,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.018691588785046e-06,
+      "loss": 4.2175,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.995327102803739e-06,
+      "loss": 4.0643,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.971962616822429e-06,
+      "loss": 3.958,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.948598130841122e-06,
+      "loss": 3.9298,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.925233644859812e-06,
+      "loss": 4.4535,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.901869158878506e-06,
+      "loss": 4.1644,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.878504672897196e-06,
+      "loss": 3.9213,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.855140186915887e-06,
+      "loss": 4.0797,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.831775700934579e-06,
+      "loss": 3.896,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.80841121495327e-06,
+      "loss": 3.9582,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.785046728971963e-06,
+      "loss": 3.9769,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.761682242990654e-06,
+      "loss": 4.4139,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.738317757009346e-06,
+      "loss": 4.2634,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.714953271028038e-06,
+      "loss": 4.2068,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.69158878504673e-06,
+      "loss": 4.3161,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.668224299065421e-06,
+      "loss": 4.1566,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.644859813084111e-06,
+      "loss": 4.0683,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.621495327102804e-06,
+      "loss": 4.1944,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.598130841121494e-06,
+      "loss": 3.9461,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.574766355140188e-06,
+      "loss": 4.1013,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.551401869158878e-06,
+      "loss": 4.2032,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.528037383177571e-06,
+      "loss": 4.138,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.504672897196261e-06,
+      "loss": 4.0538,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.481308411214953e-06,
+      "loss": 4.2792,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.457943925233645e-06,
+      "loss": 4.3122,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.434579439252336e-06,
+      "loss": 4.0478,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.411214953271028e-06,
+      "loss": 4.0643,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.38785046728972e-06,
+      "loss": 4.192,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.364485981308411e-06,
+      "loss": 4.3633,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.341121495327103e-06,
+      "loss": 4.2542,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.317757009345793e-06,
+      "loss": 4.1836,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.294392523364487e-06,
+      "loss": 4.3832,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.271028037383177e-06,
+      "loss": 4.0672,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.24766355140187e-06,
+      "loss": 3.926,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.22429906542056e-06,
+      "loss": 3.9126,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.200934579439253e-06,
+      "loss": 4.3026,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.177570093457943e-06,
+      "loss": 3.9898,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.154205607476635e-06,
+      "loss": 4.2334,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.130841121495327e-06,
+      "loss": 4.1731,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.107476635514018e-06,
+      "loss": 4.2045,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.08411214953271e-06,
+      "loss": 4.0328,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.060747663551402e-06,
+      "loss": 4.0335,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.037383177570094e-06,
+      "loss": 4.1178,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.014018691588785e-06,
+      "loss": 4.2312,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.990654205607475e-06,
+      "loss": 4.3789,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.967289719626169e-06,
+      "loss": 4.2199,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.943925233644859e-06,
+      "loss": 4.5116,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.920560747663552e-06,
+      "loss": 4.5537,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.897196261682242e-06,
+      "loss": 4.1504,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.873831775700936e-06,
+      "loss": 4.0838,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.850467289719626e-06,
+      "loss": 4.1082,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.827102803738317e-06,
+      "loss": 3.9692,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.803738317757009e-06,
+      "loss": 3.911,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.7803738317757e-06,
+      "loss": 4.018,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.757009345794392e-06,
+      "loss": 4.0924,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.733644859813084e-06,
+      "loss": 4.3227,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.710280373831776e-06,
+      "loss": 4.1212,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.686915887850467e-06,
+      "loss": 4.0404,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.663551401869157e-06,
+      "loss": 4.373,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.64018691588785e-06,
+      "loss": 4.1756,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.616822429906542e-06,
+      "loss": 4.2344,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.593457943925234e-06,
+      "loss": 4.1485,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.570093457943924e-06,
+      "loss": 4.2634,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.546728971962617e-06,
+      "loss": 4.3867,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.523364485981308e-06,
+      "loss": 3.8309,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.5e-06,
+      "loss": 4.0638,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.476635514018692e-06,
+      "loss": 4.1175,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.453271028037384e-06,
+      "loss": 3.8194,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.429906542056075e-06,
+      "loss": 3.8826,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.406542056074766e-06,
+      "loss": 4.0709,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.383177570093458e-06,
+      "loss": 3.8683,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3598130841121496e-06,
+      "loss": 4.045,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.336448598130841e-06,
+      "loss": 3.9591,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.313084112149533e-06,
+      "loss": 4.2549,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.289719626168225e-06,
+      "loss": 4.1843,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.266355140186916e-06,
+      "loss": 4.2318,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.242990654205607e-06,
+      "loss": 3.8091,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.219626168224299e-06,
+      "loss": 4.1688,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.196261682242991e-06,
+      "loss": 3.9455,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.172897196261682e-06,
+      "loss": 4.0188,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.149532710280374e-06,
+      "loss": 4.2662,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.126168224299066e-06,
+      "loss": 4.1384,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1028037383177574e-06,
+      "loss": 3.7554,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.079439252336449e-06,
+      "loss": 4.0879,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.05607476635514e-06,
+      "loss": 4.249,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.032710280373832e-06,
+      "loss": 4.2345,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.009345794392523e-06,
+      "loss": 4.2024,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.985981308411215e-06,
+      "loss": 4.1977,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.962616822429907e-06,
+      "loss": 3.9674,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9392523364485985e-06,
+      "loss": 4.4402,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.91588785046729e-06,
+      "loss": 4.113,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.892523364485981e-06,
+      "loss": 3.7594,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.869158878504673e-06,
+      "loss": 3.9534,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8457943925233644e-06,
+      "loss": 4.0974,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.822429906542056e-06,
+      "loss": 4.5972,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.799065420560748e-06,
+      "loss": 4.1733,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7757009345794396e-06,
+      "loss": 4.4564,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.752336448598131e-06,
+      "loss": 3.969,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.728971962616822e-06,
+      "loss": 4.2715,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 4.193031311035156,
+      "eval_runtime": 914.8759,
+      "eval_samples_per_second": 2.888,
+      "eval_steps_per_second": 0.362,
+      "eval_wer": 1.8950614835382784,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.705607476635514e-06,
+      "loss": 4.2,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6822429906542055e-06,
+      "loss": 4.1476,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.658878504672897e-06,
+      "loss": 4.126,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.635514018691589e-06,
+      "loss": 4.0626,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.612149532710281e-06,
+      "loss": 4.2062,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.588785046728972e-06,
+      "loss": 4.0804,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.565420560747664e-06,
+      "loss": 3.9782,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.542056074766355e-06,
+      "loss": 4.1243,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.518691588785047e-06,
+      "loss": 4.1398,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.495327102803738e-06,
+      "loss": 4.073,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.47196261682243e-06,
+      "loss": 3.9131,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.448598130841122e-06,
+      "loss": 4.2426,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.425233644859813e-06,
+      "loss": 3.8792,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.401869158878505e-06,
+      "loss": 4.0344,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.378504672897196e-06,
+      "loss": 4.0333,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.355140186915888e-06,
+      "loss": 4.405,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.331775700934579e-06,
+      "loss": 3.9909,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.308411214953271e-06,
+      "loss": 4.1977,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.285046728971963e-06,
+      "loss": 4.3117,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.2616822429906544e-06,
+      "loss": 4.0664,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.238317757009346e-06,
+      "loss": 4.0256,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.214953271028037e-06,
+      "loss": 4.0164,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.191588785046729e-06,
+      "loss": 4.2854,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.16822429906542e-06,
+      "loss": 3.9579,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.144859813084112e-06,
+      "loss": 4.1027,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.121495327102804e-06,
+      "loss": 3.927,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0981308411214955e-06,
+      "loss": 4.0293,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.074766355140187e-06,
+      "loss": 4.0612,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.051401869158879e-06,
+      "loss": 3.6883,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.02803738317757e-06,
+      "loss": 3.926,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0046728971962615e-06,
+      "loss": 3.9036,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.981308411214953e-06,
+      "loss": 3.9013,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.957943925233645e-06,
+      "loss": 4.0288,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.934579439252337e-06,
+      "loss": 4.0523,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.911214953271028e-06,
+      "loss": 4.0859,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.88785046728972e-06,
+      "loss": 3.8448,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.864485981308411e-06,
+      "loss": 4.0659,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.8411214953271025e-06,
+      "loss": 4.3504,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.817757009345794e-06,
+      "loss": 4.021,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.794392523364486e-06,
+      "loss": 4.1701,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.771028037383178e-06,
+      "loss": 4.2345,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.747663551401869e-06,
+      "loss": 3.8187,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.724299065420561e-06,
+      "loss": 4.2731,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.700934579439252e-06,
+      "loss": 4.1202,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.677570093457944e-06,
+      "loss": 4.5758,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.654205607476635e-06,
+      "loss": 3.9231,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.630841121495327e-06,
+      "loss": 4.2574,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.607476635514019e-06,
+      "loss": 4.2299,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.58411214953271e-06,
+      "loss": 4.5003,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.560747663551402e-06,
+      "loss": 3.9673,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.537383177570093e-06,
+      "loss": 4.3467,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.514018691588785e-06,
+      "loss": 4.1563,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.490654205607476e-06,
+      "loss": 4.0208,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.467289719626168e-06,
+      "loss": 3.9448,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.44392523364486e-06,
+      "loss": 4.2711,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.4205607476635515e-06,
+      "loss": 4.4785,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.397196261682243e-06,
+      "loss": 4.2061,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.373831775700935e-06,
+      "loss": 4.2728,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.350467289719626e-06,
+      "loss": 3.9019,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3271028037383174e-06,
+      "loss": 4.2276,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.303738317757009e-06,
+      "loss": 4.1641,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.280373831775701e-06,
+      "loss": 4.0502,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.2570093457943925e-06,
+      "loss": 4.2459,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.233644859813084e-06,
+      "loss": 4.0027,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.210280373831776e-06,
+      "loss": 4.2141,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.186915887850467e-06,
+      "loss": 3.9442,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.1635514018691585e-06,
+      "loss": 3.8354,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.14018691588785e-06,
+      "loss": 4.2116,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.116822429906542e-06,
+      "loss": 4.1787,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.093457943925234e-06,
+      "loss": 4.0119,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.070093457943925e-06,
+      "loss": 4.0782,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.046728971962617e-06,
+      "loss": 4.3198,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.023364485981308e-06,
+      "loss": 4.2457,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9999999999999996e-06,
+      "loss": 3.9455,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.976635514018691e-06,
+      "loss": 3.8951,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.953271028037383e-06,
+      "loss": 4.1544,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.929906542056075e-06,
+      "loss": 3.9227,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.906542056074766e-06,
+      "loss": 4.2647,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.883177570093458e-06,
+      "loss": 4.183,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.85981308411215e-06,
+      "loss": 4.3395,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.836448598130841e-06,
+      "loss": 4.2104,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.813084112149532e-06,
+      "loss": 4.3061,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.789719626168224e-06,
+      "loss": 4.1218,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.766355140186916e-06,
+      "loss": 4.0206,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7429906542056074e-06,
+      "loss": 3.9921,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719626168224299e-06,
+      "loss": 4.057,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.696261682242991e-06,
+      "loss": 4.0119,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.672897196261682e-06,
+      "loss": 4.2471,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.649532710280373e-06,
+      "loss": 4.0784,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.626168224299065e-06,
+      "loss": 3.9207,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.602803738317757e-06,
+      "loss": 4.0995,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5794392523364485e-06,
+      "loss": 3.985,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.55607476635514e-06,
+      "loss": 3.9945,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.532710280373832e-06,
+      "loss": 4.1583,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.509345794392523e-06,
+      "loss": 4.0745,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4859813084112145e-06,
+      "loss": 4.3321,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.462616822429906e-06,
+      "loss": 4.2305,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439252336448598e-06,
+      "loss": 4.4488,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4158878504672896e-06,
+      "loss": 5.0833,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.392523364485981e-06,
+      "loss": 4.6915,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.369158878504673e-06,
+      "loss": 4.0953,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.345794392523365e-06,
+      "loss": 3.9254,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3224299065420555e-06,
+      "loss": 4.0254,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.299065420560747e-06,
+      "loss": 4.1379,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.275700934579439e-06,
+      "loss": 3.9132,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.252336448598131e-06,
+      "loss": 4.1435,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.228971962616822e-06,
+      "loss": 4.0473,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.205607476635514e-06,
+      "loss": 4.0391,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.182242990654206e-06,
+      "loss": 4.2699,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.158878504672897e-06,
+      "loss": 3.934,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.135514018691588e-06,
+      "loss": 4.2621,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.11214953271028e-06,
+      "loss": 3.9878,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.088785046728972e-06,
+      "loss": 4.1561,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.065420560747663e-06,
+      "loss": 3.8834,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.042056074766355e-06,
+      "loss": 4.1451,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.018691588785047e-06,
+      "loss": 4.0874,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.995327102803738e-06,
+      "loss": 4.0917,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.971962616822429e-06,
+      "loss": 4.0869,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.948598130841121e-06,
+      "loss": 4.1209,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.925233644859813e-06,
+      "loss": 4.1191,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9018691588785045e-06,
+      "loss": 4.1615,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.878504672897196e-06,
+      "loss": 3.8686,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.855140186915888e-06,
+      "loss": 3.9337,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.831775700934579e-06,
+      "loss": 3.9501,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.808411214953271e-06,
+      "loss": 4.0796,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785046728971962e-06,
+      "loss": 4.2892,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.761682242990654e-06,
+      "loss": 4.1272,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.738317757009346e-06,
+      "loss": 3.9954,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7149532710280376e-06,
+      "loss": 4.2131,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.691588785046729e-06,
+      "loss": 3.9276,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6682242990654206e-06,
+      "loss": 4.1506,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6448598130841123e-06,
+      "loss": 4.1425,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6214953271028036e-06,
+      "loss": 4.1276,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5981308411214953e-06,
+      "loss": 4.0423,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.574766355140187e-06,
+      "loss": 4.1386,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5514018691588787e-06,
+      "loss": 3.9813,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.52803738317757e-06,
+      "loss": 4.0666,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5046728971962617e-06,
+      "loss": 4.1824,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4813084112149534e-06,
+      "loss": 3.9852,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.457943925233645e-06,
+      "loss": 4.1086,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4345794392523364e-06,
+      "loss": 4.2499,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.411214953271028e-06,
+      "loss": 4.1355,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3878504672897198e-06,
+      "loss": 4.0382,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.364485981308411e-06,
+      "loss": 4.1679,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3411214953271028e-06,
+      "loss": 4.3356,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3177570093457945e-06,
+      "loss": 4.5541,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.294392523364486e-06,
+      "loss": 3.9198,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2710280373831774e-06,
+      "loss": 4.5766,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247663551401869e-06,
+      "loss": 4.4485,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.224299065420561e-06,
+      "loss": 4.0286,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2009345794392525e-06,
+      "loss": 4.2244,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.177570093457944e-06,
+      "loss": 4.0821,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1542056074766355e-06,
+      "loss": 4.1241,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1308411214953272e-06,
+      "loss": 4.0387,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1074766355140185e-06,
+      "loss": 3.9334,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.08411214953271e-06,
+      "loss": 4.0468,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.060747663551402e-06,
+      "loss": 4.1902,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0373831775700936e-06,
+      "loss": 4.1277,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.014018691588785e-06,
+      "loss": 3.9706,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9906542056074766e-06,
+      "loss": 4.1782,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9672897196261683e-06,
+      "loss": 4.2156,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.94392523364486e-06,
+      "loss": 4.1851,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9205607476635513e-06,
+      "loss": 3.8209,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.897196261682243e-06,
+      "loss": 4.1538,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8738317757009347e-06,
+      "loss": 3.9055,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.850467289719626e-06,
+      "loss": 4.0566,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8271028037383177e-06,
+      "loss": 3.9957,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8037383177570094e-06,
+      "loss": 4.0295,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.780373831775701e-06,
+      "loss": 4.0519,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7570093457943923e-06,
+      "loss": 4.0107,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.733644859813084e-06,
+      "loss": 4.0358,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7102803738317757e-06,
+      "loss": 4.2164,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6869158878504674e-06,
+      "loss": 3.9759,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6635514018691587e-06,
+      "loss": 4.1425,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6401869158878504e-06,
+      "loss": 4.35,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.616822429906542e-06,
+      "loss": 4.0161,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5934579439252334e-06,
+      "loss": 3.8357,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.570093457943925e-06,
+      "loss": 4.0876,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.546728971962617e-06,
+      "loss": 4.2216,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5233644859813085e-06,
+      "loss": 4.2303,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4999999999999998e-06,
+      "loss": 3.7891,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4766355140186915e-06,
+      "loss": 3.8248,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.453271028037383e-06,
+      "loss": 4.3417,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.429906542056075e-06,
+      "loss": 3.8486,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.406542056074766e-06,
+      "loss": 4.1564,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.383177570093458e-06,
+      "loss": 4.1484,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3598130841121496e-06,
+      "loss": 3.9626,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.336448598130841e-06,
+      "loss": 4.2201,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3130841121495325e-06,
+      "loss": 4.1458,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2897196261682242e-06,
+      "loss": 3.9603,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.266355140186916e-06,
+      "loss": 4.3576,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2429906542056072e-06,
+      "loss": 4.1521,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.219626168224299e-06,
+      "loss": 4.4204,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1962616822429906e-06,
+      "loss": 4.4146,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1728971962616823e-06,
+      "loss": 4.4811,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1495327102803736e-06,
+      "loss": 4.92,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1261682242990653e-06,
+      "loss": 4.2099,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.102803738317757e-06,
+      "loss": 4.667,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0794392523364483e-06,
+      "loss": 4.542,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.05607476635514e-06,
+      "loss": 4.5096,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0327102803738317e-06,
+      "loss": 4.3746,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0093457943925234e-06,
+      "loss": 3.9448,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9859813084112147e-06,
+      "loss": 3.9043,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9626168224299064e-06,
+      "loss": 3.9582,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.939252336448598e-06,
+      "loss": 4.0408,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9158878504672894e-06,
+      "loss": 3.9079,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.892523364485981e-06,
+      "loss": 3.9643,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.869158878504673e-06,
+      "loss": 4.1182,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8457943925233645e-06,
+      "loss": 4.2909,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8224299065420562e-06,
+      "loss": 4.2164,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7990654205607477e-06,
+      "loss": 3.9166,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7757009345794394e-06,
+      "loss": 4.1746,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7523364485981308e-06,
+      "loss": 4.3629,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7289719626168225e-06,
+      "loss": 4.0523,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.705607476635514e-06,
+      "loss": 4.0632,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6822429906542055e-06,
+      "loss": 4.1973,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6588785046728972e-06,
+      "loss": 4.4761,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6355140186915887e-06,
+      "loss": 4.3851,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6121495327102804e-06,
+      "loss": 3.9143,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588785046728972e-06,
+      "loss": 3.9556,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5654205607476636e-06,
+      "loss": 4.0736,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.542056074766355e-06,
+      "loss": 4.2222,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5186915887850468e-06,
+      "loss": 4.2051,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4953271028037383e-06,
+      "loss": 4.1366,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.47196261682243e-06,
+      "loss": 4.0968,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4485981308411215e-06,
+      "loss": 4.0091,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.425233644859813e-06,
+      "loss": 4.1852,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4018691588785047e-06,
+      "loss": 4.281,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3785046728971962e-06,
+      "loss": 4.3431,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3551401869158879e-06,
+      "loss": 4.1475,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3317757009345794e-06,
+      "loss": 3.8348,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.308411214953271e-06,
+      "loss": 4.028,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2850467289719625e-06,
+      "loss": 4.0512,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2616822429906543e-06,
+      "loss": 4.2173,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2383177570093457e-06,
+      "loss": 4.511,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2149532710280374e-06,
+      "loss": 4.5334,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.191588785046729e-06,
+      "loss": 3.8144,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1682242990654204e-06,
+      "loss": 4.081,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1448598130841121e-06,
+      "loss": 4.0677,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1214953271028036e-06,
+      "loss": 3.9835,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0981308411214953e-06,
+      "loss": 4.1051,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0747663551401868e-06,
+      "loss": 4.1073,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0514018691588785e-06,
+      "loss": 3.923,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.02803738317757e-06,
+      "loss": 3.9201,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0046728971962617e-06,
+      "loss": 4.2432,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.813084112149532e-07,
+      "loss": 4.2978,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.579439252336447e-07,
+      "loss": 4.2097,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.345794392523365e-07,
+      "loss": 4.1209,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.112149532710281e-07,
+      "loss": 4.2323,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.878504672897197e-07,
+      "loss": 4.1248,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.644859813084113e-07,
+      "loss": 3.8854,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.411214953271028e-07,
+      "loss": 4.063,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.177570093457944e-07,
+      "loss": 4.2555,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.94392523364486e-07,
+      "loss": 4.0455,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.710280373831776e-07,
+      "loss": 4.2648,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.476635514018691e-07,
+      "loss": 3.9024,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.242990654205607e-07,
+      "loss": 3.9635,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.009345794392523e-07,
+      "loss": 4.0277,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.775700934579439e-07,
+      "loss": 4.0771,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.542056074766355e-07,
+      "loss": 3.9424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.308411214953271e-07,
+      "loss": 4.2301,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.074766355140187e-07,
+      "loss": 3.8629,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.841121495327102e-07,
+      "loss": 4.0201,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.607476635514018e-07,
+      "loss": 4.2464,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.373831775700934e-07,
+      "loss": 3.8469,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.14018691588785e-07,
+      "loss": 4.095,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.906542056074766e-07,
+      "loss": 3.9691,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6728971962616824e-07,
+      "loss": 4.0012,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4392523364485984e-07,
+      "loss": 4.0355,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.205607476635514e-07,
+      "loss": 3.9852,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.97196261682243e-07,
+      "loss": 4.049,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7383177570093457e-07,
+      "loss": 4.0473,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5046728971962617e-07,
+      "loss": 4.0632,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2710280373831776e-07,
+      "loss": 4.1366,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0373831775700936e-07,
+      "loss": 4.0087,
+      "step": 1775
+    },
     {
       "epoch": 1.0,
-      "learning_rate": 5.9e-06,
-      "loss": 4.6971,
-      "step": 296
+      "learning_rate": 2.803738317757009e-07,
+      "loss": 4.1291,
+      "step": 1776
     },
     {
       "epoch": 1.0,
-      "learning_rate": 5.92e-06,
-      "loss": 4.5581,
-      "step": 297
+      "learning_rate": 2.570093457943925e-07,
+      "loss": 4.0186,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3364485981308412e-07,
+      "loss": 4.1484,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.102803738317757e-07,
+      "loss": 4.377,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8691588785046729e-07,
+      "loss": 4.0732,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6355140186915888e-07,
+      "loss": 4.3938,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4018691588785045e-07,
+      "loss": 4.0189,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1682242990654206e-07,
+      "loss": 4.3766,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.345794392523364e-08,
+      "loss": 3.5853,
+      "step": 1784
     },
     {
       "epoch": 1.0,
-      "step": 297,
+      "step": 1784,
       "total_flos": 0.0,
-      "train_loss": 4.444549621556343,
-      "train_runtime": 4661.3109,
-      "train_samples_per_second": 6.122,
-      "train_steps_per_second": 0.064
+      "train_loss": 4.241005906208748,
+      "train_runtime": 8352.9418,
+      "train_samples_per_second": 3.417,
+      "train_steps_per_second": 0.214
     }
   ],
-  "max_steps": 297,
+  "max_steps": 1784,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,