diff --git "a/checkpoint-9268/trainer_state.json" "b/checkpoint-9268/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-9268/trainer_state.json"
@@ -0,0 +1,55757 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.938282261545101,
+  "eval_steps": 580,
+  "global_step": 9268,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-08,
+      "loss": 1.0668,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.2826168537139893,
+      "eval_runtime": 32.1462,
+      "eval_samples_per_second": 38.729,
+      "eval_steps_per_second": 19.38,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 0.8466,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-08,
+      "loss": 1.2213,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 0.8393,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 0.8577,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 1.2108,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1000000000000003e-07,
+      "loss": 1.1288,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 1.2298,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7e-07,
+      "loss": 0.9675,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.0311,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-07,
+      "loss": 1.1669,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 1.6216,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9e-07,
+      "loss": 0.9954,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 0.9629,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-07,
+      "loss": 1.351,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.439,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.100000000000001e-07,
+      "loss": 0.8192,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 1.1661,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7e-07,
+      "loss": 1.4198,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.2475,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3e-07,
+      "loss": 1.1388,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6e-07,
+      "loss": 1.1168,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.900000000000001e-07,
+      "loss": 1.1218,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 1.1156,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5e-07,
+      "loss": 1.1735,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.8e-07,
+      "loss": 0.8957,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.100000000000001e-07,
+      "loss": 1.2886,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 0.9394,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.699999999999999e-07,
+      "loss": 1.2861,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-07,
+      "loss": 0.9532,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.3e-07,
+      "loss": 1.1305,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.0239,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9e-07,
+      "loss": 1.142,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 1.5263,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.05e-06,
+      "loss": 0.8998,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.08e-06,
+      "loss": 1.0518,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.11e-06,
+      "loss": 1.3894,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.14e-06,
+      "loss": 0.9561,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.17e-06,
+      "loss": 0.966,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.9768,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2299999999999999e-06,
+      "loss": 0.8566,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.26e-06,
+      "loss": 1.0171,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.29e-06,
+      "loss": 1.0312,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.32e-06,
+      "loss": 0.8851,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.35e-06,
+      "loss": 0.7443,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.8973,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.41e-06,
+      "loss": 1.1784,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.44e-06,
+      "loss": 0.9239,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 1.0246,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 1.2476,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.53e-06,
+      "loss": 0.9587,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.56e-06,
+      "loss": 1.2841,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.59e-06,
+      "loss": 1.0869,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 1.1247,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.65e-06,
+      "loss": 1.1272,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 0.9138,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.71e-06,
+      "loss": 1.438,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7399999999999999e-06,
+      "loss": 1.3224,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.77e-06,
+      "loss": 0.9889,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-06,
+      "loss": 0.8523,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.83e-06,
+      "loss": 1.1939,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.86e-06,
+      "loss": 1.1811,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8900000000000001e-06,
+      "loss": 1.0024,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.1867,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.95e-06,
+      "loss": 0.7939,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.98e-06,
+      "loss": 1.3442,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0100000000000002e-06,
+      "loss": 0.8832,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 1.1674,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.07e-06,
+      "loss": 1.2114,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1e-06,
+      "loss": 1.106,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.13e-06,
+      "loss": 1.1039,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.16e-06,
+      "loss": 1.1667,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19e-06,
+      "loss": 0.7964,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.22e-06,
+      "loss": 1.0769,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.25e-06,
+      "loss": 0.9755,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.28e-06,
+      "loss": 1.1283,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.31e-06,
+      "loss": 1.0775,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.34e-06,
+      "loss": 0.8156,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.37e-06,
+      "loss": 1.2048,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.2663,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.43e-06,
+      "loss": 0.912,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4599999999999997e-06,
+      "loss": 1.2863,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.49e-06,
+      "loss": 1.2931,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.52e-06,
+      "loss": 1.2188,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.55e-06,
+      "loss": 1.2086,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.58e-06,
+      "loss": 1.0728,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.61e-06,
+      "loss": 1.1219,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.64e-06,
+      "loss": 0.83,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6700000000000003e-06,
+      "loss": 1.0,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7e-06,
+      "loss": 1.3198,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.73e-06,
+      "loss": 1.1578,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 1.5981,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7900000000000004e-06,
+      "loss": 0.9523,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.82e-06,
+      "loss": 1.2175,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.85e-06,
+      "loss": 1.4129,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.88e-06,
+      "loss": 1.3108,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.91e-06,
+      "loss": 0.9933,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.7392,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.97e-06,
+      "loss": 1.1855,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 1.1607,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9999999119332853e-06,
+      "loss": 1.4228,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9999996477331517e-06,
+      "loss": 1.1232,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99999920739963e-06,
+      "loss": 1.2884,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999998590932772e-06,
+      "loss": 1.0169,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999977983326505e-06,
+      "loss": 0.9785,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999996829599358e-06,
+      "loss": 1.0029,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999995684733008e-06,
+      "loss": 1.1143,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999943637337356e-06,
+      "loss": 1.0736,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999992866601696e-06,
+      "loss": 0.8849,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999911933370647e-06,
+      "loss": 0.8923,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999893439400377e-06,
+      "loss": 0.8946,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999873184108326e-06,
+      "loss": 0.8351,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999851167496876e-06,
+      "loss": 1.1443,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999827389568604e-06,
+      "loss": 0.9161,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999801850326316e-06,
+      "loss": 1.0299,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999774549772992e-06,
+      "loss": 1.2613,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999745487911856e-06,
+      "loss": 0.8642,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999971466474631e-06,
+      "loss": 1.0066,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999968208027997e-06,
+      "loss": 1.2261,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999647734516673e-06,
+      "loss": 1.0871,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999961162746044e-06,
+      "loss": 1.1891,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999573759115524e-06,
+      "loss": 1.0304,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999534129486363e-06,
+      "loss": 1.0613,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999949273857761e-06,
+      "loss": 1.3572,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999449586394126e-06,
+      "loss": 0.6936,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9999404672940985e-06,
+      "loss": 1.1993,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999935799822345e-06,
+      "loss": 1.2396,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999309562247012e-06,
+      "loss": 1.5032,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999259365017352e-06,
+      "loss": 1.2632,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999920740654036e-06,
+      "loss": 1.2663,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999915368682215e-06,
+      "loss": 1.4309,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999909820586902e-06,
+      "loss": 0.9828,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999040963687485e-06,
+      "loss": 1.0214,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998981960284274e-06,
+      "loss": 1.0738,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998921195666305e-06,
+      "loss": 0.9619,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999885866984072e-06,
+      "loss": 0.9129,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999879438281486e-06,
+      "loss": 1.2847,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999872833459627e-06,
+      "loss": 1.3372,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998660525192714e-06,
+      "loss": 0.8183,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998590954612148e-06,
+      "loss": 0.9985,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998519622862737e-06,
+      "loss": 0.9149,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998446529952864e-06,
+      "loss": 0.9572,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998371675891113e-06,
+      "loss": 0.7757,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998295060686263e-06,
+      "loss": 1.1083,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998216684347323e-06,
+      "loss": 0.854,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9998136546883488e-06,
+      "loss": 1.18,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999805464830417e-06,
+      "loss": 1.3177,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999797098861899e-06,
+      "loss": 1.4433,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999788556783777e-06,
+      "loss": 0.8144,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9997798385970528e-06,
+      "loss": 1.0203,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997709443027517e-06,
+      "loss": 1.2566,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999761873901918e-06,
+      "loss": 0.9373,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997526273956155e-06,
+      "loss": 1.0276,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997432047849313e-06,
+      "loss": 1.0681,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997336060709708e-06,
+      "loss": 0.8111,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997238312548616e-06,
+      "loss": 1.2236,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997138803377512e-06,
+      "loss": 0.8277,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999703753320809e-06,
+      "loss": 1.0342,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999693450205223e-06,
+      "loss": 1.1637,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999682970992203e-06,
+      "loss": 1.0025,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996723156829808e-06,
+      "loss": 0.9595,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996614842788063e-06,
+      "loss": 0.9179,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996504767809514e-06,
+      "loss": 1.0879,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996392931907092e-06,
+      "loss": 1.3426,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999627933509393e-06,
+      "loss": 0.8249,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996163977383363e-06,
+      "loss": 1.1791,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996046858788937e-06,
+      "loss": 1.2256,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99959279793244e-06,
+      "loss": 1.0899,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999580733900372e-06,
+      "loss": 1.1101,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995684937841056e-06,
+      "loss": 1.2585,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999556077585078e-06,
+      "loss": 1.2754,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995434853047484e-06,
+      "loss": 0.7603,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999530716944594e-06,
+      "loss": 1.0474,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995177725061144e-06,
+      "loss": 0.7295,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995046519908298e-06,
+      "loss": 1.0647,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994913554002806e-06,
+      "loss": 0.9525,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994778827360287e-06,
+      "loss": 1.2188,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999464233999655e-06,
+      "loss": 1.3339,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994504091927633e-06,
+      "loss": 1.2849,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999436408316976e-06,
+      "loss": 1.2665,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994222313739387e-06,
+      "loss": 1.0031,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999407878365314e-06,
+      "loss": 0.991,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993933492927883e-06,
+      "loss": 0.99,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999378644158068e-06,
+      "loss": 0.9756,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993637629628788e-06,
+      "loss": 1.213,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993487057089693e-06,
+      "loss": 1.1675,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993334723981064e-06,
+      "loss": 1.0255,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993180630320795e-06,
+      "loss": 0.6903,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993024776126977e-06,
+      "loss": 1.157,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992867161417915e-06,
+      "loss": 1.027,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992707786212113e-06,
+      "loss": 0.8395,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999254665052829e-06,
+      "loss": 0.8944,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992383754385356e-06,
+      "loss": 0.6681,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999221909780245e-06,
+      "loss": 0.8785,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.99920526807989e-06,
+      "loss": 1.0981,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991884503394254e-06,
+      "loss": 1.3946,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.999171456560825e-06,
+      "loss": 1.0832,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9991542867460844e-06,
+      "loss": 0.8644,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.999136940897221e-06,
+      "loss": 1.0757,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99911941901627e-06,
+      "loss": 1.2222,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.999101721105289e-06,
+      "loss": 1.1069,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9990838471663574e-06,
+      "loss": 1.271,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9990657972015732e-06,
+      "loss": 1.1228,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9990475712130558e-06,
+      "loss": 1.246,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.999029169202945e-06,
+      "loss": 1.0935,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9990105911734025e-06,
+      "loss": 0.8214,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998991837126609e-06,
+      "loss": 1.3406,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998972907064767e-06,
+      "loss": 0.6968,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9989538009900995e-06,
+      "loss": 0.9547,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9989345189048496e-06,
+      "loss": 1.1176,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9989150608112816e-06,
+      "loss": 1.1492,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99889542671168e-06,
+      "loss": 1.3651,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998875616608351e-06,
+      "loss": 0.9929,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9988556305036204e-06,
+      "loss": 0.9426,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9988354683998346e-06,
+      "loss": 1.3901,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998815130299362e-06,
+      "loss": 1.1086,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987946162045893e-06,
+      "loss": 1.0862,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998773926117927e-06,
+      "loss": 0.8931,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998753060041803e-06,
+      "loss": 1.2757,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987320179786684e-06,
+      "loss": 1.0679,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9987107999309943e-06,
+      "loss": 1.3329,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998689405901271e-06,
+      "loss": 1.1216,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9986678358920118e-06,
+      "loss": 1.1358,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998646089905749e-06,
+      "loss": 0.9812,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998624167945036e-06,
+      "loss": 0.9288,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9986020700124472e-06,
+      "loss": 1.2316,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9985797961105767e-06,
+      "loss": 1.2547,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9985573462420406e-06,
+      "loss": 1.361,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998534720409475e-06,
+      "loss": 0.9254,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9985119186155367e-06,
+      "loss": 0.7791,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998488940862903e-06,
+      "loss": 0.8094,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9984657871542716e-06,
+      "loss": 1.2043,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9984424574923617e-06,
+      "loss": 0.705,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998418951879913e-06,
+      "loss": 1.0795,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998395270319685e-06,
+      "loss": 1.1087,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9983714128144586e-06,
+      "loss": 1.1531,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9983473793670352e-06,
+      "loss": 1.0689,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998323169980237e-06,
+      "loss": 1.001,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998298784656907e-06,
+      "loss": 1.291,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982742233999083e-06,
+      "loss": 1.0116,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982494862121242e-06,
+      "loss": 1.3948,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998224573096461e-06,
+      "loss": 1.3862,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981994840558425e-06,
+      "loss": 1.0777,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9981742190932157e-06,
+      "loss": 0.8814,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9981487782115467e-06,
+      "loss": 1.1321,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9981231614138232e-06,
+      "loss": 1.0211,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.998097368703053e-06,
+      "loss": 1.0228,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.998071400082265e-06,
+      "loss": 1.3191,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.998045255554508e-06,
+      "loss": 1.0901,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.998018935122852e-06,
+      "loss": 0.6744,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9979924387903883e-06,
+      "loss": 0.9755,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9979657665602272e-06,
+      "loss": 0.8063,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9979389184355016e-06,
+      "loss": 1.2921,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997911894419363e-06,
+      "loss": 0.9761,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978846945149857e-06,
+      "loss": 0.5217,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997857318725563e-06,
+      "loss": 0.7982,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978297670543096e-06,
+      "loss": 1.0839,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978020395044604e-06,
+      "loss": 1.1431,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977741360792714e-06,
+      "loss": 1.1052,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997746056782019e-06,
+      "loss": 1.0418,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977178016160003e-06,
+      "loss": 0.887,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976893705845335e-06,
+      "loss": 0.9713,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976607636909567e-06,
+      "loss": 0.7808,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997631980938629e-06,
+      "loss": 0.9612,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.99760302233093e-06,
+      "loss": 0.4437,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9975738878712604e-06,
+      "loss": 0.7912,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9975445775630413e-06,
+      "loss": 1.0223,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997515091409714e-06,
+      "loss": 1.4806,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997485429414741e-06,
+      "loss": 1.1292,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997455591581605e-06,
+      "loss": 1.0272,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974255779138103e-06,
+      "loss": 0.9612,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973953884148806e-06,
+      "loss": 0.9025,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973650230883614e-06,
+      "loss": 1.1176,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973344819378176e-06,
+      "loss": 0.9855,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973037649668354e-06,
+      "loss": 1.352,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997272872179022e-06,
+      "loss": 1.1453,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997241803578005e-06,
+      "loss": 1.059,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972105591674327e-06,
+      "loss": 1.3455,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971791389509734e-06,
+      "loss": 0.7754,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997147542932317e-06,
+      "loss": 1.1969,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997115771115172e-06,
+      "loss": 1.0207,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970838235032718e-06,
+      "loss": 0.9992,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997051700100366e-06,
+      "loss": 1.0629,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970194009102266e-06,
+      "loss": 1.3236,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.996986925936647e-06,
+      "loss": 0.6324,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99695427518344e-06,
+      "loss": 0.7605,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9969214486544396e-06,
+      "loss": 1.0257,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9968884463535e-06,
+      "loss": 1.1605,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9968552682844976e-06,
+      "loss": 1.3859,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968219144513274e-06,
+      "loss": 0.7384,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996788384857905e-06,
+      "loss": 1.6353,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967546795081692e-06,
+      "loss": 1.2559,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967207984060772e-06,
+      "loss": 1.0661,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966867415556067e-06,
+      "loss": 1.1527,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966525089607574e-06,
+      "loss": 0.8175,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996618100625549e-06,
+      "loss": 1.3451,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996583516554021e-06,
+      "loss": 0.8253,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9965487567502356e-06,
+      "loss": 1.0656,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9965138212182737e-06,
+      "loss": 1.2254,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9964787099622372e-06,
+      "loss": 1.6709,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9964434229862496e-06,
+      "loss": 0.8667,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996407960294454e-06,
+      "loss": 1.1178,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9963723218910144e-06,
+      "loss": 0.8534,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9963365077801156e-06,
+      "loss": 0.945,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9963005179659637e-06,
+      "loss": 1.0545,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996264352452784e-06,
+      "loss": 0.9398,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9962280112448225e-06,
+      "loss": 1.1415,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9961914943463476e-06,
+      "loss": 0.9068,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996154801761647e-06,
+      "loss": 1.2247,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9961179334950286e-06,
+      "loss": 0.9266,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9960808895508226e-06,
+      "loss": 1.0856,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9960436699333775e-06,
+      "loss": 0.9896,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996006274647065e-06,
+      "loss": 1.0751,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9959687036962753e-06,
+      "loss": 1.2232,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.99593095708542e-06,
+      "loss": 1.0885,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995893034818932e-06,
+      "loss": 0.7979,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995854936901264e-06,
+      "loss": 0.8242,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995816663336889e-06,
+      "loss": 1.2643,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9957782141303018e-06,
+      "loss": 1.0537,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9957395892860175e-06,
+      "loss": 1.1472,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9957007888085703e-06,
+      "loss": 0.8924,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995661812702517e-06,
+      "loss": 1.2104,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9956226609724343e-06,
+      "loss": 0.9627,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.99558333362292e-06,
+      "loss": 0.9223,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9955438306585913e-06,
+      "loss": 0.8288,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9955041520840864e-06,
+      "loss": 1.0866,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995464297904065e-06,
+      "loss": 0.7675,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995424268123207e-06,
+      "loss": 1.0087,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995384062746212e-06,
+      "loss": 1.1538,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9953436817778023e-06,
+      "loss": 1.0517,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9953031252227186e-06,
+      "loss": 0.9318,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9952623930857233e-06,
+      "loss": 1.0032,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9952214853715993e-06,
+      "loss": 0.9167,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9951804020851497e-06,
+      "loss": 1.0245,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.995139143231199e-06,
+      "loss": 1.185,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995097708814592e-06,
+      "loss": 0.8543,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9950560988401945e-06,
+      "loss": 0.8485,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995014313312891e-06,
+      "loss": 1.1215,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994972352237589e-06,
+      "loss": 0.8368,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994930215619216e-06,
+      "loss": 1.2811,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9948879034627195e-06,
+      "loss": 1.2706,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994845415773067e-06,
+      "loss": 1.3247,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9948027525552487e-06,
+      "loss": 1.5791,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994759913814274e-06,
+      "loss": 1.1183,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994716899555172e-06,
+      "loss": 0.9277,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994673709782995e-06,
+      "loss": 1.1398,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9946303445028136e-06,
+      "loss": 0.9636,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9945868037197203e-06,
+      "loss": 1.0631,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9945430874388274e-06,
+      "loss": 1.0427,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9944991956652686e-06,
+      "loss": 0.9942,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994455128404197e-06,
+      "loss": 1.0084,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9944108856607874e-06,
+      "loss": 0.9839,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9943664674402357e-06,
+      "loss": 1.1796,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9943218737477564e-06,
+      "loss": 1.3111,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994277104588587e-06,
+      "loss": 0.9406,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9942321599679833e-06,
+      "loss": 0.9456,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9941870398912232e-06,
+      "loss": 1.0495,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9941417443636046e-06,
+      "loss": 1.033,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.994096273390447e-06,
+      "loss": 1.0279,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9940506269770886e-06,
+      "loss": 0.8449,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.99400480512889e-06,
+      "loss": 1.0743,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.993958807851232e-06,
+      "loss": 1.1996,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.993912635149515e-06,
+      "loss": 1.2913,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9938662870291604e-06,
+      "loss": 1.0936,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9938197634956115e-06,
+      "loss": 0.9745,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9937730645543307e-06,
+      "loss": 1.0071,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9937261902108017e-06,
+      "loss": 1.1672,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9936791404705285e-06,
+      "loss": 1.7294,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9936319153390354e-06,
+      "loss": 1.1222,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9935845148218684e-06,
+      "loss": 1.0851,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9935369389245925e-06,
+      "loss": 1.1429,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9934891876527952e-06,
+      "loss": 1.5181,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9934412610120826e-06,
+      "loss": 1.3844,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.993393159008083e-06,
+      "loss": 1.4567,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9933448816464444e-06,
+      "loss": 1.095,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9932964289328356e-06,
+      "loss": 1.1531,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9932478008729464e-06,
+      "loss": 0.9975,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.993198997472486e-06,
+      "loss": 1.2656,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.993150018737186e-06,
+      "loss": 1.4983,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.993100864672797e-06,
+      "loss": 1.125,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.993051535285091e-06,
+      "loss": 1.0912,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9930020305798596e-06,
+      "loss": 0.9772,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9929523505629175e-06,
+      "loss": 0.8309,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9929024952400966e-06,
+      "loss": 1.1034,
+      "step": 384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9928524646172517e-06,
+      "loss": 1.1266,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.992802258700257e-06,
+      "loss": 0.7076,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.992751877495009e-06,
+      "loss": 1.0077,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.992701321007422e-06,
+      "loss": 1.0197,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9926505892434334e-06,
+      "loss": 1.1103,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9925996822090006e-06,
+      "loss": 0.9896,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9925485999101e-06,
+      "loss": 1.1594,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.992497342352731e-06,
+      "loss": 0.926,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9924459095429116e-06,
+      "loss": 1.168,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9923943014866813e-06,
+      "loss": 0.5834,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9923425181901004e-06,
+      "loss": 1.0797,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.992290559659249e-06,
+      "loss": 0.8887,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9922384259002286e-06,
+      "loss": 0.8068,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9921861169191607e-06,
+      "loss": 0.9613,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9921336327221876e-06,
+      "loss": 1.1375,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9920809733154715e-06,
+      "loss": 1.0678,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9920281387051965e-06,
+      "loss": 0.7456,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9919751288975664e-06,
+      "loss": 0.9705,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9919219438988055e-06,
+      "loss": 1.1648,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9918685837151594e-06,
+      "loss": 1.0508,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9918150483528932e-06,
+      "loss": 0.6299,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9917613378182937e-06,
+      "loss": 1.1415,
+      "step": 406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9917074521176664e-06,
+      "loss": 0.9854,
+      "step": 407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.991653391257341e-06,
+      "loss": 1.0443,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9915991552436633e-06,
+      "loss": 1.0263,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.991544744083003e-06,
+      "loss": 1.0909,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9914901577817486e-06,
+      "loss": 1.2664,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.99143539634631e-06,
+      "loss": 1.266,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9913804597831178e-06,
+      "loss": 1.428,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9913253480986223e-06,
+      "loss": 1.2817,
+      "step": 414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9912700612992946e-06,
+      "loss": 0.9364,
+      "step": 415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9912145993916273e-06,
+      "loss": 0.9314,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.991158962382132e-06,
+      "loss": 1.0493,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9911031502773425e-06,
+      "loss": 1.1937,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.991047163083812e-06,
+      "loss": 0.8765,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.990991000808115e-06,
+      "loss": 0.9615,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.990934663456846e-06,
+      "loss": 0.9097,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.99087815103662e-06,
+      "loss": 1.0979,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9908214635540725e-06,
+      "loss": 1.1106,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.990764601015861e-06,
+      "loss": 1.2386,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.990707563428662e-06,
+      "loss": 1.1282,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9906503507991723e-06,
+      "loss": 1.0755,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9905929631341105e-06,
+      "loss": 1.1004,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9905354004402154e-06,
+      "loss": 1.0042,
+      "step": 428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.990477662724246e-06,
+      "loss": 1.0024,
+      "step": 429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9904197499929815e-06,
+      "loss": 1.0323,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9903616622532224e-06,
+      "loss": 1.27,
+      "step": 431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9903033995117897e-06,
+      "loss": 1.2566,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9902449617755253e-06,
+      "loss": 0.7985,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9901863490512897e-06,
+      "loss": 1.1392,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9901275613459667e-06,
+      "loss": 1.2173,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.990068598666458e-06,
+      "loss": 0.9578,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.990009461019688e-06,
+      "loss": 1.001,
+      "step": 437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.989950148412601e-06,
+      "loss": 0.6646,
+      "step": 438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.989890660852161e-06,
+      "loss": 0.8681,
+      "step": 439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9898309983453533e-06,
+      "loss": 0.9956,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9897711608991836e-06,
+      "loss": 1.3169,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9897111485206785e-06,
+      "loss": 0.9874,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9896509612168843e-06,
+      "loss": 0.781,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9895905989948687e-06,
+      "loss": 1.0075,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9895300618617192e-06,
+      "loss": 1.4006,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9894693498245447e-06,
+      "loss": 0.9516,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9894084628904736e-06,
+      "loss": 1.1393,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9893474010666558e-06,
+      "loss": 1.1598,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9892861643602614e-06,
+      "loss": 0.7951,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.98922475277848e-06,
+      "loss": 0.8049,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9891631663285244e-06,
+      "loss": 1.1294,
+      "step": 451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.989101405017625e-06,
+      "loss": 0.698,
+      "step": 452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.989039468853034e-06,
+      "loss": 0.681,
+      "step": 453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.988977357842024e-06,
+      "loss": 0.9209,
+      "step": 454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9889150719918887e-06,
+      "loss": 1.2015,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.988852611309942e-06,
+      "loss": 1.1002,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9887899758035173e-06,
+      "loss": 0.9994,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.98872716547997e-06,
+      "loss": 0.8622,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9886641803466757e-06,
+      "loss": 1.1229,
+      "step": 459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.98860102041103e-06,
+      "loss": 1.1371,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9885376856804487e-06,
+      "loss": 0.8039,
+      "step": 461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9884741761623696e-06,
+      "loss": 0.959,
+      "step": 462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.98841049186425e-06,
+      "loss": 1.2513,
+      "step": 463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.988346632793567e-06,
+      "loss": 1.0403,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.98828259895782e-06,
+      "loss": 1.3199,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9882183903645274e-06,
+      "loss": 0.9289,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9881540070212294e-06,
+      "loss": 0.7974,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9880894489354853e-06,
+      "loss": 0.8132,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.988024716114876e-06,
+      "loss": 1.3109,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.987959808567003e-06,
+      "loss": 1.0003,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.987894726299487e-06,
+      "loss": 0.7431,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9878294693199706e-06,
+      "loss": 0.9142,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9877640376361165e-06,
+      "loss": 0.8275,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9876984312556076e-06,
+      "loss": 0.9978,
+      "step": 474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9876326501861483e-06,
+      "loss": 0.9026,
+      "step": 475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.987566694435462e-06,
+      "loss": 0.9614,
+      "step": 476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9875005640112928e-06,
+      "loss": 1.1506,
+      "step": 477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.987434258921407e-06,
+      "loss": 1.0007,
+      "step": 478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9873677791735898e-06,
+      "loss": 0.9798,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.987301124775648e-06,
+      "loss": 0.8309,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9872342957354075e-06,
+      "loss": 0.7859,
+      "step": 481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9871672920607156e-06,
+      "loss": 1.6491,
+      "step": 482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.987100113759441e-06,
+      "loss": 1.0152,
+      "step": 483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.98703276083947e-06,
+      "loss": 1.2719,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.986965233308714e-06,
+      "loss": 1.0153,
+      "step": 485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9868975311750997e-06,
+      "loss": 1.4362,
+      "step": 486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.986829654446578e-06,
+      "loss": 1.3895,
+      "step": 487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.986761603131119e-06,
+      "loss": 1.1723,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9866933772367137e-06,
+      "loss": 0.9854,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9866249767713733e-06,
+      "loss": 1.4223,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.986556401743129e-06,
+      "loss": 0.7006,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9864876521600335e-06,
+      "loss": 1.3342,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9864187280301595e-06,
+      "loss": 1.1225,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9863496293616005e-06,
+      "loss": 0.8446,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.98628035616247e-06,
+      "loss": 1.0535,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9862109084409015e-06,
+      "loss": 1.1367,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9861412862050503e-06,
+      "loss": 0.9806,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.986071489463092e-06,
+      "loss": 1.1439,
+      "step": 498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9860015182232212e-06,
+      "loss": 1.1149,
+      "step": 499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.985931372493656e-06,
+      "loss": 1.1547,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.985861052282631e-06,
+      "loss": 0.8806,
+      "step": 501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9857905575984044e-06,
+      "loss": 1.216,
+      "step": 502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9857198884492538e-06,
+      "loss": 0.8075,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.985649044843477e-06,
+      "loss": 0.6957,
+      "step": 504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.985578026789393e-06,
+      "loss": 1.0724,
+      "step": 505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9855068342953405e-06,
+      "loss": 1.1164,
+      "step": 506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9854354673696797e-06,
+      "loss": 0.8721,
+      "step": 507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.98536392602079e-06,
+      "loss": 1.1491,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9852922102570723e-06,
+      "loss": 0.9626,
+      "step": 509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9852203200869473e-06,
+      "loss": 1.053,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.985148255518857e-06,
+      "loss": 1.067,
+      "step": 511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9850760165612633e-06,
+      "loss": 1.2675,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9850036032226484e-06,
+      "loss": 1.1246,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9849310155115155e-06,
+      "loss": 1.1673,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9848582534363872e-06,
+      "loss": 1.5118,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9847853170058087e-06,
+      "loss": 0.9582,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.984712206228344e-06,
+      "loss": 0.9707,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.984638921112577e-06,
+      "loss": 1.2606,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9845654616671136e-06,
+      "loss": 1.2437,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.98449182790058e-06,
+      "loss": 1.2143,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.984418019821622e-06,
+      "loss": 0.7296,
+      "step": 521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9843440374389066e-06,
+      "loss": 0.8619,
+      "step": 522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9842698807611205e-06,
+      "loss": 1.2762,
+      "step": 523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9841955497969716e-06,
+      "loss": 0.8513,
+      "step": 524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9841210445551882e-06,
+      "loss": 0.8414,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.984046365044519e-06,
+      "loss": 0.8397,
+      "step": 526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.983971511273732e-06,
+      "loss": 1.1441,
+      "step": 527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.983896483251618e-06,
+      "loss": 0.9721,
+      "step": 528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9838212809869863e-06,
+      "loss": 1.1894,
+      "step": 529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9837459044886673e-06,
+      "loss": 0.861,
+      "step": 530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.983670353765512e-06,
+      "loss": 1.4871,
+      "step": 531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.983594628826392e-06,
+      "loss": 1.0617,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9835187296801988e-06,
+      "loss": 0.8969,
+      "step": 533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9834426563358446e-06,
+      "loss": 0.8227,
+      "step": 534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9833664088022624e-06,
+      "loss": 1.1846,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.983289987088405e-06,
+      "loss": 1.0767,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.983213391203246e-06,
+      "loss": 0.9383,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.98313662115578e-06,
+      "loss": 1.2809,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.983059676955021e-06,
+      "loss": 0.6862,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9829825586100036e-06,
+      "loss": 0.5148,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9829052661297846e-06,
+      "loss": 1.007,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9828277995234386e-06,
+      "loss": 0.877,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.982750158800062e-06,
+      "loss": 0.9812,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9826723439687723e-06,
+      "loss": 0.9038,
+      "step": 544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9825943550387064e-06,
+      "loss": 1.0283,
+      "step": 545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.982516192019021e-06,
+      "loss": 1.1665,
+      "step": 546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.982437854918896e-06,
+      "loss": 1.0829,
+      "step": 547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.982359343747528e-06,
+      "loss": 1.1128,
+      "step": 548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9822806585141373e-06,
+      "loss": 0.8643,
+      "step": 549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9822017992279627e-06,
+      "loss": 0.7539,
+      "step": 550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9821227658982646e-06,
+      "loss": 1.2795,
+      "step": 551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.982043558534322e-06,
+      "loss": 0.7064,
+      "step": 552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.981964177145437e-06,
+      "loss": 1.2372,
+      "step": 553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9818846217409304e-06,
+      "loss": 0.9669,
+      "step": 554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9818048923301433e-06,
+      "loss": 0.9619,
+      "step": 555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.981724988922438e-06,
+      "loss": 1.2044,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9816449115271975e-06,
+      "loss": 0.9373,
+      "step": 557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9815646601538234e-06,
+      "loss": 1.1138,
+      "step": 558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.98148423481174e-06,
+      "loss": 1.2086,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9814036355103904e-06,
+      "loss": 1.3177,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9813228622592394e-06,
+      "loss": 1.1477,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9812419150677716e-06,
+      "loss": 0.7987,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.981160793945491e-06,
+      "loss": 0.8819,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.981079498901924e-06,
+      "loss": 0.9212,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9809980299466164e-06,
+      "loss": 1.0855,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9809163870891337e-06,
+      "loss": 0.9349,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9808345703390636e-06,
+      "loss": 1.2243,
+      "step": 567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9807525797060125e-06,
+      "loss": 0.7934,
+      "step": 568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.980670415199608e-06,
+      "loss": 0.5616,
+      "step": 569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9805880768294988e-06,
+      "loss": 0.8519,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.980505564605352e-06,
+      "loss": 0.985,
+      "step": 571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.980422878536857e-06,
+      "loss": 1.0743,
+      "step": 572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9803400186337236e-06,
+      "loss": 0.7903,
+      "step": 573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.980256984905681e-06,
+      "loss": 1.1292,
+      "step": 574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9801737773624784e-06,
+      "loss": 1.0639,
+      "step": 575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.980090396013887e-06,
+      "loss": 0.8221,
+      "step": 576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9800068408696975e-06,
+      "loss": 0.8653,
+      "step": 577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.979923111939721e-06,
+      "loss": 1.1338,
+      "step": 578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9798392092337892e-06,
+      "loss": 1.0976,
+      "step": 579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9797551327617546e-06,
+      "loss": 0.9408,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.2613449096679688,
+      "eval_runtime": 32.8425,
+      "eval_samples_per_second": 37.908,
+      "eval_steps_per_second": 18.969,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.979670882533489e-06,
+      "loss": 1.1245,
+      "step": 581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9795864585588855e-06,
+      "loss": 1.0534,
+      "step": 582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9795018608478575e-06,
+      "loss": 0.7156,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9794170894103386e-06,
+      "loss": 1.0678,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9793321442562822e-06,
+      "loss": 1.3353,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.979247025395664e-06,
+      "loss": 1.1273,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.979161732838478e-06,
+      "loss": 0.7786,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.979076266594739e-06,
+      "loss": 0.6931,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.978990626674484e-06,
+      "loss": 1.0972,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9789048130877682e-06,
+      "loss": 1.1221,
+      "step": 590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.978818825844668e-06,
+      "loss": 0.9518,
+      "step": 591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9787326649552802e-06,
+      "loss": 0.9277,
+      "step": 592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.978646330429723e-06,
+      "loss": 1.1388,
+      "step": 593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9785598222781325e-06,
+      "loss": 1.014,
+      "step": 594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.978473140510667e-06,
+      "loss": 0.8841,
+      "step": 595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.978386285137506e-06,
+      "loss": 1.055,
+      "step": 596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9782992561688473e-06,
+      "loss": 1.0099,
+      "step": 597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9782120536149096e-06,
+      "loss": 0.8939,
+      "step": 598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9781246774859334e-06,
+      "loss": 0.7555,
+      "step": 599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9780371277921785e-06,
+      "loss": 0.9133,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9779494045439243e-06,
+      "loss": 1.1065,
+      "step": 601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9778615077514725e-06,
+      "loss": 0.9152,
+      "step": 602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9777734374251442e-06,
+      "loss": 1.0829,
+      "step": 603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9776851935752796e-06,
+      "loss": 0.9682,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9775967762122417e-06,
+      "loss": 1.0187,
+      "step": 605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9775081853464117e-06,
+      "loss": 1.1062,
+      "step": 606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9774194209881933e-06,
+      "loss": 1.0507,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9773304831480084e-06,
+      "loss": 1.1243,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.977241371836301e-06,
+      "loss": 0.62,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.977152087063534e-06,
+      "loss": 0.7882,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9770626288401917e-06,
+      "loss": 0.6661,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.976972997176779e-06,
+      "loss": 0.9701,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9768831920838205e-06,
+      "loss": 1.047,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9767932135718606e-06,
+      "loss": 0.8731,
+      "step": 614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9767030616514655e-06,
+      "loss": 0.561,
+      "step": 615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9766127363332207e-06,
+      "loss": 1.0698,
+      "step": 616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9765222376277326e-06,
+      "loss": 0.8451,
+      "step": 617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9764315655456273e-06,
+      "loss": 1.1086,
+      "step": 618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9763407200975525e-06,
+      "loss": 1.1081,
+      "step": 619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9762497012941756e-06,
+      "loss": 0.9823,
+      "step": 620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.976158509146183e-06,
+      "loss": 1.2448,
+      "step": 621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.976067143664284e-06,
+      "loss": 0.9679,
+      "step": 622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.975975604859206e-06,
+      "loss": 0.844,
+      "step": 623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9758838927416974e-06,
+      "loss": 1.106,
+      "step": 624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.975792007322529e-06,
+      "loss": 0.9266,
+      "step": 625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9756999486124886e-06,
+      "loss": 1.0137,
+      "step": 626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9756077166223867e-06,
+      "loss": 0.7806,
+      "step": 627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.975515311363053e-06,
+      "loss": 0.6932,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.975422732845338e-06,
+      "loss": 1.0675,
+      "step": 629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9753299810801127e-06,
+      "loss": 1.0283,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.975237056078268e-06,
+      "loss": 1.0374,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9751439578507156e-06,
+      "loss": 1.2975,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9750506864083865e-06,
+      "loss": 0.7285,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9749572417622345e-06,
+      "loss": 0.9599,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9748636239232303e-06,
+      "loss": 1.1432,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.974769832902368e-06,
+      "loss": 0.8334,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.97467586871066e-06,
+      "loss": 1.0316,
+      "step": 637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.97458173135914e-06,
+      "loss": 1.2535,
+      "step": 638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9744874208588624e-06,
+      "loss": 1.0193,
+      "step": 639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9743929372209e-06,
+      "loss": 1.0466,
+      "step": 640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.974298280456349e-06,
+      "loss": 1.1382,
+      "step": 641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9742034505763228e-06,
+      "loss": 1.1603,
+      "step": 642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.974108447591957e-06,
+      "loss": 1.077,
+      "step": 643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.974013271514407e-06,
+      "loss": 1.2367,
+      "step": 644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9739179223548492e-06,
+      "loss": 1.1627,
+      "step": 645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.973822400124479e-06,
+      "loss": 1.0716,
+      "step": 646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9737267048345133e-06,
+      "loss": 0.9375,
+      "step": 647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9736308364961883e-06,
+      "loss": 0.8722,
+      "step": 648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9735347951207616e-06,
+      "loss": 1.0777,
+      "step": 649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.973438580719511e-06,
+      "loss": 0.9632,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9733421933037325e-06,
+      "loss": 1.1566,
+      "step": 651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.973245632884746e-06,
+      "loss": 1.1901,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.973148899473889e-06,
+      "loss": 1.1624,
+      "step": 653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9730519930825203e-06,
+      "loss": 1.0595,
+      "step": 654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9729549137220185e-06,
+      "loss": 0.9972,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9728576614037836e-06,
+      "loss": 1.0206,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9727602361392348e-06,
+      "loss": 1.2351,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9726626379398117e-06,
+      "loss": 0.8946,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.972564866816975e-06,
+      "loss": 1.1029,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.972466922782205e-06,
+      "loss": 1.2262,
+      "step": 660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.972368805847003e-06,
+      "loss": 1.1586,
+      "step": 661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.972270516022889e-06,
+      "loss": 0.9681,
+      "step": 662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9721720533214052e-06,
+      "loss": 1.0917,
+      "step": 663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9720734177541133e-06,
+      "loss": 1.2183,
+      "step": 664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9719746093325953e-06,
+      "loss": 0.9879,
+      "step": 665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9718756280684534e-06,
+      "loss": 0.9093,
+      "step": 666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9717764739733096e-06,
+      "loss": 1.0094,
+      "step": 667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9716771470588083e-06,
+      "loss": 1.0687,
+      "step": 668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9715776473366113e-06,
+      "loss": 1.0676,
+      "step": 669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9714779748184026e-06,
+      "loss": 0.8461,
+      "step": 670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.971378129515886e-06,
+      "loss": 0.6711,
+      "step": 671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.971278111440786e-06,
+      "loss": 1.3177,
+      "step": 672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.971177920604846e-06,
+      "loss": 0.8529,
+      "step": 673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9710775570198308e-06,
+      "loss": 1.3491,
+      "step": 674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9709770206975262e-06,
+      "loss": 1.0669,
+      "step": 675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9708763116497365e-06,
+      "loss": 0.8623,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.970775429888288e-06,
+      "loss": 1.0946,
+      "step": 677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9706743754250256e-06,
+      "loss": 0.8058,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.970573148271816e-06,
+      "loss": 1.1618,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.970471748440545e-06,
+      "loss": 0.9758,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9703701759431194e-06,
+      "loss": 0.6717,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.970268430791466e-06,
+      "loss": 1.1163,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9701665129975325e-06,
+      "loss": 1.2558,
+      "step": 683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.970064422573286e-06,
+      "loss": 1.31,
+      "step": 684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.969962159530714e-06,
+      "loss": 0.9986,
+      "step": 685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9698597238818242e-06,
+      "loss": 1.0617,
+      "step": 686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9697571156386454e-06,
+      "loss": 0.783,
+      "step": 687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.969654334813226e-06,
+      "loss": 1.2464,
+      "step": 688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9695513814176347e-06,
+      "loss": 1.1797,
+      "step": 689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9694482554639603e-06,
+      "loss": 0.8208,
+      "step": 690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.969344956964312e-06,
+      "loss": 1.2101,
+      "step": 691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.96924148593082e-06,
+      "loss": 0.9559,
+      "step": 692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.969137842375634e-06,
+      "loss": 1.4473,
+      "step": 693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.969034026310923e-06,
+      "loss": 1.381,
+      "step": 694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9689300377488784e-06,
+      "loss": 1.0125,
+      "step": 695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.968825876701711e-06,
+      "loss": 0.8278,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.96872154318165e-06,
+      "loss": 1.0497,
+      "step": 697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.968617037200948e-06,
+      "loss": 1.0431,
+      "step": 698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9685123587718767e-06,
+      "loss": 0.9797,
+      "step": 699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.968407507906726e-06,
+      "loss": 0.6056,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.968302484617809e-06,
+      "loss": 1.0324,
+      "step": 701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.968197288917457e-06,
+      "loss": 0.9439,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9680919208180233e-06,
+      "loss": 1.1148,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9679863803318793e-06,
+      "loss": 1.1658,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9678806674714186e-06,
+      "loss": 0.8455,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9677747822490544e-06,
+      "loss": 1.0937,
+      "step": 706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9676687246772185e-06,
+      "loss": 1.0732,
+      "step": 707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9675624947683667e-06,
+      "loss": 1.1065,
+      "step": 708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.967456092534971e-06,
+      "loss": 0.9994,
+      "step": 709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9673495179895262e-06,
+      "loss": 1.0244,
+      "step": 710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.967242771144546e-06,
+      "loss": 1.0501,
+      "step": 711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9671358520125653e-06,
+      "loss": 1.0324,
+      "step": 712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.967028760606139e-06,
+      "loss": 0.7647,
+      "step": 713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.966921496937841e-06,
+      "loss": 1.0331,
+      "step": 714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.966814061020268e-06,
+      "loss": 1.25,
+      "step": 715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.966706452866034e-06,
+      "loss": 1.1545,
+      "step": 716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.966598672487775e-06,
+      "loss": 1.3594,
+      "step": 717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.966490719898147e-06,
+      "loss": 1.3007,
+      "step": 718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9663825951098256e-06,
+      "loss": 1.0648,
+      "step": 719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.966274298135508e-06,
+      "loss": 1.0983,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.966165828987909e-06,
+      "loss": 1.3102,
+      "step": 721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9660571876797673e-06,
+      "loss": 1.2444,
+      "step": 722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9659483742238392e-06,
+      "loss": 1.0806,
+      "step": 723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9658393886329007e-06,
+      "loss": 0.9444,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9657302309197507e-06,
+      "loss": 0.7826,
+      "step": 725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9656209010972054e-06,
+      "loss": 0.4399,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9655113991781033e-06,
+      "loss": 1.1048,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.965401725175302e-06,
+      "loss": 1.1805,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9652918791016805e-06,
+      "loss": 0.9463,
+      "step": 729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.965181860970136e-06,
+      "loss": 0.998,
+      "step": 730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.965071670793588e-06,
+      "loss": 0.9595,
+      "step": 731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9649613085849747e-06,
+      "loss": 0.7779,
+      "step": 732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9648507743572553e-06,
+      "loss": 0.702,
+      "step": 733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.964740068123409e-06,
+      "loss": 1.0655,
+      "step": 734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.964629189896435e-06,
+      "loss": 0.9643,
+      "step": 735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.964518139689354e-06,
+      "loss": 1.2267,
+      "step": 736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.964406917515204e-06,
+      "loss": 1.0465,
+      "step": 737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.964295523387046e-06,
+      "loss": 1.2266,
+      "step": 738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9641839573179597e-06,
+      "loss": 0.9642,
+      "step": 739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9640722193210457e-06,
+      "loss": 0.827,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9639603094094246e-06,
+      "loss": 0.8156,
+      "step": 741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.963848227596237e-06,
+      "loss": 1.0656,
+      "step": 742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.963735973894644e-06,
+      "loss": 1.1434,
+      "step": 743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9636235483178264e-06,
+      "loss": 0.9287,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9635109508789858e-06,
+      "loss": 0.8704,
+      "step": 745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9633981815913433e-06,
+      "loss": 1.0512,
+      "step": 746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.96328524046814e-06,
+      "loss": 1.1009,
+      "step": 747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9631721275226393e-06,
+      "loss": 1.0748,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9630588427681217e-06,
+      "loss": 0.9792,
+      "step": 749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.96294538621789e-06,
+      "loss": 0.8939,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9628317578852665e-06,
+      "loss": 0.825,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.962717957783594e-06,
+      "loss": 0.6481,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.962603985926234e-06,
+      "loss": 0.8289,
+      "step": 753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9624898423265704e-06,
+      "loss": 1.1016,
+      "step": 754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.962375526998006e-06,
+      "loss": 1.1215,
+      "step": 755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9622610399539643e-06,
+      "loss": 1.3029,
+      "step": 756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.962146381207888e-06,
+      "loss": 1.2482,
+      "step": 757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9620315507732404e-06,
+      "loss": 1.2224,
+      "step": 758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9619165486635057e-06,
+      "loss": 0.8364,
+      "step": 759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.961801374892188e-06,
+      "loss": 1.4814,
+      "step": 760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9616860294728104e-06,
+      "loss": 1.3116,
+      "step": 761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9615705124189183e-06,
+      "loss": 1.0055,
+      "step": 762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.961454823744074e-06,
+      "loss": 0.9561,
+      "step": 763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.961338963461864e-06,
+      "loss": 1.1432,
+      "step": 764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9612229315858914e-06,
+      "loss": 1.2983,
+      "step": 765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9611067281297817e-06,
+      "loss": 0.9539,
+      "step": 766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9609903531071795e-06,
+      "loss": 1.0341,
+      "step": 767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.96087380653175e-06,
+      "loss": 0.9092,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.960757088417178e-06,
+      "loss": 0.8342,
+      "step": 769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9606401987771692e-06,
+      "loss": 1.1344,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9605231376254485e-06,
+      "loss": 0.8652,
+      "step": 771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9604059049757628e-06,
+      "loss": 1.1636,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9602885008418763e-06,
+      "loss": 1.0486,
+      "step": 773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9601709252375753e-06,
+      "loss": 1.19,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9600531781766665e-06,
+      "loss": 1.0038,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9599352596729752e-06,
+      "loss": 0.9223,
+      "step": 776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9598171697403482e-06,
+      "loss": 0.9922,
+      "step": 777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9596989083926515e-06,
+      "loss": 1.1491,
+      "step": 778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.959580475643772e-06,
+      "loss": 0.8081,
+      "step": 779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.959461871507616e-06,
+      "loss": 1.1238,
+      "step": 780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.95934309599811e-06,
+      "loss": 1.0524,
+      "step": 781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9592241491292015e-06,
+      "loss": 0.8835,
+      "step": 782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9591050309148574e-06,
+      "loss": 1.093,
+      "step": 783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9589857413690648e-06,
+      "loss": 1.0641,
+      "step": 784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.958866280505831e-06,
+      "loss": 1.0908,
+      "step": 785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.958746648339183e-06,
+      "loss": 1.0964,
+      "step": 786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9586268448831685e-06,
+      "loss": 1.3304,
+      "step": 787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9585068701518547e-06,
+      "loss": 1.1892,
+      "step": 788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9583867241593307e-06,
+      "loss": 1.0283,
+      "step": 789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9582664069197025e-06,
+      "loss": 1.5036,
+      "step": 790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.958145918447099e-06,
+      "loss": 1.0514,
+      "step": 791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9580252587556684e-06,
+      "loss": 1.1048,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9579044278595785e-06,
+      "loss": 0.6791,
+      "step": 793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9577834257730177e-06,
+      "loss": 0.7435,
+      "step": 794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.957662252510194e-06,
+      "loss": 0.9605,
+      "step": 795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.957540908085336e-06,
+      "loss": 1.2126,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9574193925126924e-06,
+      "loss": 1.4153,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9572977058065317e-06,
+      "loss": 1.0654,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.957175847981143e-06,
+      "loss": 1.1137,
+      "step": 799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9570538190508346e-06,
+      "loss": 1.0485,
+      "step": 800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9569316190299355e-06,
+      "loss": 1.1704,
+      "step": 801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9568092479327946e-06,
+      "loss": 0.8121,
+      "step": 802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9566867057737815e-06,
+      "loss": 0.9908,
+      "step": 803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9565639925672855e-06,
+      "loss": 0.9766,
+      "step": 804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9564411083277147e-06,
+      "loss": 1.0819,
+      "step": 805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9563180530694996e-06,
+      "loss": 1.3171,
+      "step": 806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9561948268070892e-06,
+      "loss": 1.0199,
+      "step": 807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.956071429554953e-06,
+      "loss": 0.9818,
+      "step": 808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.955947861327581e-06,
+      "loss": 0.7922,
+      "step": 809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9558241221394815e-06,
+      "loss": 1.3451,
+      "step": 810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9557002120051862e-06,
+      "loss": 1.1229,
+      "step": 811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9555761309392436e-06,
+      "loss": 1.1411,
+      "step": 812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.955451878956224e-06,
+      "loss": 0.8565,
+      "step": 813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.955327456070717e-06,
+      "loss": 0.8403,
+      "step": 814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9552028622973332e-06,
+      "loss": 0.8337,
+      "step": 815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.955078097650702e-06,
+      "loss": 0.9181,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9549531621454743e-06,
+      "loss": 0.8733,
+      "step": 817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9548280557963197e-06,
+      "loss": 1.2485,
+      "step": 818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9547027786179296e-06,
+      "loss": 0.7096,
+      "step": 819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9545773306250127e-06,
+      "loss": 0.8002,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9544517118323e-06,
+      "loss": 1.0956,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9543259222545423e-06,
+      "loss": 1.1533,
+      "step": 822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.95419996190651e-06,
+      "loss": 1.1669,
+      "step": 823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.954073830802993e-06,
+      "loss": 0.8459,
+      "step": 824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953947528958803e-06,
+      "loss": 1.196,
+      "step": 825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9538210563887705e-06,
+      "loss": 0.6672,
+      "step": 826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953694413107745e-06,
+      "loss": 1.0257,
+      "step": 827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9535675991305984e-06,
+      "loss": 0.9098,
+      "step": 828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953440614472221e-06,
+      "loss": 1.0491,
+      "step": 829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953313459147524e-06,
+      "loss": 0.8466,
+      "step": 830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9531861331714373e-06,
+      "loss": 1.1391,
+      "step": 831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9530586365589134e-06,
+      "loss": 0.9923,
+      "step": 832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.952930969324922e-06,
+      "loss": 0.7221,
+      "step": 833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9528031314844545e-06,
+      "loss": 1.1418,
+      "step": 834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9526751230525217e-06,
+      "loss": 1.1161,
+      "step": 835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.952546944044155e-06,
+      "loss": 1.2731,
+      "step": 836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9524185944744047e-06,
+      "loss": 1.0576,
+      "step": 837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9522900743583435e-06,
+      "loss": 1.299,
+      "step": 838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.952161383711061e-06,
+      "loss": 1.1729,
+      "step": 839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.952032522547669e-06,
+      "loss": 1.2685,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.951903490883298e-06,
+      "loss": 0.7864,
+      "step": 841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9517742887331e-06,
+      "loss": 0.9869,
+      "step": 842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9516449161122464e-06,
+      "loss": 1.0114,
+      "step": 843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9515153730359275e-06,
+      "loss": 1.0993,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9513856595193553e-06,
+      "loss": 0.7679,
+      "step": 845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9512557755777605e-06,
+      "loss": 1.0497,
+      "step": 846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9511257212263946e-06,
+      "loss": 1.0331,
+      "step": 847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.950995496480529e-06,
+      "loss": 1.0987,
+      "step": 848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9508651013554548e-06,
+      "loss": 1.0119,
+      "step": 849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9507345358664834e-06,
+      "loss": 0.6745,
+      "step": 850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9506038000289465e-06,
+      "loss": 0.759,
+      "step": 851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9504728938581945e-06,
+      "loss": 1.0333,
+      "step": 852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9503418173695993e-06,
+      "loss": 0.714,
+      "step": 853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.950210570578552e-06,
+      "loss": 1.0262,
+      "step": 854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9500791535004643e-06,
+      "loss": 1.1013,
+      "step": 855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.949947566150767e-06,
+      "loss": 0.977,
+      "step": 856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.949815808544912e-06,
+      "loss": 0.9921,
+      "step": 857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9496838806983696e-06,
+      "loss": 1.2548,
+      "step": 858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9495517826266315e-06,
+      "loss": 1.0739,
+      "step": 859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9494195143452095e-06,
+      "loss": 0.8372,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9492870758696344e-06,
+      "loss": 1.1985,
+      "step": 861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9491544672154574e-06,
+      "loss": 1.2638,
+      "step": 862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.94902168839825e-06,
+      "loss": 1.1846,
+      "step": 863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9488887394336023e-06,
+      "loss": 0.8291,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.948755620337127e-06,
+      "loss": 0.973,
+      "step": 865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.948622331124455e-06,
+      "loss": 0.9326,
+      "step": 866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.948488871811236e-06,
+      "loss": 1.1494,
+      "step": 867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9483552424131425e-06,
+      "loss": 1.2494,
+      "step": 868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9482214429458652e-06,
+      "loss": 1.2245,
+      "step": 869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9480874734251152e-06,
+      "loss": 1.0355,
+      "step": 870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9479533338666235e-06,
+      "loss": 0.7586,
+      "step": 871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.947819024286141e-06,
+      "loss": 0.6242,
+      "step": 872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9476845446994383e-06,
+      "loss": 0.7493,
+      "step": 873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.947549895122307e-06,
+      "loss": 1.0136,
+      "step": 874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9474150755705573e-06,
+      "loss": 0.9459,
+      "step": 875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.94728008606002e-06,
+      "loss": 1.0873,
+      "step": 876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.947144926606547e-06,
+      "loss": 1.3212,
+      "step": 877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9470095972260078e-06,
+      "loss": 1.1887,
+      "step": 878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9468740979342935e-06,
+      "loss": 0.993,
+      "step": 879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9467384287473145e-06,
+      "loss": 1.405,
+      "step": 880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.946602589681002e-06,
+      "loss": 1.0319,
+      "step": 881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.946466580751306e-06,
+      "loss": 1.3933,
+      "step": 882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9463304019741973e-06,
+      "loss": 1.0465,
+      "step": 883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9461940533656656e-06,
+      "loss": 1.0652,
+      "step": 884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9460575349417227e-06,
+      "loss": 1.2942,
+      "step": 885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9459208467183975e-06,
+      "loss": 1.0143,
+      "step": 886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9457839887117408e-06,
+      "loss": 0.8945,
+      "step": 887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.945646960937823e-06,
+      "loss": 1.051,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9455097634127337e-06,
+      "loss": 1.0819,
+      "step": 889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9453723961525833e-06,
+      "loss": 1.4003,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9452348591735017e-06,
+      "loss": 1.0707,
+      "step": 891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9450971524916385e-06,
+      "loss": 0.6619,
+      "step": 892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.944959276123164e-06,
+      "loss": 0.9369,
+      "step": 893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9448212300842682e-06,
+      "loss": 1.2272,
+      "step": 894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.94468301439116e-06,
+      "loss": 1.2543,
+      "step": 895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.94454462906007e-06,
+      "loss": 1.0865,
+      "step": 896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.944406074107246e-06,
+      "loss": 1.2047,
+      "step": 897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9442673495489595e-06,
+      "loss": 0.9421,
+      "step": 898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.944128455401498e-06,
+      "loss": 0.9132,
+      "step": 899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9439893916811723e-06,
+      "loss": 1.2146,
+      "step": 900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.943850158404311e-06,
+      "loss": 0.7985,
+      "step": 901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9437107555872632e-06,
+      "loss": 1.1325,
+      "step": 902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9435711832463975e-06,
+      "loss": 1.0415,
+      "step": 903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.943431441398103e-06,
+      "loss": 1.1776,
+      "step": 904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9432915300587886e-06,
+      "loss": 0.9809,
+      "step": 905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9431514492448836e-06,
+      "loss": 0.8944,
+      "step": 906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9430111989728353e-06,
+      "loss": 0.9421,
+      "step": 907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.942870779259114e-06,
+      "loss": 0.8007,
+      "step": 908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.942730190120206e-06,
+      "loss": 0.8999,
+      "step": 909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9425894315726213e-06,
+      "loss": 1.3596,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.942448503632887e-06,
+      "loss": 1.0807,
+      "step": 911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9423074063175524e-06,
+      "loss": 0.7669,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.942166139643184e-06,
+      "loss": 0.6416,
+      "step": 913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.942024703626371e-06,
+      "loss": 0.835,
+      "step": 914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.94188309828372e-06,
+      "loss": 0.9678,
+      "step": 915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9417413236318593e-06,
+      "loss": 1.1446,
+      "step": 916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.941599379687436e-06,
+      "loss": 0.8227,
+      "step": 917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9414572664671182e-06,
+      "loss": 0.9563,
+      "step": 918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.941314983987592e-06,
+      "loss": 1.1794,
+      "step": 919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9411725322655658e-06,
+      "loss": 0.9981,
+      "step": 920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9410299113177657e-06,
+      "loss": 1.1344,
+      "step": 921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.940887121160939e-06,
+      "loss": 1.0589,
+      "step": 922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9407441618118523e-06,
+      "loss": 1.0436,
+      "step": 923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9406010332872925e-06,
+      "loss": 0.8618,
+      "step": 924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9404577356040655e-06,
+      "loss": 1.1137,
+      "step": 925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.940314268778998e-06,
+      "loss": 1.1607,
+      "step": 926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9401706328289365e-06,
+      "loss": 0.9021,
+      "step": 927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9400268277707464e-06,
+      "loss": 1.1061,
+      "step": 928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.939882853621314e-06,
+      "loss": 0.8798,
+      "step": 929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9397387103975448e-06,
+      "loss": 1.0488,
+      "step": 930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.939594398116365e-06,
+      "loss": 1.2219,
+      "step": 931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9394499167947198e-06,
+      "loss": 0.5683,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9393052664495744e-06,
+      "loss": 0.9425,
+      "step": 933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.939160447097914e-06,
+      "loss": 1.1467,
+      "step": 934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9390154587567436e-06,
+      "loss": 1.1428,
+      "step": 935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.938870301443088e-06,
+      "loss": 0.8801,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9387249751739924e-06,
+      "loss": 1.1294,
+      "step": 937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9385794799665204e-06,
+      "loss": 1.2765,
+      "step": 938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9384338158377576e-06,
+      "loss": 0.8579,
+      "step": 939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.938287982804807e-06,
+      "loss": 1.2596,
+      "step": 940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9381419808847934e-06,
+      "loss": 1.2239,
+      "step": 941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.937995810094861e-06,
+      "loss": 1.0503,
+      "step": 942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.937849470452172e-06,
+      "loss": 0.9454,
+      "step": 943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9377029619739116e-06,
+      "loss": 1.0779,
+      "step": 944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9375562846772827e-06,
+      "loss": 0.978,
+      "step": 945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9374094385795075e-06,
+      "loss": 0.9579,
+      "step": 946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9372624236978306e-06,
+      "loss": 1.1602,
+      "step": 947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9371152400495133e-06,
+      "loss": 1.1092,
+      "step": 948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.936967887651839e-06,
+      "loss": 1.1953,
+      "step": 949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9368203665221106e-06,
+      "loss": 0.9316,
+      "step": 950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.93667267667765e-06,
+      "loss": 1.0263,
+      "step": 951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9365248181357982e-06,
+      "loss": 0.8453,
+      "step": 952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9363767909139187e-06,
+      "loss": 1.2129,
+      "step": 953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.936228595029392e-06,
+      "loss": 1.0142,
+      "step": 954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.936080230499621e-06,
+      "loss": 0.8644,
+      "step": 955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9359316973420258e-06,
+      "loss": 0.9413,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.935782995574047e-06,
+      "loss": 0.9315,
+      "step": 957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9356341252131473e-06,
+      "loss": 0.805,
+      "step": 958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.935485086276806e-06,
+      "loss": 1.1874,
+      "step": 959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.935335878782524e-06,
+      "loss": 0.8654,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9351865027478223e-06,
+      "loss": 0.7039,
+      "step": 961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9350369581902397e-06,
+      "loss": 0.6442,
+      "step": 962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9348872451273364e-06,
+      "loss": 1.0855,
+      "step": 963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9347373635766932e-06,
+      "loss": 1.1794,
+      "step": 964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.934587313555908e-06,
+      "loss": 1.1012,
+      "step": 965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9344370950826008e-06,
+      "loss": 0.6425,
+      "step": 966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.93428670817441e-06,
+      "loss": 1.0236,
+      "step": 967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9341361528489954e-06,
+      "loss": 1.0628,
+      "step": 968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9339854291240346e-06,
+      "loss": 1.1964,
+      "step": 969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9338345370172265e-06,
+      "loss": 1.1374,
+      "step": 970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9336834765462886e-06,
+      "loss": 0.8944,
+      "step": 971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.93353224772896e-06,
+      "loss": 1.0101,
+      "step": 972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9333808505829966e-06,
+      "loss": 0.9595,
+      "step": 973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9332292851261774e-06,
+      "loss": 1.5989,
+      "step": 974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.933077551376298e-06,
+      "loss": 0.9131,
+      "step": 975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.932925649351177e-06,
+      "loss": 1.2336,
+      "step": 976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.93277357906865e-06,
+      "loss": 0.9783,
+      "step": 977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.932621340546574e-06,
+      "loss": 0.8633,
+      "step": 978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9324689338028243e-06,
+      "loss": 1.1258,
+      "step": 979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.932316358855298e-06,
+      "loss": 1.1425,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.93216361572191e-06,
+      "loss": 0.6341,
+      "step": 981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9320107044205954e-06,
+      "loss": 0.955,
+      "step": 982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.931857624969311e-06,
+      "loss": 1.2249,
+      "step": 983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.93170437738603e-06,
+      "loss": 1.0899,
+      "step": 984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9315509616887482e-06,
+      "loss": 1.2387,
+      "step": 985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9313973778954795e-06,
+      "loss": 1.2573,
+      "step": 986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9312436260242582e-06,
+      "loss": 1.1757,
+      "step": 987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.931089706093138e-06,
+      "loss": 0.8837,
+      "step": 988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.930935618120193e-06,
+      "loss": 0.7885,
+      "step": 989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9307813621235164e-06,
+      "loss": 1.2819,
+      "step": 990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9306269381212206e-06,
+      "loss": 0.9802,
+      "step": 991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.930472346131439e-06,
+      "loss": 0.5953,
+      "step": 992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.930317586172325e-06,
+      "loss": 1.0562,
+      "step": 993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9301626582620493e-06,
+      "loss": 1.0448,
+      "step": 994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.930007562418805e-06,
+      "loss": 1.3228,
+      "step": 995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.929852298660803e-06,
+      "loss": 0.9678,
+      "step": 996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9296968670062755e-06,
+      "loss": 0.8337,
+      "step": 997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.929541267473473e-06,
+      "loss": 1.0928,
+      "step": 998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9293855000806665e-06,
+      "loss": 0.7941,
+      "step": 999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9292295648461465e-06,
+      "loss": 1.1375,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9290734617882237e-06,
+      "loss": 0.9867,
+      "step": 1001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9289171909252284e-06,
+      "loss": 0.9038,
+      "step": 1002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.928760752275509e-06,
+      "loss": 0.8326,
+      "step": 1003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.928604145857436e-06,
+      "loss": 1.009,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9284473716893978e-06,
+      "loss": 0.7926,
+      "step": 1005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9282904297898035e-06,
+      "loss": 1.2134,
+      "step": 1006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9281333201770813e-06,
+      "loss": 1.0612,
+      "step": 1007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9279760428696804e-06,
+      "loss": 0.8213,
+      "step": 1008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9278185978860674e-06,
+      "loss": 0.8472,
+      "step": 1009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9276609852447304e-06,
+      "loss": 0.9122,
+      "step": 1010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.927503204964176e-06,
+      "loss": 1.0025,
+      "step": 1011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9273452570629325e-06,
+      "loss": 0.9094,
+      "step": 1012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.927187141559545e-06,
+      "loss": 1.3681,
+      "step": 1013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9270288584725812e-06,
+      "loss": 1.3082,
+      "step": 1014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.926870407820626e-06,
+      "loss": 0.7559,
+      "step": 1015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9267117896222848e-06,
+      "loss": 0.8261,
+      "step": 1016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.926553003896184e-06,
+      "loss": 0.9812,
+      "step": 1017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.926394050660968e-06,
+      "loss": 0.95,
+      "step": 1018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.926234929935302e-06,
+      "loss": 1.2821,
+      "step": 1019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.926075641737869e-06,
+      "loss": 0.8821,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.925916186087374e-06,
+      "loss": 1.0386,
+      "step": 1021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9257565630025408e-06,
+      "loss": 1.3073,
+      "step": 1022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.925596772502112e-06,
+      "loss": 1.1673,
+      "step": 1023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9254368146048517e-06,
+      "loss": 1.1257,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.925276689329541e-06,
+      "loss": 0.7219,
+      "step": 1025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.925116396694983e-06,
+      "loss": 0.9967,
+      "step": 1026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.92495593672e-06,
+      "loss": 1.0274,
+      "step": 1027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9247953094234324e-06,
+      "loss": 0.7967,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9246345148241423e-06,
+      "loss": 0.9687,
+      "step": 1029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9244735529410107e-06,
+      "loss": 0.9867,
+      "step": 1030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.924312423792938e-06,
+      "loss": 0.9928,
+      "step": 1031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9241511273988443e-06,
+      "loss": 0.4831,
+      "step": 1032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.923989663777669e-06,
+      "loss": 0.8635,
+      "step": 1033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.923828032948372e-06,
+      "loss": 1.0183,
+      "step": 1034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.923666234929932e-06,
+      "loss": 1.2433,
+      "step": 1035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.923504269741348e-06,
+      "loss": 1.016,
+      "step": 1036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.923342137401638e-06,
+      "loss": 1.0538,
+      "step": 1037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9231798379298404e-06,
+      "loss": 1.2156,
+      "step": 1038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9230173713450124e-06,
+      "loss": 1.2698,
+      "step": 1039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.922854737666231e-06,
+      "loss": 1.0256,
+      "step": 1040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9226919369125937e-06,
+      "loss": 1.087,
+      "step": 1041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9225289691032164e-06,
+      "loss": 1.1647,
+      "step": 1042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9223658342572356e-06,
+      "loss": 1.3297,
+      "step": 1043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.922202532393807e-06,
+      "loss": 1.1143,
+      "step": 1044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9220390635321045e-06,
+      "loss": 1.044,
+      "step": 1045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.921875427691325e-06,
+      "loss": 1.0758,
+      "step": 1046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9217116248906816e-06,
+      "loss": 0.8036,
+      "step": 1047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9215476551494086e-06,
+      "loss": 1.0451,
+      "step": 1048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9213835184867604e-06,
+      "loss": 1.3635,
+      "step": 1049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9212192149220095e-06,
+      "loss": 1.04,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.92105474447445e-06,
+      "loss": 0.8852,
+      "step": 1051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9208901071633927e-06,
+      "loss": 1.0754,
+      "step": 1052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9207253030081707e-06,
+      "loss": 1.0441,
+      "step": 1053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9205603320281356e-06,
+      "loss": 0.5929,
+      "step": 1054
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9203951942426586e-06,
+      "loss": 0.912,
+      "step": 1055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9202298896711304e-06,
+      "loss": 0.9385,
+      "step": 1056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.920064418332962e-06,
+      "loss": 0.8818,
+      "step": 1057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9198987802475827e-06,
+      "loss": 0.8678,
+      "step": 1058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9197329754344424e-06,
+      "loss": 1.0032,
+      "step": 1059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9195670039130103e-06,
+      "loss": 1.1993,
+      "step": 1060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.919400865702775e-06,
+      "loss": 0.9237,
+      "step": 1061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9192345608232457e-06,
+      "loss": 1.1928,
+      "step": 1062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9190680892939487e-06,
+      "loss": 0.8844,
+      "step": 1063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.918901451134433e-06,
+      "loss": 1.3083,
+      "step": 1064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9187346463642646e-06,
+      "loss": 0.8613,
+      "step": 1065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9185676750030307e-06,
+      "loss": 1.0481,
+      "step": 1066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.918400537070337e-06,
+      "loss": 0.5445,
+      "step": 1067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9182332325858094e-06,
+      "loss": 0.7222,
+      "step": 1068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9180657615690933e-06,
+      "loss": 0.7606,
+      "step": 1069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9178981240398533e-06,
+      "loss": 1.01,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9177303200177745e-06,
+      "loss": 1.0333,
+      "step": 1071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9175623495225595e-06,
+      "loss": 0.9392,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.917394212573933e-06,
+      "loss": 1.0995,
+      "step": 1073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.917225909191638e-06,
+      "loss": 0.7307,
+      "step": 1074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.917057439395436e-06,
+      "loss": 1.4723,
+      "step": 1075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.91688880320511e-06,
+      "loss": 0.9406,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.916720000640461e-06,
+      "loss": 0.8594,
+      "step": 1077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9165510317213113e-06,
+      "loss": 1.1443,
+      "step": 1078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9163818964675e-06,
+      "loss": 0.5679,
+      "step": 1079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.916212594898889e-06,
+      "loss": 0.8811,
+      "step": 1080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9160431270353572e-06,
+      "loss": 1.1736,
+      "step": 1081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.915873492896804e-06,
+      "loss": 1.2539,
+      "step": 1082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.915703692503148e-06,
+      "loss": 0.5927,
+      "step": 1083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.915533725874328e-06,
+      "loss": 0.6974,
+      "step": 1084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.915363593030302e-06,
+      "loss": 1.1239,
+      "step": 1085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9151932939910465e-06,
+      "loss": 0.7877,
+      "step": 1086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9150228287765592e-06,
+      "loss": 1.2002,
+      "step": 1087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.914852197406856e-06,
+      "loss": 1.1548,
+      "step": 1088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.914681399901974e-06,
+      "loss": 0.972,
+      "step": 1089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9145104362819666e-06,
+      "loss": 1.0343,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.914339306566911e-06,
+      "loss": 1.1537,
+      "step": 1091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9141680107768997e-06,
+      "loss": 0.8685,
+      "step": 1092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9139965489320477e-06,
+      "loss": 1.2266,
+      "step": 1093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9138249210524877e-06,
+      "loss": 1.0454,
+      "step": 1094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.913653127158374e-06,
+      "loss": 1.1902,
+      "step": 1095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.913481167269878e-06,
+      "loss": 1.3076,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9133090414071914e-06,
+      "loss": 0.8986,
+      "step": 1097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9131367495905263e-06,
+      "loss": 1.0318,
+      "step": 1098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.912964291840113e-06,
+      "loss": 1.1292,
+      "step": 1099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.912791668176203e-06,
+      "loss": 1.2041,
+      "step": 1100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9126188786190646e-06,
+      "loss": 0.9368,
+      "step": 1101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.912445923188988e-06,
+      "loss": 1.0162,
+      "step": 1102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.912272801906282e-06,
+      "loss": 1.0391,
+      "step": 1103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.912099514791275e-06,
+      "loss": 0.8176,
+      "step": 1104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9119260618643145e-06,
+      "loss": 1.0232,
+      "step": 1105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9117524431457678e-06,
+      "loss": 1.0424,
+      "step": 1106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9115786586560216e-06,
+      "loss": 0.784,
+      "step": 1107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.911404708415482e-06,
+      "loss": 1.0392,
+      "step": 1108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9112305924445748e-06,
+      "loss": 1.348,
+      "step": 1109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.911056310763745e-06,
+      "loss": 0.9373,
+      "step": 1110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9108818633934577e-06,
+      "loss": 0.8979,
+      "step": 1111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.910707250354196e-06,
+      "loss": 0.7092,
+      "step": 1112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.910532471666464e-06,
+      "loss": 1.0774,
+      "step": 1113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9103575273507838e-06,
+      "loss": 0.9377,
+      "step": 1114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9101824174276993e-06,
+      "loss": 1.2412,
+      "step": 1115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9100071419177704e-06,
+      "loss": 0.8608,
+      "step": 1116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9098317008415796e-06,
+      "loss": 1.069,
+      "step": 1117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.909656094219727e-06,
+      "loss": 1.1998,
+      "step": 1118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9094803220728336e-06,
+      "loss": 0.9115,
+      "step": 1119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9093043844215387e-06,
+      "loss": 1.1323,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9091282812865002e-06,
+      "loss": 0.9386,
+      "step": 1121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9089520126883977e-06,
+      "loss": 1.3749,
+      "step": 1122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9087755786479287e-06,
+      "loss": 1.2341,
+      "step": 1123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9085989791858104e-06,
+      "loss": 0.9957,
+      "step": 1124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9084222143227796e-06,
+      "loss": 0.9658,
+      "step": 1125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.908245284079592e-06,
+      "loss": 1.0964,
+      "step": 1126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.908068188477024e-06,
+      "loss": 1.2345,
+      "step": 1127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9078909275358703e-06,
+      "loss": 0.9007,
+      "step": 1128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9077135012769445e-06,
+      "loss": 1.0418,
+      "step": 1129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9075359097210815e-06,
+      "loss": 1.1044,
+      "step": 1130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.907358152889134e-06,
+      "loss": 0.9292,
+      "step": 1131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9071802308019746e-06,
+      "loss": 1.1871,
+      "step": 1132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9070021434804952e-06,
+      "loss": 0.8854,
+      "step": 1133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.906823890945608e-06,
+      "loss": 0.9536,
+      "step": 1134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9066454732182426e-06,
+      "loss": 1.3593,
+      "step": 1135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.90646689031935e-06,
+      "loss": 1.0656,
+      "step": 1136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9062881422699e-06,
+      "loss": 0.9117,
+      "step": 1137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9061092290908806e-06,
+      "loss": 1.2503,
+      "step": 1138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.905930150803301e-06,
+      "loss": 1.062,
+      "step": 1139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9057509074281892e-06,
+      "loss": 1.1214,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9055714989865917e-06,
+      "loss": 0.4302,
+      "step": 1141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9053919254995754e-06,
+      "loss": 0.9146,
+      "step": 1142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9052121869882262e-06,
+      "loss": 1.1927,
+      "step": 1143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9050322834736494e-06,
+      "loss": 1.0486,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9048522149769695e-06,
+      "loss": 1.0296,
+      "step": 1145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9046719815193305e-06,
+      "loss": 0.7783,
+      "step": 1146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9044915831218965e-06,
+      "loss": 1.2398,
+      "step": 1147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9043110198058495e-06,
+      "loss": 1.0766,
+      "step": 1148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.904130291592392e-06,
+      "loss": 0.9817,
+      "step": 1149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9039493985027455e-06,
+      "loss": 1.1198,
+      "step": 1150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.903768340558151e-06,
+      "loss": 0.9857,
+      "step": 1151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.903587117779869e-06,
+      "loss": 1.0523,
+      "step": 1152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.903405730189178e-06,
+      "loss": 0.7595,
+      "step": 1153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.903224177807378e-06,
+      "loss": 1.256,
+      "step": 1154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9030424606557867e-06,
+      "loss": 1.2903,
+      "step": 1155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9028605787557424e-06,
+      "loss": 0.958,
+      "step": 1156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9026785321286014e-06,
+      "loss": 0.9544,
+      "step": 1157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9024963207957404e-06,
+      "loss": 1.0159,
+      "step": 1158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9023139447785553e-06,
+      "loss": 1.0906,
+      "step": 1159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9021314040984605e-06,
+      "loss": 1.2121,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.2559232711791992,
+      "eval_runtime": 32.0093,
+      "eval_samples_per_second": 38.895,
+      "eval_steps_per_second": 19.463,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9019486987768905e-06,
+      "loss": 0.9211,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.901765828835299e-06,
+      "loss": 0.9292,
+      "step": 1162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9015827942951596e-06,
+      "loss": 1.0443,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.901399595177964e-06,
+      "loss": 1.0231,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.901216231505224e-06,
+      "loss": 0.9451,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9010327032984707e-06,
+      "loss": 0.9782,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.900849010579254e-06,
+      "loss": 1.1621,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9006651533691443e-06,
+      "loss": 1.0195,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9004811316897295e-06,
+      "loss": 0.8629,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9002969455626186e-06,
+      "loss": 0.8675,
+      "step": 1170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.900112595009439e-06,
+      "loss": 0.9815,
+      "step": 1171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8999280800518373e-06,
+      "loss": 1.0908,
+      "step": 1172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.89974340071148e-06,
+      "loss": 0.9196,
+      "step": 1173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8995585570100527e-06,
+      "loss": 1.3169,
+      "step": 1174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8993735489692597e-06,
+      "loss": 1.3139,
+      "step": 1175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8991883766108254e-06,
+      "loss": 1.1849,
+      "step": 1176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.899003039956493e-06,
+      "loss": 0.9625,
+      "step": 1177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8988175390280252e-06,
+      "loss": 1.1945,
+      "step": 1178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.898631873847204e-06,
+      "loss": 0.6556,
+      "step": 1179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.898446044435831e-06,
+      "loss": 1.1115,
+      "step": 1180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8982600508157257e-06,
+      "loss": 0.9984,
+      "step": 1181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8980738930087286e-06,
+      "loss": 0.7876,
+      "step": 1182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8978875710366987e-06,
+      "loss": 0.7712,
+      "step": 1183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8977010849215146e-06,
+      "loss": 0.8851,
+      "step": 1184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8975144346850733e-06,
+      "loss": 0.7486,
+      "step": 1185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8973276203492923e-06,
+      "loss": 0.6672,
+      "step": 1186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8971406419361075e-06,
+      "loss": 1.02,
+      "step": 1187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.896953499467474e-06,
+      "loss": 0.6818,
+      "step": 1188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.896766192965367e-06,
+      "loss": 0.9379,
+      "step": 1189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8965787224517803e-06,
+      "loss": 0.6417,
+      "step": 1190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.896391087948727e-06,
+      "loss": 1.23,
+      "step": 1191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.89620328947824e-06,
+      "loss": 1.0698,
+      "step": 1192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8960153270623705e-06,
+      "loss": 1.0197,
+      "step": 1193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.89582720072319e-06,
+      "loss": 0.7748,
+      "step": 1194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.895638910482788e-06,
+      "loss": 0.7834,
+      "step": 1195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8954504563632746e-06,
+      "loss": 0.9227,
+      "step": 1196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.895261838386778e-06,
+      "loss": 1.0536,
+      "step": 1197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8950730565754468e-06,
+      "loss": 1.0397,
+      "step": 1198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8948841109514473e-06,
+      "loss": 1.0011,
+      "step": 1199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.894695001536967e-06,
+      "loss": 0.7613,
+      "step": 1200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8945057283542114e-06,
+      "loss": 0.6283,
+      "step": 1201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8943162914254044e-06,
+      "loss": 0.9877,
+      "step": 1202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.894126690772791e-06,
+      "loss": 0.8313,
+      "step": 1203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.893936926418634e-06,
+      "loss": 0.6709,
+      "step": 1204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8937469983852166e-06,
+      "loss": 0.652,
+      "step": 1205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.89355690669484e-06,
+      "loss": 1.0175,
+      "step": 1206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8933666513698255e-06,
+      "loss": 0.6683,
+      "step": 1207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.893176232432513e-06,
+      "loss": 1.164,
+      "step": 1208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8929856499052627e-06,
+      "loss": 0.8025,
+      "step": 1209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.892794903810452e-06,
+      "loss": 1.1844,
+      "step": 1210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.89260399417048e-06,
+      "loss": 0.9697,
+      "step": 1211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.892412921007763e-06,
+      "loss": 0.8668,
+      "step": 1212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8922216843447377e-06,
+      "loss": 0.8056,
+      "step": 1213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.892030284203859e-06,
+      "loss": 1.3095,
+      "step": 1214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.891838720607602e-06,
+      "loss": 0.8904,
+      "step": 1215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.89164699357846e-06,
+      "loss": 0.6892,
+      "step": 1216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.891455103138947e-06,
+      "loss": 1.2712,
+      "step": 1217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8912630493115945e-06,
+      "loss": 1.2733,
+      "step": 1218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8910708321189537e-06,
+      "loss": 1.0814,
+      "step": 1219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8908784515835955e-06,
+      "loss": 0.9849,
+      "step": 1220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8906859077281095e-06,
+      "loss": 0.9462,
+      "step": 1221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.890493200575105e-06,
+      "loss": 1.0852,
+      "step": 1222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.89030033014721e-06,
+      "loss": 1.0688,
+      "step": 1223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.890107296467071e-06,
+      "loss": 0.8875,
+      "step": 1224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8899140995573566e-06,
+      "loss": 1.0368,
+      "step": 1225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8897207394407498e-06,
+      "loss": 0.9513,
+      "step": 1226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.889527216139957e-06,
+      "loss": 0.9748,
+      "step": 1227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8893335296777017e-06,
+      "loss": 1.0107,
+      "step": 1228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8891396800767267e-06,
+      "loss": 0.6807,
+      "step": 1229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8889456673597948e-06,
+      "loss": 1.1789,
+      "step": 1230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.888751491549687e-06,
+      "loss": 1.3071,
+      "step": 1231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8885571526692045e-06,
+      "loss": 1.0462,
+      "step": 1232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8883626507411665e-06,
+      "loss": 0.7864,
+      "step": 1233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.888167985788412e-06,
+      "loss": 0.8989,
+      "step": 1234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8879731578337985e-06,
+      "loss": 1.0803,
+      "step": 1235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8877781669002036e-06,
+      "loss": 1.1229,
+      "step": 1236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8875830130105237e-06,
+      "loss": 0.8333,
+      "step": 1237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.887387696187674e-06,
+      "loss": 1.3724,
+      "step": 1238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8871922164545897e-06,
+      "loss": 1.4597,
+      "step": 1239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8869965738342235e-06,
+      "loss": 1.0465,
+      "step": 1240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8868007683495483e-06,
+      "loss": 0.8199,
+      "step": 1241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8866048000235565e-06,
+      "loss": 1.1498,
+      "step": 1242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8864086688792593e-06,
+      "loss": 1.4164,
+      "step": 1243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8862123749396863e-06,
+      "loss": 0.6391,
+      "step": 1244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.886015918227887e-06,
+      "loss": 1.3452,
+      "step": 1245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.88581929876693e-06,
+      "loss": 0.9862,
+      "step": 1246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8856225165799028e-06,
+      "loss": 1.04,
+      "step": 1247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8854255716899118e-06,
+      "loss": 0.9922,
+      "step": 1248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8852284641200824e-06,
+      "loss": 1.0557,
+      "step": 1249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8850311938935603e-06,
+      "loss": 1.199,
+      "step": 1250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.884833761033508e-06,
+      "loss": 0.7847,
+      "step": 1251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8846361655631106e-06,
+      "loss": 0.7972,
+      "step": 1252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8844384075055685e-06,
+      "loss": 1.0405,
+      "step": 1253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8842404868841033e-06,
+      "loss": 1.2016,
+      "step": 1254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.884042403721956e-06,
+      "loss": 0.722,
+      "step": 1255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.883844158042385e-06,
+      "loss": 1.1715,
+      "step": 1256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.883645749868669e-06,
+      "loss": 0.9869,
+      "step": 1257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.883447179224106e-06,
+      "loss": 0.8567,
+      "step": 1258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.883248446132012e-06,
+      "loss": 0.9001,
+      "step": 1259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8830495506157236e-06,
+      "loss": 0.6941,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8828504926985946e-06,
+      "loss": 0.8824,
+      "step": 1261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8826512724039994e-06,
+      "loss": 0.9862,
+      "step": 1262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.882451889755331e-06,
+      "loss": 1.0098,
+      "step": 1263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8822523447760006e-06,
+      "loss": 1.2138,
+      "step": 1264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.88205263748944e-06,
+      "loss": 1.2082,
+      "step": 1265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8818527679190987e-06,
+      "loss": 0.7295,
+      "step": 1266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8816527360884467e-06,
+      "loss": 1.1502,
+      "step": 1267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.881452542020971e-06,
+      "loss": 1.5621,
+      "step": 1268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8812521857401805e-06,
+      "loss": 1.3778,
+      "step": 1269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.881051667269599e-06,
+      "loss": 1.1541,
+      "step": 1270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.880850986632775e-06,
+      "loss": 1.119,
+      "step": 1271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8806501438532704e-06,
+      "loss": 1.3808,
+      "step": 1272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8804491389546694e-06,
+      "loss": 0.6305,
+      "step": 1273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8802479719605744e-06,
+      "loss": 1.0649,
+      "step": 1274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.880046642894607e-06,
+      "loss": 0.8037,
+      "step": 1275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.879845151780408e-06,
+      "loss": 0.9648,
+      "step": 1276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.879643498641636e-06,
+      "loss": 1.1234,
+      "step": 1277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.879441683501971e-06,
+      "loss": 0.8318,
+      "step": 1278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.879239706385109e-06,
+      "loss": 1.0203,
+      "step": 1279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8790375673147683e-06,
+      "loss": 1.0568,
+      "step": 1280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8788352663146834e-06,
+      "loss": 0.954,
+      "step": 1281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.878632803408609e-06,
+      "loss": 1.2038,
+      "step": 1282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.878430178620319e-06,
+      "loss": 0.9098,
+      "step": 1283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8782273919736057e-06,
+      "loss": 0.8451,
+      "step": 1284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8780244434922816e-06,
+      "loss": 1.1456,
+      "step": 1285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8778213332001767e-06,
+      "loss": 0.9952,
+      "step": 1286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.877618061121141e-06,
+      "loss": 1.0445,
+      "step": 1287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8774146272790423e-06,
+      "loss": 0.9341,
+      "step": 1288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8772110316977697e-06,
+      "loss": 1.2445,
+      "step": 1289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.877007274401229e-06,
+      "loss": 1.2265,
+      "step": 1290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8768033554133463e-06,
+      "loss": 0.971,
+      "step": 1291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.876599274758065e-06,
+      "loss": 0.9309,
+      "step": 1292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.87639503245935e-06,
+      "loss": 0.856,
+      "step": 1293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.876190628541184e-06,
+      "loss": 0.9755,
+      "step": 1294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8759860630275684e-06,
+      "loss": 1.1926,
+      "step": 1295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8757813359425226e-06,
+      "loss": 0.7647,
+      "step": 1296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8755764473100876e-06,
+      "loss": 1.0114,
+      "step": 1297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8753713971543204e-06,
+      "loss": 0.9405,
+      "step": 1298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8751661854993e-06,
+      "loss": 0.6947,
+      "step": 1299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8749608123691223e-06,
+      "loss": 0.8775,
+      "step": 1300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8747552777879023e-06,
+      "loss": 1.066,
+      "step": 1301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8745495817797747e-06,
+      "loss": 1.1364,
+      "step": 1302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8743437243688925e-06,
+      "loss": 1.0033,
+      "step": 1303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.874137705579428e-06,
+      "loss": 0.9826,
+      "step": 1304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8739315254355727e-06,
+      "loss": 0.8425,
+      "step": 1305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8737251839615366e-06,
+      "loss": 0.7282,
+      "step": 1306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8735186811815492e-06,
+      "loss": 1.0768,
+      "step": 1307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8733120171198572e-06,
+      "loss": 0.8397,
+      "step": 1308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8731051918007286e-06,
+      "loss": 0.8264,
+      "step": 1309
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8728982052484502e-06,
+      "loss": 0.9964,
+      "step": 1310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8726910574873244e-06,
+      "loss": 1.0444,
+      "step": 1311
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.872483748541677e-06,
+      "loss": 0.7449,
+      "step": 1312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.87227627843585e-06,
+      "loss": 1.0238,
+      "step": 1313
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.872068647194205e-06,
+      "loss": 1.1267,
+      "step": 1314
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.871860854841123e-06,
+      "loss": 1.0272,
+      "step": 1315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8716529014010024e-06,
+      "loss": 0.7685,
+      "step": 1316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.871444786898262e-06,
+      "loss": 0.9681,
+      "step": 1317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.87123651135734e-06,
+      "loss": 0.5305,
+      "step": 1318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8710280748026917e-06,
+      "loss": 0.8744,
+      "step": 1319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8708194772587926e-06,
+      "loss": 0.9501,
+      "step": 1320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.870610718750136e-06,
+      "loss": 0.7043,
+      "step": 1321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8704017993012358e-06,
+      "loss": 0.9231,
+      "step": 1322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.870192718936623e-06,
+      "loss": 0.8569,
+      "step": 1323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.869983477680848e-06,
+      "loss": 1.1905,
+      "step": 1324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8697740755584816e-06,
+      "loss": 0.9734,
+      "step": 1325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8695645125941115e-06,
+      "loss": 1.1233,
+      "step": 1326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.869354788812345e-06,
+      "loss": 0.9231,
+      "step": 1327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8691449042378088e-06,
+      "loss": 1.0363,
+      "step": 1328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.868934858895148e-06,
+      "loss": 0.8991,
+      "step": 1329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.868724652809026e-06,
+      "loss": 1.068,
+      "step": 1330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8685142860041264e-06,
+      "loss": 1.0249,
+      "step": 1331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.86830375850515e-06,
+      "loss": 1.4233,
+      "step": 1332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.868093070336818e-06,
+      "loss": 0.7852,
+      "step": 1333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8678822215238707e-06,
+      "loss": 0.9044,
+      "step": 1334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.867671212091065e-06,
+      "loss": 1.2712,
+      "step": 1335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8674600420631795e-06,
+      "loss": 1.0156,
+      "step": 1336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.867248711465009e-06,
+      "loss": 0.8303,
+      "step": 1337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8670372203213687e-06,
+      "loss": 0.9628,
+      "step": 1338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.866825568657093e-06,
+      "loss": 0.9611,
+      "step": 1339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8666137564970344e-06,
+      "loss": 0.8304,
+      "step": 1340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8664017838660637e-06,
+      "loss": 1.3079,
+      "step": 1341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8661896507890715e-06,
+      "loss": 0.6073,
+      "step": 1342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8659773572909673e-06,
+      "loss": 1.2698,
+      "step": 1343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.865764903396679e-06,
+      "loss": 0.9982,
+      "step": 1344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.865552289131153e-06,
+      "loss": 0.9936,
+      "step": 1345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8653395145193556e-06,
+      "loss": 1.1147,
+      "step": 1346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8651265795862707e-06,
+      "loss": 0.9309,
+      "step": 1347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.864913484356902e-06,
+      "loss": 1.383,
+      "step": 1348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8647002288562714e-06,
+      "loss": 0.8988,
+      "step": 1349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.86448681310942e-06,
+      "loss": 0.98,
+      "step": 1350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.864273237141407e-06,
+      "loss": 0.87,
+      "step": 1351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8640595009773124e-06,
+      "loss": 0.9079,
+      "step": 1352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8638456046422317e-06,
+      "loss": 1.0663,
+      "step": 1353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8636315481612824e-06,
+      "loss": 1.1011,
+      "step": 1354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8634173315595994e-06,
+      "loss": 1.0382,
+      "step": 1355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.863202954862336e-06,
+      "loss": 0.9427,
+      "step": 1356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.862988418094665e-06,
+      "loss": 1.1291,
+      "step": 1357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8627737212817776e-06,
+      "loss": 1.2597,
+      "step": 1358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8625588644488845e-06,
+      "loss": 1.3439,
+      "step": 1359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8623438476212144e-06,
+      "loss": 1.084,
+      "step": 1360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8621286708240154e-06,
+      "loss": 1.1777,
+      "step": 1361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8619133340825535e-06,
+      "loss": 0.8326,
+      "step": 1362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8616978374221144e-06,
+      "loss": 0.7669,
+      "step": 1363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.861482180868002e-06,
+      "loss": 0.8352,
+      "step": 1364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.861266364445539e-06,
+      "loss": 0.9821,
+      "step": 1365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8610503881800674e-06,
+      "loss": 0.9366,
+      "step": 1366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.860834252096948e-06,
+      "loss": 0.9139,
+      "step": 1367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.860617956221559e-06,
+      "loss": 1.1078,
+      "step": 1368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.860401500579299e-06,
+      "loss": 0.8547,
+      "step": 1369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.860184885195585e-06,
+      "loss": 1.1325,
+      "step": 1370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.859968110095852e-06,
+      "loss": 0.9406,
+      "step": 1371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8597511753055537e-06,
+      "loss": 0.7187,
+      "step": 1372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8595340808501643e-06,
+      "loss": 1.2668,
+      "step": 1373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.859316826755175e-06,
+      "loss": 1.1281,
+      "step": 1374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8590994130460953e-06,
+      "loss": 0.9823,
+      "step": 1375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8588818397484562e-06,
+      "loss": 1.3906,
+      "step": 1376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8586641068878047e-06,
+      "loss": 1.1821,
+      "step": 1377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8584462144897076e-06,
+      "loss": 0.7809,
+      "step": 1378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.85822816257975e-06,
+      "loss": 0.9761,
+      "step": 1379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.858009951183536e-06,
+      "loss": 0.7976,
+      "step": 1380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.857791580326689e-06,
+      "loss": 1.0952,
+      "step": 1381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.857573050034851e-06,
+      "loss": 1.0203,
+      "step": 1382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8573543603336814e-06,
+      "loss": 0.7621,
+      "step": 1383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8571355112488594e-06,
+      "loss": 0.853,
+      "step": 1384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8569165028060836e-06,
+      "loss": 0.7916,
+      "step": 1385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.856697335031069e-06,
+      "loss": 0.7838,
+      "step": 1386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.856478007949552e-06,
+      "loss": 0.9742,
+      "step": 1387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.856258521587286e-06,
+      "loss": 0.7081,
+      "step": 1388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8560388759700437e-06,
+      "loss": 0.6295,
+      "step": 1389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8558190711236165e-06,
+      "loss": 0.846,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.855599107073814e-06,
+      "loss": 0.8066,
+      "step": 1391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8553789838464653e-06,
+      "loss": 0.8413,
+      "step": 1392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8551587014674176e-06,
+      "loss": 1.0046,
+      "step": 1393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.854938259962537e-06,
+      "loss": 1.2745,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.854717659357708e-06,
+      "loss": 1.1488,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8544968996788345e-06,
+      "loss": 1.078,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8542759809518385e-06,
+      "loss": 1.3938,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8540549032026606e-06,
+      "loss": 1.2644,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8538336664572598e-06,
+      "loss": 1.4649,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.853612270741615e-06,
+      "loss": 1.1606,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.853390716081723e-06,
+      "loss": 1.3306,
+      "step": 1401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8531690025035984e-06,
+      "loss": 0.9533,
+      "step": 1402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8529471300332764e-06,
+      "loss": 0.8077,
+      "step": 1403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8527250986968087e-06,
+      "loss": 1.0382,
+      "step": 1404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8525029085202677e-06,
+      "loss": 1.0954,
+      "step": 1405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.852280559529743e-06,
+      "loss": 0.8784,
+      "step": 1406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.852058051751343e-06,
+      "loss": 1.1747,
+      "step": 1407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.851835385211196e-06,
+      "loss": 1.0972,
+      "step": 1408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8516125599354474e-06,
+      "loss": 1.0045,
+      "step": 1409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8513895759502616e-06,
+      "loss": 1.193,
+      "step": 1410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8511664332818223e-06,
+      "loss": 1.1646,
+      "step": 1411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.850943131956332e-06,
+      "loss": 0.7612,
+      "step": 1412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8507196720000096e-06,
+      "loss": 0.7672,
+      "step": 1413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8504960534390956e-06,
+      "loss": 0.836,
+      "step": 1414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8502722762998474e-06,
+      "loss": 0.964,
+      "step": 1415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8500483406085418e-06,
+      "loss": 0.9028,
+      "step": 1416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.849824246391473e-06,
+      "loss": 0.8188,
+      "step": 1417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8495999936749553e-06,
+      "loss": 0.8371,
+      "step": 1418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8493755824853214e-06,
+      "loss": 1.5862,
+      "step": 1419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.849151012848921e-06,
+      "loss": 0.9196,
+      "step": 1420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.848926284792125e-06,
+      "loss": 0.8783,
+      "step": 1421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.84870139834132e-06,
+      "loss": 0.7698,
+      "step": 1422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8484763535229137e-06,
+      "loss": 1.1716,
+      "step": 1423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.848251150363331e-06,
+      "loss": 1.1575,
+      "step": 1424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8480257888890153e-06,
+      "loss": 1.1282,
+      "step": 1425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.84780026912643e-06,
+      "loss": 0.8266,
+      "step": 1426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.847574591102056e-06,
+      "loss": 0.6357,
+      "step": 1427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8473487548423923e-06,
+      "loss": 0.8049,
+      "step": 1428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8471227603739576e-06,
+      "loss": 1.2364,
+      "step": 1429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.846896607723288e-06,
+      "loss": 0.9172,
+      "step": 1430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8466702969169403e-06,
+      "loss": 0.7338,
+      "step": 1431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.846443827981487e-06,
+      "loss": 1.0989,
+      "step": 1432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.846217200943521e-06,
+      "loss": 1.1736,
+      "step": 1433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8459904158296544e-06,
+      "loss": 1.0712,
+      "step": 1434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.845763472666515e-06,
+      "loss": 1.0882,
+      "step": 1435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.845536371480752e-06,
+      "loss": 1.0729,
+      "step": 1436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8453091122990324e-06,
+      "loss": 1.0537,
+      "step": 1437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.845081695148041e-06,
+      "loss": 0.8953,
+      "step": 1438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8448541200544817e-06,
+      "loss": 1.2555,
+      "step": 1439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8446263870450773e-06,
+      "loss": 1.1132,
+      "step": 1440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8443984961465683e-06,
+      "loss": 0.9217,
+      "step": 1441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.844170447385714e-06,
+      "loss": 0.9865,
+      "step": 1442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.843942240789293e-06,
+      "loss": 1.1243,
+      "step": 1443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.843713876384102e-06,
+      "loss": 1.0678,
+      "step": 1444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8434853541969546e-06,
+      "loss": 0.9748,
+      "step": 1445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.843256674254686e-06,
+      "loss": 1.0298,
+      "step": 1446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8430278365841473e-06,
+      "loss": 0.9752,
+      "step": 1447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.84279884121221e-06,
+      "loss": 1.1187,
+      "step": 1448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.842569688165763e-06,
+      "loss": 1.3254,
+      "step": 1449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8423403774717128e-06,
+      "loss": 1.0475,
+      "step": 1450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8421109091569878e-06,
+      "loss": 5.8068,
+      "step": 1451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8418812832485307e-06,
+      "loss": 1.097,
+      "step": 1452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8416514997733053e-06,
+      "loss": 0.6675,
+      "step": 1453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.841421558758294e-06,
+      "loss": 0.8556,
+      "step": 1454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8411914602304967e-06,
+      "loss": 0.8473,
+      "step": 1455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8409612042169314e-06,
+      "loss": 0.9964,
+      "step": 1456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8407307907446354e-06,
+      "loss": 1.0028,
+      "step": 1457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.840500219840665e-06,
+      "loss": 0.6862,
+      "step": 1458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.840269491532095e-06,
+      "loss": 1.2108,
+      "step": 1459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.840038605846016e-06,
+      "loss": 1.1008,
+      "step": 1460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.839807562809541e-06,
+      "loss": 1.3761,
+      "step": 1461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8395763624497985e-06,
+      "loss": 1.2219,
+      "step": 1462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.839345004793937e-06,
+      "loss": 1.2252,
+      "step": 1463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.839113489869123e-06,
+      "loss": 0.7466,
+      "step": 1464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8388818177025415e-06,
+      "loss": 0.83,
+      "step": 1465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.838649988321396e-06,
+      "loss": 1.0158,
+      "step": 1466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8384180017529087e-06,
+      "loss": 0.982,
+      "step": 1467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8381858580243195e-06,
+      "loss": 0.8162,
+      "step": 1468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.837953557162887e-06,
+      "loss": 0.9714,
+      "step": 1469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8377210991958894e-06,
+      "loss": 0.9446,
+      "step": 1470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.837488484150622e-06,
+      "loss": 1.1125,
+      "step": 1471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8372557120543988e-06,
+      "loss": 0.8445,
+      "step": 1472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.837022782934553e-06,
+      "loss": 0.9898,
+      "step": 1473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.836789696818435e-06,
+      "loss": 1.2115,
+      "step": 1474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8365564537334145e-06,
+      "loss": 1.0549,
+      "step": 1475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8363230537068794e-06,
+      "loss": 0.6008,
+      "step": 1476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.836089496766237e-06,
+      "loss": 0.8543,
+      "step": 1477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8358557829389103e-06,
+      "loss": 0.8156,
+      "step": 1478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.835621912252344e-06,
+      "loss": 0.9438,
+      "step": 1479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.835387884733999e-06,
+      "loss": 0.9066,
+      "step": 1480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8351537004113558e-06,
+      "loss": 1.0932,
+      "step": 1481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8349193593119125e-06,
+      "loss": 0.976,
+      "step": 1482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.834684861463186e-06,
+      "loss": 1.1006,
+      "step": 1483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8344502068927123e-06,
+      "loss": 1.1035,
+      "step": 1484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.834215395628044e-06,
+      "loss": 0.7535,
+      "step": 1485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8339804276967538e-06,
+      "loss": 0.9751,
+      "step": 1486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8337453031264313e-06,
+      "loss": 1.0613,
+      "step": 1487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.833510021944687e-06,
+      "loss": 0.8478,
+      "step": 1488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8332745841791464e-06,
+      "loss": 0.825,
+      "step": 1489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8330389898574566e-06,
+      "loss": 1.2027,
+      "step": 1490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.832803239007281e-06,
+      "loss": 1.0121,
+      "step": 1491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.832567331656302e-06,
+      "loss": 0.8381,
+      "step": 1492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8323312678322196e-06,
+      "loss": 1.1024,
+      "step": 1493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.832095047562754e-06,
+      "loss": 0.9515,
+      "step": 1494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.831858670875643e-06,
+      "loss": 1.2391,
+      "step": 1495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8316221377986414e-06,
+      "loss": 0.6622,
+      "step": 1496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8313854483595242e-06,
+      "loss": 0.7804,
+      "step": 1497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.831148602586084e-06,
+      "loss": 0.7682,
+      "step": 1498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830911600506131e-06,
+      "loss": 1.16,
+      "step": 1499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8306744421474956e-06,
+      "loss": 1.3063,
+      "step": 1500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830437127538025e-06,
+      "loss": 0.8106,
+      "step": 1501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830199656705585e-06,
+      "loss": 0.9242,
+      "step": 1502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8299620296780603e-06,
+      "loss": 1.0045,
+      "step": 1503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8297242464833533e-06,
+      "loss": 0.9316,
+      "step": 1504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8294863071493854e-06,
+      "loss": 1.0499,
+      "step": 1505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8292482117040957e-06,
+      "loss": 1.2331,
+      "step": 1506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.829009960175442e-06,
+      "loss": 0.9895,
+      "step": 1507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8287715525914e-06,
+      "loss": 4.3708,
+      "step": 1508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8285329889799654e-06,
+      "loss": 3.997,
+      "step": 1509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.828294269369149e-06,
+      "loss": 1.1453,
+      "step": 1510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.828055393786983e-06,
+      "loss": 0.8472,
+      "step": 1511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.827816362261517e-06,
+      "loss": 1.2101,
+      "step": 1512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.827577174820817e-06,
+      "loss": 0.7871,
+      "step": 1513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8273378314929708e-06,
+      "loss": 0.8492,
+      "step": 1514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8270983323060816e-06,
+      "loss": 1.0352,
+      "step": 1515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.826858677288272e-06,
+      "loss": 0.91,
+      "step": 1516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.826618866467683e-06,
+      "loss": 0.8147,
+      "step": 1517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.826378899872474e-06,
+      "loss": 1.5598,
+      "step": 1518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8261387775308223e-06,
+      "loss": 0.9277,
+      "step": 1519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.825898499470923e-06,
+      "loss": 1.0618,
+      "step": 1520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8256580657209908e-06,
+      "loss": 1.108,
+      "step": 1521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.825417476309258e-06,
+      "loss": 0.7319,
+      "step": 1522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8251767312639745e-06,
+      "loss": 1.0143,
+      "step": 1523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.82493583061341e-06,
+      "loss": 1.1462,
+      "step": 1524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.824694774385851e-06,
+      "loss": 1.1857,
+      "step": 1525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.824453562609603e-06,
+      "loss": 1.1915,
+      "step": 1526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8242121953129896e-06,
+      "loss": 1.0094,
+      "step": 1527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8239706725243527e-06,
+      "loss": 1.1081,
+      "step": 1528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8237289942720527e-06,
+      "loss": 1.0034,
+      "step": 1529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.823487160584468e-06,
+      "loss": 0.9318,
+      "step": 1530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8232451714899946e-06,
+      "loss": 1.073,
+      "step": 1531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8230030270170486e-06,
+      "loss": 0.8153,
+      "step": 1532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8227607271940623e-06,
+      "loss": 0.8888,
+      "step": 1533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.822518272049487e-06,
+      "loss": 0.8465,
+      "step": 1534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.822275661611793e-06,
+      "loss": 0.9778,
+      "step": 1535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8220328959094676e-06,
+      "loss": 1.1103,
+      "step": 1536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8217899749710166e-06,
+      "loss": 0.8398,
+      "step": 1537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8215468988249653e-06,
+      "loss": 0.7427,
+      "step": 1538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.821303667499856e-06,
+      "loss": 1.1008,
+      "step": 1539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8210602810242486e-06,
+      "loss": 0.7864,
+      "step": 1540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8208167394267233e-06,
+      "loss": 1.1185,
+      "step": 1541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.820573042735877e-06,
+      "loss": 1.0565,
+      "step": 1542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8203291909803243e-06,
+      "loss": 1.2891,
+      "step": 1543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8200851841887e-06,
+      "loss": 1.1968,
+      "step": 1544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8198410223896547e-06,
+      "loss": 0.8648,
+      "step": 1545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8195967056118596e-06,
+      "loss": 1.0128,
+      "step": 1546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.819352233884002e-06,
+      "loss": 0.8317,
+      "step": 1547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8191076072347895e-06,
+      "loss": 0.7891,
+      "step": 1548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8188628256929453e-06,
+      "loss": 1.151,
+      "step": 1549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8186178892872128e-06,
+      "loss": 0.9798,
+      "step": 1550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8183727980463537e-06,
+      "loss": 0.893,
+      "step": 1551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.818127551999146e-06,
+      "loss": 0.8141,
+      "step": 1552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.817882151174388e-06,
+      "loss": 1.2635,
+      "step": 1553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8176365956008943e-06,
+      "loss": 1.3159,
+      "step": 1554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8173908853074994e-06,
+      "loss": 0.8533,
+      "step": 1555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.817145020323055e-06,
+      "loss": 1.0295,
+      "step": 1556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.816899000676431e-06,
+      "loss": 1.1692,
+      "step": 1557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.816652826396516e-06,
+      "loss": 0.7737,
+      "step": 1558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8164064975122152e-06,
+      "loss": 1.0096,
+      "step": 1559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8161600140524545e-06,
+      "loss": 1.1788,
+      "step": 1560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.815913376046176e-06,
+      "loss": 0.7865,
+      "step": 1561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8156665835223394e-06,
+      "loss": 1.4645,
+      "step": 1562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8154196365099255e-06,
+      "loss": 0.779,
+      "step": 1563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8151725350379303e-06,
+      "loss": 0.9827,
+      "step": 1564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8149252791353693e-06,
+      "loss": 0.878,
+      "step": 1565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8146778688312757e-06,
+      "loss": 0.7525,
+      "step": 1566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.814430304154701e-06,
+      "loss": 1.0199,
+      "step": 1567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8141825851347154e-06,
+      "loss": 1.0762,
+      "step": 1568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8139347118004056e-06,
+      "loss": 1.0751,
+      "step": 1569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.813686684180878e-06,
+      "loss": 0.8735,
+      "step": 1570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.813438502305257e-06,
+      "loss": 1.0184,
+      "step": 1571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.813190166202684e-06,
+      "loss": 0.9839,
+      "step": 1572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8129416759023195e-06,
+      "loss": 0.9603,
+      "step": 1573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8126930314333416e-06,
+      "loss": 0.9204,
+      "step": 1574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.812444232824947e-06,
+      "loss": 0.9971,
+      "step": 1575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.81219528010635e-06,
+      "loss": 0.8384,
+      "step": 1576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.811946173306783e-06,
+      "loss": 1.0625,
+      "step": 1577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.811696912455497e-06,
+      "loss": 1.2776,
+      "step": 1578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.811447497581761e-06,
+      "loss": 1.2727,
+      "step": 1579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8111979287148615e-06,
+      "loss": 0.695,
+      "step": 1580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8109482058841032e-06,
+      "loss": 0.7486,
+      "step": 1581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8106983291188097e-06,
+      "loss": 0.8528,
+      "step": 1582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8104482984483217e-06,
+      "loss": 0.8875,
+      "step": 1583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8101981139019987e-06,
+      "loss": 1.2364,
+      "step": 1584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8099477755092173e-06,
+      "loss": 0.4561,
+      "step": 1585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.809697283299374e-06,
+      "loss": 1.1597,
+      "step": 1586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8094466373018807e-06,
+      "loss": 0.9985,
+      "step": 1587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.80919583754617e-06,
+      "loss": 1.2715,
+      "step": 1588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8089448840616908e-06,
+      "loss": 1.1516,
+      "step": 1589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.808693776877911e-06,
+      "loss": 0.8472,
+      "step": 1590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8084425160243155e-06,
+      "loss": 0.7923,
+      "step": 1591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8081911015304085e-06,
+      "loss": 1.0609,
+      "step": 1592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8079395334257115e-06,
+      "loss": 1.264,
+      "step": 1593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.807687811739765e-06,
+      "loss": 0.971,
+      "step": 1594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.807435936502125e-06,
+      "loss": 1.0774,
+      "step": 1595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8071839077423686e-06,
+      "loss": 0.9008,
+      "step": 1596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.806931725490089e-06,
+      "loss": 0.9806,
+      "step": 1597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8066793897748986e-06,
+      "loss": 0.909,
+      "step": 1598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8064269006264267e-06,
+      "loss": 0.9031,
+      "step": 1599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8061742580743216e-06,
+      "loss": 0.7706,
+      "step": 1600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.805921462148249e-06,
+      "loss": 0.9147,
+      "step": 1601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.805668512877892e-06,
+      "loss": 1.1784,
+      "step": 1602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8054154102929536e-06,
+      "loss": 0.7485,
+      "step": 1603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.805162154423153e-06,
+      "loss": 1.0183,
+      "step": 1604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8049087452982287e-06,
+      "loss": 0.972,
+      "step": 1605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.804655182947936e-06,
+      "loss": 0.7604,
+      "step": 1606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.804401467402049e-06,
+      "loss": 1.3831,
+      "step": 1607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8041475986903594e-06,
+      "loss": 1.0214,
+      "step": 1608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.803893576842678e-06,
+      "loss": 0.8725,
+      "step": 1609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.803639401888831e-06,
+      "loss": 1.0248,
+      "step": 1610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.803385073858665e-06,
+      "loss": 0.7826,
+      "step": 1611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8031305927820442e-06,
+      "loss": 0.891,
+      "step": 1612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8028759586888497e-06,
+      "loss": 1.1391,
+      "step": 1613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.802621171608981e-06,
+      "loss": 0.9124,
+      "step": 1614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.802366231572357e-06,
+      "loss": 0.7872,
+      "step": 1615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8021111386089123e-06,
+      "loss": 1.0316,
+      "step": 1616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.801855892748601e-06,
+      "loss": 1.3106,
+      "step": 1617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8016004940213942e-06,
+      "loss": 0.92,
+      "step": 1618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8013449424572817e-06,
+      "loss": 1.0113,
+      "step": 1619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8010892380862706e-06,
+      "loss": 1.1241,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8008333809383866e-06,
+      "loss": 1.2412,
+      "step": 1621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8005773710436735e-06,
+      "loss": 1.0896,
+      "step": 1622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8003212084321917e-06,
+      "loss": 0.9571,
+      "step": 1623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.800064893134021e-06,
+      "loss": 0.951,
+      "step": 1624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7998084251792583e-06,
+      "loss": 0.8304,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7995518045980183e-06,
+      "loss": 1.004,
+      "step": 1626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.799295031420435e-06,
+      "loss": 1.3834,
+      "step": 1627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7990381056766585e-06,
+      "loss": 0.8884,
+      "step": 1628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.798781027396857e-06,
+      "loss": 0.946,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7985237966112184e-06,
+      "loss": 1.1809,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.798266413349947e-06,
+      "loss": 1.1244,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.798008877643265e-06,
+      "loss": 0.8263,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7977511895214136e-06,
+      "loss": 0.9677,
+      "step": 1633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7974933490146497e-06,
+      "loss": 0.9647,
+      "step": 1634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7972353561532507e-06,
+      "loss": 1.233,
+      "step": 1635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.796977210967511e-06,
+      "loss": 1.0374,
+      "step": 1636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.796718913487741e-06,
+      "loss": 1.123,
+      "step": 1637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7964604637442727e-06,
+      "loss": 1.1527,
+      "step": 1638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.796201861767452e-06,
+      "loss": 0.7956,
+      "step": 1639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.795943107587645e-06,
+      "loss": 0.8825,
+      "step": 1640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7956842012352362e-06,
+      "loss": 0.9789,
+      "step": 1641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.795425142740626e-06,
+      "loss": 0.9239,
+      "step": 1642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.795165932134234e-06,
+      "loss": 0.8772,
+      "step": 1643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7949065694464966e-06,
+      "loss": 1.1812,
+      "step": 1644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.79464705470787e-06,
+      "loss": 0.9869,
+      "step": 1645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.794387387948827e-06,
+      "loss": 0.9445,
+      "step": 1646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7941275691998566e-06,
+      "loss": 0.7572,
+      "step": 1647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7938675984914685e-06,
+      "loss": 0.8018,
+      "step": 1648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.793607475854189e-06,
+      "loss": 0.8999,
+      "step": 1649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7933472013185623e-06,
+      "loss": 0.6102,
+      "step": 1650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.79308677491515e-06,
+      "loss": 1.2503,
+      "step": 1651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.792826196674533e-06,
+      "loss": 0.7011,
+      "step": 1652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7925654666273076e-06,
+      "loss": 1.0235,
+      "step": 1653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7923045848040905e-06,
+      "loss": 0.7841,
+      "step": 1654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7920435512355144e-06,
+      "loss": 1.3398,
+      "step": 1655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.79178236595223e-06,
+      "loss": 0.6116,
+      "step": 1656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.791521028984908e-06,
+      "loss": 1.2239,
+      "step": 1657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7912595403642335e-06,
+      "loss": 0.8026,
+      "step": 1658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.790997900120912e-06,
+      "loss": 0.9876,
+      "step": 1659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7907361082856648e-06,
+      "loss": 1.1478,
+      "step": 1660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.790474164889233e-06,
+      "loss": 0.8213,
+      "step": 1661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7902120699623747e-06,
+      "loss": 1.0713,
+      "step": 1662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7899498235358654e-06,
+      "loss": 1.0332,
+      "step": 1663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.789687425640499e-06,
+      "loss": 1.0135,
+      "step": 1664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7894248763070862e-06,
+      "loss": 0.8572,
+      "step": 1665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7891621755664567e-06,
+      "loss": 0.8719,
+      "step": 1666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.788899323449457e-06,
+      "loss": 1.004,
+      "step": 1667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7886363199869526e-06,
+      "loss": 0.9356,
+      "step": 1668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7883731652098247e-06,
+      "loss": 1.011,
+      "step": 1669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.788109859148975e-06,
+      "loss": 0.7754,
+      "step": 1670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.78784640183532e-06,
+      "loss": 1.0254,
+      "step": 1671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7875827932997962e-06,
+      "loss": 1.0442,
+      "step": 1672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7873190335733573e-06,
+      "loss": 0.8948,
+      "step": 1673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7870551226869746e-06,
+      "loss": 0.7348,
+      "step": 1674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.786791060671637e-06,
+      "loss": 1.193,
+      "step": 1675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7865268475583506e-06,
+      "loss": 0.752,
+      "step": 1676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7862624833781404e-06,
+      "loss": 1.0418,
+      "step": 1677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.785997968162049e-06,
+      "loss": 0.7823,
+      "step": 1678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7857333019411365e-06,
+      "loss": 1.0183,
+      "step": 1679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7854684847464794e-06,
+      "loss": 0.8226,
+      "step": 1680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7852035166091743e-06,
+      "loss": 1.0641,
+      "step": 1681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7849383975603344e-06,
+      "loss": 0.5058,
+      "step": 1682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7846731276310898e-06,
+      "loss": 1.1609,
+      "step": 1683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7844077068525897e-06,
+      "loss": 1.1131,
+      "step": 1684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7841421352560003e-06,
+      "loss": 1.1396,
+      "step": 1685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.783876412872505e-06,
+      "loss": 0.766,
+      "step": 1686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7836105397333066e-06,
+      "loss": 0.8691,
+      "step": 1687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7833445158696236e-06,
+      "loss": 1.2389,
+      "step": 1688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.783078341312694e-06,
+      "loss": 1.2497,
+      "step": 1689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.782812016093772e-06,
+      "loss": 0.9751,
+      "step": 1690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7825455402441298e-06,
+      "loss": 0.7084,
+      "step": 1691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7822789137950587e-06,
+      "loss": 0.9633,
+      "step": 1692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7820121367778653e-06,
+      "loss": 0.7067,
+      "step": 1693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7817452092238763e-06,
+      "loss": 0.8895,
+      "step": 1694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7814781311644344e-06,
+      "loss": 1.1589,
+      "step": 1695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7812109026309007e-06,
+      "loss": 1.1057,
+      "step": 1696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7809435236546535e-06,
+      "loss": 0.7408,
+      "step": 1697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7806759942670893e-06,
+      "loss": 0.9337,
+      "step": 1698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.780408314499622e-06,
+      "loss": 0.8197,
+      "step": 1699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7801404843836832e-06,
+      "loss": 1.1266,
+      "step": 1700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7798725039507214e-06,
+      "loss": 0.7201,
+      "step": 1701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.779604373232205e-06,
+      "loss": 0.9494,
+      "step": 1702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.779336092259617e-06,
+      "loss": 1.2628,
+      "step": 1703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.77906766106446e-06,
+      "loss": 1.127,
+      "step": 1704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.778799079678255e-06,
+      "loss": 1.261,
+      "step": 1705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.778530348132538e-06,
+      "loss": 0.7737,
+      "step": 1706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.778261466458864e-06,
+      "loss": 0.8413,
+      "step": 1707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7779924346888066e-06,
+      "loss": 0.8495,
+      "step": 1708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7777232528539555e-06,
+      "loss": 0.8209,
+      "step": 1709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7774539209859195e-06,
+      "loss": 0.9472,
+      "step": 1710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.777184439116323e-06,
+      "loss": 0.9668,
+      "step": 1711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.77691480727681e-06,
+      "loss": 0.6263,
+      "step": 1712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7766450254990404e-06,
+      "loss": 1.0582,
+      "step": 1713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.776375093814694e-06,
+      "loss": 1.1913,
+      "step": 1714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7761050122554655e-06,
+      "loss": 1.0014,
+      "step": 1715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7758347808530695e-06,
+      "loss": 0.8679,
+      "step": 1716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.775564399639236e-06,
+      "loss": 1.1106,
+      "step": 1717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7752938686457154e-06,
+      "loss": 1.0736,
+      "step": 1718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.775023187904272e-06,
+      "loss": 1.0913,
+      "step": 1719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7747523574466917e-06,
+      "loss": 1.1649,
+      "step": 1720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.774481377304775e-06,
+      "loss": 0.8655,
+      "step": 1721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7742102475103412e-06,
+      "loss": 1.1786,
+      "step": 1722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.773938968095227e-06,
+      "loss": 1.1442,
+      "step": 1723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7736675390912868e-06,
+      "loss": 1.0197,
+      "step": 1724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.773395960530392e-06,
+      "loss": 1.3805,
+      "step": 1725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7731242324444322e-06,
+      "loss": 0.8954,
+      "step": 1726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7728523548653144e-06,
+      "loss": 1.1246,
+      "step": 1727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7725803278249633e-06,
+      "loss": 0.949,
+      "step": 1728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7723081513553207e-06,
+      "loss": 1.0645,
+      "step": 1729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.772035825488346e-06,
+      "loss": 0.8046,
+      "step": 1730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7717633502560167e-06,
+      "loss": 0.765,
+      "step": 1731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.771490725690327e-06,
+      "loss": 0.756,
+      "step": 1732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7712179518232896e-06,
+      "loss": 1.2065,
+      "step": 1733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.770945028686934e-06,
+      "loss": 1.1483,
+      "step": 1734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7706719563133072e-06,
+      "loss": 0.9729,
+      "step": 1735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.770398734734474e-06,
+      "loss": 0.9017,
+      "step": 1736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.770125363982518e-06,
+      "loss": 1.0909,
+      "step": 1737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7698518440895375e-06,
+      "loss": 0.8115,
+      "step": 1738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.76957817508765e-06,
+      "loss": 0.9085,
+      "step": 1739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7693043570089908e-06,
+      "loss": 0.9644,
+      "step": 1740
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.256152868270874,
+      "eval_runtime": 32.1734,
+      "eval_samples_per_second": 38.697,
+      "eval_steps_per_second": 19.364,
+      "step": 1740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.769030389885712e-06,
+      "loss": 0.9486,
+      "step": 1741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7687562737499835e-06,
+      "loss": 1.4981,
+      "step": 1742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.768482008633993e-06,
+      "loss": 1.0327,
+      "step": 1743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7682075945699446e-06,
+      "loss": 1.1069,
+      "step": 1744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7679330315900616e-06,
+      "loss": 1.3067,
+      "step": 1745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.767658319726582e-06,
+      "loss": 0.9682,
+      "step": 1746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.767383459011766e-06,
+      "loss": 0.9697,
+      "step": 1747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7671084494778856e-06,
+      "loss": 0.716,
+      "step": 1748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.766833291157235e-06,
+      "loss": 0.8137,
+      "step": 1749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7665579840821223e-06,
+      "loss": 1.114,
+      "step": 1750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7662825282848756e-06,
+      "loss": 1.3975,
+      "step": 1751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7660069237978395e-06,
+      "loss": 0.7992,
+      "step": 1752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7657311706533765e-06,
+      "loss": 0.9781,
+      "step": 1753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7654552688838653e-06,
+      "loss": 1.0528,
+      "step": 1754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7651792185217033e-06,
+      "loss": 0.6321,
+      "step": 1755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7649030195993046e-06,
+      "loss": 1.223,
+      "step": 1756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764626672149102e-06,
+      "loss": 0.7618,
+      "step": 1757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764350176203544e-06,
+      "loss": 1.1135,
+      "step": 1758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764073531795098e-06,
+      "loss": 0.9138,
+      "step": 1759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7637967389562476e-06,
+      "loss": 0.9548,
+      "step": 1760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7635197977194952e-06,
+      "loss": 0.8678,
+      "step": 1761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7632427081173594e-06,
+      "loss": 0.9922,
+      "step": 1762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762965470182377e-06,
+      "loss": 1.0087,
+      "step": 1763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7626880839471013e-06,
+      "loss": 1.032,
+      "step": 1764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7624105494441037e-06,
+      "loss": 1.0674,
+      "step": 1765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762132866705974e-06,
+      "loss": 1.1482,
+      "step": 1766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7618550357653165e-06,
+      "loss": 1.3429,
+      "step": 1767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7615770566547566e-06,
+      "loss": 0.9124,
+      "step": 1768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761298929406934e-06,
+      "loss": 1.0739,
+      "step": 1769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761020654054508e-06,
+      "loss": 0.8846,
+      "step": 1770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7607422306301537e-06,
+      "loss": 1.1748,
+      "step": 1771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760463659166564e-06,
+      "loss": 0.9312,
+      "step": 1772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7601849396964503e-06,
+      "loss": 0.9731,
+      "step": 1773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7599060722525395e-06,
+      "loss": 0.8547,
+      "step": 1774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7596270568675774e-06,
+      "loss": 0.9153,
+      "step": 1775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7593478935743264e-06,
+      "loss": 1.104,
+      "step": 1776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7590685824055667e-06,
+      "loss": 1.0985,
+      "step": 1777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7587891233940954e-06,
+      "loss": 1.0755,
+      "step": 1778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7585095165727274e-06,
+      "loss": 1.022,
+      "step": 1779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7582297619742945e-06,
+      "loss": 1.2668,
+      "step": 1780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.757949859631647e-06,
+      "loss": 0.4957,
+      "step": 1781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75766980957765e-06,
+      "loss": 1.1962,
+      "step": 1782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7573896118451896e-06,
+      "loss": 0.9233,
+      "step": 1783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.757109266467166e-06,
+      "loss": 0.6768,
+      "step": 1784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7568287734764985e-06,
+      "loss": 1.0978,
+      "step": 1785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7565481329061223e-06,
+      "loss": 0.8083,
+      "step": 1786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7562673447889924e-06,
+      "loss": 0.8803,
+      "step": 1787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755986409158078e-06,
+      "loss": 0.8858,
+      "step": 1788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7557053260463685e-06,
+      "loss": 1.0541,
+      "step": 1789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7554240954868688e-06,
+      "loss": 1.0821,
+      "step": 1790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7551427175126014e-06,
+      "loss": 0.6321,
+      "step": 1791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7548611921566066e-06,
+      "loss": 0.5906,
+      "step": 1792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.754579519451942e-06,
+      "loss": 0.6906,
+      "step": 1793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7542976994316817e-06,
+      "loss": 1.3772,
+      "step": 1794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7540157321289177e-06,
+      "loss": 1.0565,
+      "step": 1795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.75373361757676e-06,
+      "loss": 1.0466,
+      "step": 1796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7534513558083344e-06,
+      "loss": 0.6046,
+      "step": 1797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7531689468567844e-06,
+      "loss": 1.2846,
+      "step": 1798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7528863907552725e-06,
+      "loss": 1.0251,
+      "step": 1799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7526036875369757e-06,
+      "loss": 0.7412,
+      "step": 1800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.75232083723509e-06,
+      "loss": 1.172,
+      "step": 1801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.752037839882829e-06,
+      "loss": 1.1497,
+      "step": 1802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7517546955134224e-06,
+      "loss": 0.7238,
+      "step": 1803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7514714041601175e-06,
+      "loss": 0.9286,
+      "step": 1804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.751187965856179e-06,
+      "loss": 1.1538,
+      "step": 1805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.750904380634889e-06,
+      "loss": 0.9193,
+      "step": 1806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.750620648529547e-06,
+      "loss": 1.1063,
+      "step": 1807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7503367695734693e-06,
+      "loss": 0.6603,
+      "step": 1808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.750052743799989e-06,
+      "loss": 0.9343,
+      "step": 1809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7497685712424585e-06,
+      "loss": 1.006,
+      "step": 1810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.749484251934245e-06,
+      "loss": 0.867,
+      "step": 1811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7491997859087336e-06,
+      "loss": 0.8108,
+      "step": 1812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7489151731993275e-06,
+      "loss": 1.0218,
+      "step": 1813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7486304138394464e-06,
+      "loss": 1.0602,
+      "step": 1814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7483455078625273e-06,
+      "loss": 1.0479,
+      "step": 1815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.748060455302025e-06,
+      "loss": 1.054,
+      "step": 1816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7477752561914106e-06,
+      "loss": 0.9799,
+      "step": 1817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.747489910564173e-06,
+      "loss": 0.7012,
+      "step": 1818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.747204418453818e-06,
+      "loss": 0.8776,
+      "step": 1819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7469187798938687e-06,
+      "loss": 0.681,
+      "step": 1820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7466329949178654e-06,
+      "loss": 0.8585,
+      "step": 1821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.746347063559366e-06,
+      "loss": 0.9953,
+      "step": 1822
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7460609858519454e-06,
+      "loss": 1.247,
+      "step": 1823
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7457747618291943e-06,
+      "loss": 1.1763,
+      "step": 1824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7454883915247222e-06,
+      "loss": 0.8952,
+      "step": 1825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.745201874972157e-06,
+      "loss": 0.8883,
+      "step": 1826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.74491521220514e-06,
+      "loss": 0.9288,
+      "step": 1827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7446284032573326e-06,
+      "loss": 0.9064,
+      "step": 1828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7443414481624128e-06,
+      "loss": 0.7462,
+      "step": 1829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7440543469540754e-06,
+      "loss": 0.7953,
+      "step": 1830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7437670996660322e-06,
+      "loss": 0.9947,
+      "step": 1831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7434797063320127e-06,
+      "loss": 0.8877,
+      "step": 1832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.743192166985763e-06,
+      "loss": 0.9243,
+      "step": 1833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.742904481661047e-06,
+      "loss": 0.6944,
+      "step": 1834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7426166503916454e-06,
+      "loss": 0.7735,
+      "step": 1835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.742328673211355e-06,
+      "loss": 0.7539,
+      "step": 1836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7420405501539924e-06,
+      "loss": 0.8782,
+      "step": 1837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.741752281253389e-06,
+      "loss": 1.1026,
+      "step": 1838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.741463866543393e-06,
+      "loss": 0.799,
+      "step": 1839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.741175306057872e-06,
+      "loss": 1.1587,
+      "step": 1840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.740886599830709e-06,
+      "loss": 0.9869,
+      "step": 1841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7405977478958043e-06,
+      "loss": 1.0784,
+      "step": 1842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7403087502870756e-06,
+      "loss": 0.8938,
+      "step": 1843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7400196070384584e-06,
+      "loss": 0.9266,
+      "step": 1844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7397303181839034e-06,
+      "loss": 0.8174,
+      "step": 1845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.73944088375738e-06,
+      "loss": 0.9389,
+      "step": 1846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.739151303792875e-06,
+      "loss": 0.8774,
+      "step": 1847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7388615783243907e-06,
+      "loss": 0.9094,
+      "step": 1848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7385717073859475e-06,
+      "loss": 0.9895,
+      "step": 1849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.738281691011583e-06,
+      "loss": 0.9573,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.737991529235351e-06,
+      "loss": 0.7659,
+      "step": 1851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7377012220913233e-06,
+      "loss": 0.8332,
+      "step": 1852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.737410769613589e-06,
+      "loss": 1.1836,
+      "step": 1853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7371201718362524e-06,
+      "loss": 1.0569,
+      "step": 1854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.736829428793437e-06,
+      "loss": 0.7988,
+      "step": 1855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7365385405192825e-06,
+      "loss": 0.8209,
+      "step": 1856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.736247507047946e-06,
+      "loss": 1.5295,
+      "step": 1857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7359563284136004e-06,
+      "loss": 1.0402,
+      "step": 1858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.735665004650437e-06,
+      "loss": 0.883,
+      "step": 1859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.735373535792664e-06,
+      "loss": 1.096,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.735081921874506e-06,
+      "loss": 0.7991,
+      "step": 1861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7347901629302054e-06,
+      "loss": 1.2145,
+      "step": 1862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.73449825899402e-06,
+      "loss": 0.6532,
+      "step": 1863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7342062101002276e-06,
+      "loss": 1.0581,
+      "step": 1864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7339140162831197e-06,
+      "loss": 0.9254,
+      "step": 1865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.733621677577007e-06,
+      "loss": 1.0395,
+      "step": 1866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7333291940162173e-06,
+      "loss": 1.1302,
+      "step": 1867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7330365656350933e-06,
+      "loss": 1.2696,
+      "step": 1868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.732743792467997e-06,
+      "loss": 0.9898,
+      "step": 1869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7324508745493064e-06,
+      "loss": 1.0142,
+      "step": 1870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.732157811913416e-06,
+      "loss": 0.849,
+      "step": 1871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7318646045947387e-06,
+      "loss": 0.9378,
+      "step": 1872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7315712526277032e-06,
+      "loss": 1.0235,
+      "step": 1873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.731277756046755e-06,
+      "loss": 1.0164,
+      "step": 1874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.730984114886358e-06,
+      "loss": 0.9755,
+      "step": 1875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7306903291809922e-06,
+      "loss": 1.3032,
+      "step": 1876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7303963989651547e-06,
+      "loss": 1.2517,
+      "step": 1877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.730102324273358e-06,
+      "loss": 1.0657,
+      "step": 1878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.729808105140135e-06,
+      "loss": 1.167,
+      "step": 1879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7295137416000323e-06,
+      "loss": 0.8718,
+      "step": 1880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7292192336876153e-06,
+      "loss": 0.944,
+      "step": 1881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7289245814374654e-06,
+      "loss": 1.0171,
+      "step": 1882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7286297848841813e-06,
+      "loss": 1.1082,
+      "step": 1883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7283348440623793e-06,
+      "loss": 0.7503,
+      "step": 1884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.728039759006692e-06,
+      "loss": 0.9794,
+      "step": 1885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.727744529751768e-06,
+      "loss": 1.0917,
+      "step": 1886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.727449156332274e-06,
+      "loss": 0.7873,
+      "step": 1887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.727153638782895e-06,
+      "loss": 0.8422,
+      "step": 1888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7268579771383294e-06,
+      "loss": 1.0037,
+      "step": 1889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.726562171433296e-06,
+      "loss": 0.8631,
+      "step": 1890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7262662217025272e-06,
+      "loss": 0.8086,
+      "step": 1891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7259701279807756e-06,
+      "loss": 1.2279,
+      "step": 1892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7256738903028087e-06,
+      "loss": 1.3531,
+      "step": 1893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7253775087034117e-06,
+      "loss": 1.0015,
+      "step": 1894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7250809832173856e-06,
+      "loss": 0.7056,
+      "step": 1895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7247843138795504e-06,
+      "loss": 0.7337,
+      "step": 1896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7244875007247403e-06,
+      "loss": 0.9497,
+      "step": 1897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.724190543787809e-06,
+      "loss": 1.2138,
+      "step": 1898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7238934431036247e-06,
+      "loss": 0.9988,
+      "step": 1899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.723596198707074e-06,
+      "loss": 1.0154,
+      "step": 1900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.723298810633061e-06,
+      "loss": 1.0931,
+      "step": 1901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7230012789165047e-06,
+      "loss": 1.4181,
+      "step": 1902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.722703603592342e-06,
+      "loss": 0.9521,
+      "step": 1903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7224057846955265e-06,
+      "loss": 0.7826,
+      "step": 1904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7221078222610297e-06,
+      "loss": 1.0808,
+      "step": 1905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.721809716323838e-06,
+      "loss": 1.0451,
+      "step": 1906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7215114669189558e-06,
+      "loss": 0.9624,
+      "step": 1907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.721213074081405e-06,
+      "loss": 1.0258,
+      "step": 1908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.720914537846223e-06,
+      "loss": 1.032,
+      "step": 1909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7206158582484644e-06,
+      "loss": 0.652,
+      "step": 1910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7203170353232016e-06,
+      "loss": 1.0965,
+      "step": 1911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.720018069105522e-06,
+      "loss": 1.3741,
+      "step": 1912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.719718959630532e-06,
+      "loss": 0.7909,
+      "step": 1913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7194197069333527e-06,
+      "loss": 1.2275,
+      "step": 1914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.719120311049124e-06,
+      "loss": 0.7889,
+      "step": 1915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.718820772013001e-06,
+      "loss": 0.7991,
+      "step": 1916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7185210898601565e-06,
+      "loss": 0.8762,
+      "step": 1917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.71822126462578e-06,
+      "loss": 1.0042,
+      "step": 1918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7179212963450774e-06,
+      "loss": 1.1289,
+      "step": 1919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7176211850532717e-06,
+      "loss": 0.7883,
+      "step": 1920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7173209307856024e-06,
+      "loss": 1.0623,
+      "step": 1921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7170205335773265e-06,
+      "loss": 1.1571,
+      "step": 1922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7167199934637173e-06,
+      "loss": 1.0592,
+      "step": 1923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7164193104800647e-06,
+      "loss": 1.1184,
+      "step": 1924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7161184846616753e-06,
+      "loss": 0.9249,
+      "step": 1925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7158175160438733e-06,
+      "loss": 1.2249,
+      "step": 1926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7155164046619997e-06,
+      "loss": 0.8732,
+      "step": 1927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.71521515055141e-06,
+      "loss": 0.8809,
+      "step": 1928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.714913753747479e-06,
+      "loss": 1.3511,
+      "step": 1929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7146122142855978e-06,
+      "loss": 0.8619,
+      "step": 1930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.714310532201174e-06,
+      "loss": 1.0753,
+      "step": 1931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7140087075296308e-06,
+      "loss": 0.985,
+      "step": 1932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7137067403064098e-06,
+      "loss": 1.0868,
+      "step": 1933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.713404630566968e-06,
+      "loss": 0.8237,
+      "step": 1934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.713102378346781e-06,
+      "loss": 1.4431,
+      "step": 1935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7127999836813395e-06,
+      "loss": 0.6948,
+      "step": 1936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.712497446606151e-06,
+      "loss": 0.8903,
+      "step": 1937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.71219476715674e-06,
+      "loss": 0.7675,
+      "step": 1938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.711891945368648e-06,
+      "loss": 0.8686,
+      "step": 1939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.711588981277433e-06,
+      "loss": 0.746,
+      "step": 1940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.71128587491867e-06,
+      "loss": 1.0857,
+      "step": 1941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.710982626327951e-06,
+      "loss": 0.9255,
+      "step": 1942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7106792355408825e-06,
+      "loss": 0.9048,
+      "step": 1943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.710375702593091e-06,
+      "loss": 0.8696,
+      "step": 1944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7100720275202167e-06,
+      "loss": 0.6697,
+      "step": 1945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.709768210357919e-06,
+      "loss": 0.929,
+      "step": 1946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7094642511418718e-06,
+      "loss": 1.0113,
+      "step": 1947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.709160149907767e-06,
+      "loss": 1.1814,
+      "step": 1948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.708855906691313e-06,
+      "loss": 1.065,
+      "step": 1949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.708551521528235e-06,
+      "loss": 1.0298,
+      "step": 1950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.708246994454274e-06,
+      "loss": 0.6606,
+      "step": 1951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.707942325505189e-06,
+      "loss": 1.149,
+      "step": 1952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7076375147167547e-06,
+      "loss": 0.8169,
+      "step": 1953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.707332562124762e-06,
+      "loss": 0.9538,
+      "step": 1954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7070274677650197e-06,
+      "loss": 0.9312,
+      "step": 1955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7067222316733528e-06,
+      "loss": 0.8729,
+      "step": 1956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7064168538856027e-06,
+      "loss": 1.2598,
+      "step": 1957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.706111334437627e-06,
+      "loss": 0.6996,
+      "step": 1958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7058056733653013e-06,
+      "loss": 0.9727,
+      "step": 1959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7054998707045166e-06,
+      "loss": 0.8095,
+      "step": 1960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.705193926491181e-06,
+      "loss": 1.2524,
+      "step": 1961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7048878407612193e-06,
+      "loss": 0.9796,
+      "step": 1962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7045816135505727e-06,
+      "loss": 1.0064,
+      "step": 1963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.704275244895199e-06,
+      "loss": 0.8951,
+      "step": 1964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.703968734831073e-06,
+      "loss": 1.1799,
+      "step": 1965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.703662083394185e-06,
+      "loss": 0.9532,
+      "step": 1966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.703355290620544e-06,
+      "loss": 0.8608,
+      "step": 1967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7030483565461735e-06,
+      "loss": 0.9227,
+      "step": 1968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.702741281207114e-06,
+      "loss": 0.8588,
+      "step": 1969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.702434064639424e-06,
+      "loss": 0.7552,
+      "step": 1970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.702126706879177e-06,
+      "loss": 1.0536,
+      "step": 1971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.701819207962464e-06,
+      "loss": 0.8168,
+      "step": 1972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.701511567925392e-06,
+      "loss": 1.0226,
+      "step": 1973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.701203786804084e-06,
+      "loss": 0.9292,
+      "step": 1974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7008958646346823e-06,
+      "loss": 0.9661,
+      "step": 1975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7005878014533415e-06,
+      "loss": 1.5104,
+      "step": 1976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.700279597296237e-06,
+      "loss": 0.9093,
+      "step": 1977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6999712521995575e-06,
+      "loss": 1.1781,
+      "step": 1978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6996627661995106e-06,
+      "loss": 0.985,
+      "step": 1979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.699354139332319e-06,
+      "loss": 0.8081,
+      "step": 1980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.699045371634222e-06,
+      "loss": 0.873,
+      "step": 1981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.698736463141477e-06,
+      "loss": 0.7166,
+      "step": 1982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.698427413890355e-06,
+      "loss": 0.845,
+      "step": 1983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6981182239171463e-06,
+      "loss": 1.2687,
+      "step": 1984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.697808893258157e-06,
+      "loss": 0.8495,
+      "step": 1985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.697499421949709e-06,
+      "loss": 0.9271,
+      "step": 1986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6971898100281404e-06,
+      "loss": 0.7909,
+      "step": 1987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.696880057529808e-06,
+      "loss": 1.0804,
+      "step": 1988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.696570164491083e-06,
+      "loss": 1.0022,
+      "step": 1989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.696260130948353e-06,
+      "loss": 0.7756,
+      "step": 1990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.695949956938024e-06,
+      "loss": 1.0455,
+      "step": 1991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6956396424965164e-06,
+      "loss": 1.0787,
+      "step": 1992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.695329187660269e-06,
+      "loss": 0.9612,
+      "step": 1993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6950185924657353e-06,
+      "loss": 0.9604,
+      "step": 1994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.694707856949386e-06,
+      "loss": 1.0158,
+      "step": 1995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6943969811477098e-06,
+      "loss": 1.0468,
+      "step": 1996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6940859650972084e-06,
+      "loss": 0.721,
+      "step": 1997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6937748088344037e-06,
+      "loss": 0.9034,
+      "step": 1998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.693463512395831e-06,
+      "loss": 0.8115,
+      "step": 1999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6931520758180447e-06,
+      "loss": 1.0163,
+      "step": 2000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6928404991376136e-06,
+      "loss": 0.5858,
+      "step": 2001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6925287823911243e-06,
+      "loss": 0.9527,
+      "step": 2002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6922169256151786e-06,
+      "loss": 1.3955,
+      "step": 2003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6919049288463965e-06,
+      "loss": 1.107,
+      "step": 2004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.691592792121412e-06,
+      "loss": 0.8169,
+      "step": 2005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.691280515476878e-06,
+      "loss": 1.0497,
+      "step": 2006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6909680989494615e-06,
+      "loss": 1.0604,
+      "step": 2007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.690655542575849e-06,
+      "loss": 0.6519,
+      "step": 2008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.69034284639274e-06,
+      "loss": 1.2187,
+      "step": 2009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.690030010436853e-06,
+      "loss": 1.1361,
+      "step": 2010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.689717034744921e-06,
+      "loss": 0.9057,
+      "step": 2011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.689403919353695e-06,
+      "loss": 1.1232,
+      "step": 2012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.689090664299942e-06,
+      "loss": 1.0093,
+      "step": 2013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.688777269620444e-06,
+      "loss": 1.1549,
+      "step": 2014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.688463735352002e-06,
+      "loss": 0.9776,
+      "step": 2015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.68815006153143e-06,
+      "loss": 0.873,
+      "step": 2016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6878362481955622e-06,
+      "loss": 0.8479,
+      "step": 2017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6875222953812463e-06,
+      "loss": 1.1072,
+      "step": 2018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6872082031253474e-06,
+      "loss": 0.985,
+      "step": 2019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6868939714647474e-06,
+      "loss": 1.1142,
+      "step": 2020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6865796004363435e-06,
+      "loss": 0.8956,
+      "step": 2021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.68626509007705e-06,
+      "loss": 0.9482,
+      "step": 2022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6859504404237977e-06,
+      "loss": 0.6568,
+      "step": 2023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6856356515135332e-06,
+      "loss": 0.8457,
+      "step": 2024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6853207233832204e-06,
+      "loss": 0.7659,
+      "step": 2025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6850056560698378e-06,
+      "loss": 0.9364,
+      "step": 2026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6846904496103816e-06,
+      "loss": 0.9814,
+      "step": 2027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.684375104041865e-06,
+      "loss": 0.8315,
+      "step": 2028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6840596194013157e-06,
+      "loss": 1.1382,
+      "step": 2029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6837439957257788e-06,
+      "loss": 1.0329,
+      "step": 2030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6834282330523155e-06,
+      "loss": 0.5192,
+      "step": 2031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6831123314180037e-06,
+      "loss": 1.0404,
+      "step": 2032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6827962908599373e-06,
+      "loss": 0.872,
+      "step": 2033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.682480111415226e-06,
+      "loss": 0.9305,
+      "step": 2034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6821637931209967e-06,
+      "loss": 1.0438,
+      "step": 2035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.681847336014392e-06,
+      "loss": 1.3164,
+      "step": 2036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6815307401325714e-06,
+      "loss": 0.9595,
+      "step": 2037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6812140055127096e-06,
+      "loss": 0.9813,
+      "step": 2038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.680897132191999e-06,
+      "loss": 1.0394,
+      "step": 2039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6805801202076473e-06,
+      "loss": 0.995,
+      "step": 2040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.680262969596879e-06,
+      "loss": 0.9909,
+      "step": 2041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.679945680396934e-06,
+      "loss": 1.1155,
+      "step": 2042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6796282526450704e-06,
+      "loss": 0.7887,
+      "step": 2043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6793106863785593e-06,
+      "loss": 0.8834,
+      "step": 2044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.678992981634692e-06,
+      "loss": 0.9346,
+      "step": 2045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6786751384507737e-06,
+      "loss": 0.9814,
+      "step": 2046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6783571568641253e-06,
+      "loss": 0.8506,
+      "step": 2047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6780390369120853e-06,
+      "loss": 0.9794,
+      "step": 2048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.677720778632009e-06,
+      "loss": 0.9644,
+      "step": 2049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.677402382061266e-06,
+      "loss": 1.1777,
+      "step": 2050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6770838472372435e-06,
+      "loss": 1.2116,
+      "step": 2051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6767651741973447e-06,
+      "loss": 0.7848,
+      "step": 2052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6764463629789886e-06,
+      "loss": 0.9325,
+      "step": 2053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6761274136196113e-06,
+      "loss": 0.842,
+      "step": 2054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.675808326156664e-06,
+      "loss": 1.1416,
+      "step": 2055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6754891006276145e-06,
+      "loss": 1.0791,
+      "step": 2056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.675169737069948e-06,
+      "loss": 0.7869,
+      "step": 2057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6748502355211637e-06,
+      "loss": 0.9751,
+      "step": 2058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.674530596018779e-06,
+      "loss": 1.0784,
+      "step": 2059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6742108186003264e-06,
+      "loss": 0.8685,
+      "step": 2060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6738909033033556e-06,
+      "loss": 1.3684,
+      "step": 2061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6735708501654307e-06,
+      "loss": 1.1059,
+      "step": 2062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.673250659224133e-06,
+      "loss": 1.1162,
+      "step": 2063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6729303305170615e-06,
+      "loss": 0.9774,
+      "step": 2064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6726098640818283e-06,
+      "loss": 0.9169,
+      "step": 2065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6722892599560642e-06,
+      "loss": 1.1151,
+      "step": 2066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6719685181774152e-06,
+      "loss": 0.9881,
+      "step": 2067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6716476387835434e-06,
+      "loss": 0.8807,
+      "step": 2068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.671326621812127e-06,
+      "loss": 0.8397,
+      "step": 2069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.671005467300861e-06,
+      "loss": 1.1132,
+      "step": 2070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.670684175287456e-06,
+      "loss": 0.8446,
+      "step": 2071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6703627458096383e-06,
+      "loss": 0.9369,
+      "step": 2072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.670041178905151e-06,
+      "loss": 0.8336,
+      "step": 2073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.669719474611754e-06,
+      "loss": 1.0781,
+      "step": 2074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.669397632967222e-06,
+      "loss": 1.1276,
+      "step": 2075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6690756540093466e-06,
+      "loss": 4.9433,
+      "step": 2076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6687535377759344e-06,
+      "loss": 0.676,
+      "step": 2077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6684312843048104e-06,
+      "loss": 1.1092,
+      "step": 2078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.668108893633814e-06,
+      "loss": 0.8805,
+      "step": 2079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6677863658008e-06,
+      "loss": 0.769,
+      "step": 2080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6674637008436413e-06,
+      "loss": 0.904,
+      "step": 2081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.667140898800226e-06,
+      "loss": 0.9438,
+      "step": 2082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.666817959708458e-06,
+      "loss": 0.8002,
+      "step": 2083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6664948836062575e-06,
+      "loss": 0.6627,
+      "step": 2084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.666171670531561e-06,
+      "loss": 0.955,
+      "step": 2085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6658483205223202e-06,
+      "loss": 1.0154,
+      "step": 2086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.665524833616505e-06,
+      "loss": 0.887,
+      "step": 2087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.665201209852099e-06,
+      "loss": 0.9935,
+      "step": 2088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6648774492671025e-06,
+      "loss": 0.9822,
+      "step": 2089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6645535518995334e-06,
+      "loss": 1.1637,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6642295177874235e-06,
+      "loss": 1.2246,
+      "step": 2091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6639053469688216e-06,
+      "loss": 0.8671,
+      "step": 2092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6635810394817933e-06,
+      "loss": 0.9071,
+      "step": 2093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.663256595364419e-06,
+      "loss": 1.1611,
+      "step": 2094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.662932014654796e-06,
+      "loss": 0.9294,
+      "step": 2095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6626072973910372e-06,
+      "loss": 1.3707,
+      "step": 2096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6622824436112713e-06,
+      "loss": 1.0313,
+      "step": 2097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6619574533536437e-06,
+      "loss": 0.9005,
+      "step": 2098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6616323266563152e-06,
+      "loss": 0.8663,
+      "step": 2099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.661307063557464e-06,
+      "loss": 0.7249,
+      "step": 2100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.660981664095281e-06,
+      "loss": 0.9924,
+      "step": 2101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6606561283079777e-06,
+      "loss": 0.9157,
+      "step": 2102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6603304562337782e-06,
+      "loss": 0.9306,
+      "step": 2103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6600046479109238e-06,
+      "loss": 0.8414,
+      "step": 2104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6596787033776713e-06,
+      "loss": 1.1335,
+      "step": 2105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.659352622672294e-06,
+      "loss": 1.1213,
+      "step": 2106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6590264058330817e-06,
+      "loss": 0.8544,
+      "step": 2107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.658700052898339e-06,
+      "loss": 0.7937,
+      "step": 2108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6583735639063864e-06,
+      "loss": 0.8076,
+      "step": 2109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.658046938895562e-06,
+      "loss": 0.9482,
+      "step": 2110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6577201779042185e-06,
+      "loss": 0.8815,
+      "step": 2111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6573932809707255e-06,
+      "loss": 1.0836,
+      "step": 2112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.657066248133467e-06,
+      "loss": 0.9827,
+      "step": 2113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.656739079430844e-06,
+      "loss": 0.9745,
+      "step": 2114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.656411774901274e-06,
+      "loss": 0.8799,
+      "step": 2115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6560843345831894e-06,
+      "loss": 1.0267,
+      "step": 2116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6557567585150393e-06,
+      "loss": 1.1237,
+      "step": 2117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6554290467352883e-06,
+      "loss": 1.1161,
+      "step": 2118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.655101199282417e-06,
+      "loss": 0.8325,
+      "step": 2119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6547732161949216e-06,
+      "loss": 0.8781,
+      "step": 2120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6544450975113156e-06,
+      "loss": 0.7878,
+      "step": 2121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6541168432701268e-06,
+      "loss": 0.8653,
+      "step": 2122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6537884535099e-06,
+      "loss": 1.0191,
+      "step": 2123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6534599282691947e-06,
+      "loss": 1.1653,
+      "step": 2124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6531312675865878e-06,
+      "loss": 1.1129,
+      "step": 2125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.652802471500671e-06,
+      "loss": 1.0294,
+      "step": 2126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6524735400500526e-06,
+      "loss": 0.9714,
+      "step": 2127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6521444732733564e-06,
+      "loss": 0.9569,
+      "step": 2128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.651815271209222e-06,
+      "loss": 0.9163,
+      "step": 2129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.651485933896305e-06,
+      "loss": 1.0002,
+      "step": 2130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6511564613732773e-06,
+      "loss": 0.7774,
+      "step": 2131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6508268536788257e-06,
+      "loss": 0.9512,
+      "step": 2132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6504971108516545e-06,
+      "loss": 0.8072,
+      "step": 2133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6501672329304816e-06,
+      "loss": 0.899,
+      "step": 2134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.649837219954043e-06,
+      "loss": 0.9938,
+      "step": 2135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6495070719610895e-06,
+      "loss": 1.0826,
+      "step": 2136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.649176788990387e-06,
+      "loss": 1.3658,
+      "step": 2137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.648846371080719e-06,
+      "loss": 1.0896,
+      "step": 2138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6485158182708834e-06,
+      "loss": 0.9479,
+      "step": 2139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6481851305996946e-06,
+      "loss": 0.9685,
+      "step": 2140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6478543081059832e-06,
+      "loss": 1.0831,
+      "step": 2141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.647523350828594e-06,
+      "loss": 1.0332,
+      "step": 2142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.64719225880639e-06,
+      "loss": 0.7812,
+      "step": 2143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.646861032078248e-06,
+      "loss": 0.8437,
+      "step": 2144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.646529670683062e-06,
+      "loss": 0.965,
+      "step": 2145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.646198174659741e-06,
+      "loss": 1.1639,
+      "step": 2146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.645866544047209e-06,
+      "loss": 1.0476,
+      "step": 2147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6455347788844083e-06,
+      "loss": 1.093,
+      "step": 2148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6452028792102946e-06,
+      "loss": 0.9949,
+      "step": 2149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6448708450638408e-06,
+      "loss": 0.7234,
+      "step": 2150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6445386764840352e-06,
+      "loss": 0.9393,
+      "step": 2151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.644206373509881e-06,
+      "loss": 1.1871,
+      "step": 2152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6438739361803986e-06,
+      "loss": 1.0066,
+      "step": 2153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6435413645346238e-06,
+      "loss": 0.8484,
+      "step": 2154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.643208658611607e-06,
+      "loss": 1.1043,
+      "step": 2155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.642875818450417e-06,
+      "loss": 0.7788,
+      "step": 2156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6425428440901346e-06,
+      "loss": 1.1388,
+      "step": 2157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.642209735569859e-06,
+      "loss": 0.7959,
+      "step": 2158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6418764929287055e-06,
+      "loss": 0.631,
+      "step": 2159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6415431162058038e-06,
+      "loss": 0.7646,
+      "step": 2160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.641209605440299e-06,
+      "loss": 0.9306,
+      "step": 2161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6408759606713537e-06,
+      "loss": 0.6788,
+      "step": 2162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.640542181938145e-06,
+      "loss": 0.8163,
+      "step": 2163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.640208269279865e-06,
+      "loss": 1.0438,
+      "step": 2164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639874222735724e-06,
+      "loss": 1.0853,
+      "step": 2165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639540042344945e-06,
+      "loss": 1.0671,
+      "step": 2166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6392057281467693e-06,
+      "loss": 0.9891,
+      "step": 2167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6388712801804526e-06,
+      "loss": 0.6334,
+      "step": 2168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6385366984852666e-06,
+      "loss": 0.9395,
+      "step": 2169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6382019831004984e-06,
+      "loss": 0.8579,
+      "step": 2170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6378671340654513e-06,
+      "loss": 1.0561,
+      "step": 2171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6375321514194435e-06,
+      "loss": 0.7974,
+      "step": 2172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6371970352018105e-06,
+      "loss": 1.5158,
+      "step": 2173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.636861785451901e-06,
+      "loss": 0.9171,
+      "step": 2174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.636526402209082e-06,
+      "loss": 1.0135,
+      "step": 2175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6361908855127337e-06,
+      "loss": 0.9125,
+      "step": 2176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.635855235402255e-06,
+      "loss": 0.9657,
+      "step": 2177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6355194519170575e-06,
+      "loss": 1.1745,
+      "step": 2178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6351835350965697e-06,
+      "loss": 0.991,
+      "step": 2179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.634847484980236e-06,
+      "loss": 0.9239,
+      "step": 2180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6345113016075163e-06,
+      "loss": 0.8767,
+      "step": 2181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.634174985017886e-06,
+      "loss": 1.0092,
+      "step": 2182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6338385352508355e-06,
+      "loss": 1.4239,
+      "step": 2183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6335019523458717e-06,
+      "loss": 1.1679,
+      "step": 2184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.633165236342518e-06,
+      "loss": 1.0639,
+      "step": 2185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.632828387280311e-06,
+      "loss": 1.1103,
+      "step": 2186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.632491405198805e-06,
+      "loss": 0.7895,
+      "step": 2187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6321542901375687e-06,
+      "loss": 0.7543,
+      "step": 2188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.631817042136188e-06,
+      "loss": 0.9466,
+      "step": 2189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6314796612342626e-06,
+      "loss": 0.8103,
+      "step": 2190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6311421474714075e-06,
+      "loss": 0.9168,
+      "step": 2191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6308045008872565e-06,
+      "loss": 0.8543,
+      "step": 2192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6304667215214554e-06,
+      "loss": 0.8971,
+      "step": 2193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6301288094136674e-06,
+      "loss": 1.1884,
+      "step": 2194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.629790764603571e-06,
+      "loss": 0.8462,
+      "step": 2195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.62945258713086e-06,
+      "loss": 0.8512,
+      "step": 2196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6291142770352437e-06,
+      "loss": 0.8845,
+      "step": 2197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.628775834356448e-06,
+      "loss": 1.0412,
+      "step": 2198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.628437259134213e-06,
+      "loss": 1.0042,
+      "step": 2199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.628098551408295e-06,
+      "loss": 1.1914,
+      "step": 2200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.627759711218466e-06,
+      "loss": 1.0425,
+      "step": 2201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6274207386045135e-06,
+      "loss": 0.7293,
+      "step": 2202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6270816336062405e-06,
+      "loss": 0.8341,
+      "step": 2203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.626742396263465e-06,
+      "loss": 0.8763,
+      "step": 2204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.626403026616022e-06,
+      "loss": 0.8872,
+      "step": 2205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.62606352470376e-06,
+      "loss": 1.0669,
+      "step": 2206
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6257238905665445e-06,
+      "loss": 1.0426,
+      "step": 2207
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.625384124244256e-06,
+      "loss": 1.1931,
+      "step": 2208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6250442257767907e-06,
+      "loss": 0.8706,
+      "step": 2209
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6247041952040605e-06,
+      "loss": 1.0388,
+      "step": 2210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6243640325659923e-06,
+      "loss": 0.6939,
+      "step": 2211
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6240237379025287e-06,
+      "loss": 1.0516,
+      "step": 2212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.623683311253628e-06,
+      "loss": 0.6993,
+      "step": 2213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6233427526592644e-06,
+      "loss": 0.9671,
+      "step": 2214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.623002062159426e-06,
+      "loss": 0.6895,
+      "step": 2215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6226612397941184e-06,
+      "loss": 0.9037,
+      "step": 2216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.622320285603361e-06,
+      "loss": 1.1853,
+      "step": 2217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6219791996271903e-06,
+      "loss": 1.1199,
+      "step": 2218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6216379819056566e-06,
+      "loss": 0.9729,
+      "step": 2219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6212966324788267e-06,
+      "loss": 1.0868,
+      "step": 2220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.620955151386783e-06,
+      "loss": 0.9986,
+      "step": 2221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.620613538669622e-06,
+      "loss": 1.0611,
+      "step": 2222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.620271794367458e-06,
+      "loss": 1.0859,
+      "step": 2223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.619929918520418e-06,
+      "loss": 1.1459,
+      "step": 2224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6195879111686465e-06,
+      "loss": 1.1931,
+      "step": 2225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6192457723523037e-06,
+      "loss": 1.0106,
+      "step": 2226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.618903502111563e-06,
+      "loss": 1.0907,
+      "step": 2227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.618561100486615e-06,
+      "loss": 1.0572,
+      "step": 2228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6182185675176653e-06,
+      "loss": 0.7676,
+      "step": 2229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6178759032449347e-06,
+      "loss": 1.0519,
+      "step": 2230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6175331077086606e-06,
+      "loss": 0.9305,
+      "step": 2231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6171901809490933e-06,
+      "loss": 1.3965,
+      "step": 2232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.616847123006501e-06,
+      "loss": 0.7759,
+      "step": 2233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6165039339211666e-06,
+      "loss": 0.5694,
+      "step": 2234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6161606137333867e-06,
+      "loss": 0.871,
+      "step": 2235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.615817162483477e-06,
+      "loss": 0.9658,
+      "step": 2236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.615473580211764e-06,
+      "loss": 0.9036,
+      "step": 2237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6151298669585934e-06,
+      "loss": 0.5242,
+      "step": 2238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6147860227643243e-06,
+      "loss": 0.8201,
+      "step": 2239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6144420476693314e-06,
+      "loss": 1.0767,
+      "step": 2240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.614097941714006e-06,
+      "loss": 1.1376,
+      "step": 2241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6137537049387535e-06,
+      "loss": 0.8657,
+      "step": 2242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.613409337383994e-06,
+      "loss": 0.8826,
+      "step": 2243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.613064839090165e-06,
+      "loss": 0.8488,
+      "step": 2244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6127202100977174e-06,
+      "loss": 1.017,
+      "step": 2245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.612375450447119e-06,
+      "loss": 0.8824,
+      "step": 2246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6120305601788523e-06,
+      "loss": 1.3436,
+      "step": 2247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6116855393334147e-06,
+      "loss": 1.092,
+      "step": 2248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6113403879513196e-06,
+      "loss": 1.2282,
+      "step": 2249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.610995106073095e-06,
+      "loss": 0.8347,
+      "step": 2250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.610649693739285e-06,
+      "loss": 0.9701,
+      "step": 2251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.61030415099045e-06,
+      "loss": 1.0506,
+      "step": 2252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.609958477867162e-06,
+      "loss": 0.6952,
+      "step": 2253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.609612674410012e-06,
+      "loss": 0.963,
+      "step": 2254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6092667406596055e-06,
+      "loss": 0.9516,
+      "step": 2255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6089206766565612e-06,
+      "loss": 0.8165,
+      "step": 2256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.608574482441517e-06,
+      "loss": 1.186,
+      "step": 2257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.608228158055122e-06,
+      "loss": 1.1854,
+      "step": 2258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6078817035380432e-06,
+      "loss": 1.0395,
+      "step": 2259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6075351189309618e-06,
+      "loss": 0.9028,
+      "step": 2260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6071884042745747e-06,
+      "loss": 0.8685,
+      "step": 2261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.606841559609594e-06,
+      "loss": 0.688,
+      "step": 2262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6064945849767465e-06,
+      "loss": 0.5562,
+      "step": 2263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6061474804167756e-06,
+      "loss": 0.9418,
+      "step": 2264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6058002459704385e-06,
+      "loss": 0.9117,
+      "step": 2265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6054528816785085e-06,
+      "loss": 0.9835,
+      "step": 2266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6051053875817736e-06,
+      "loss": 0.8961,
+      "step": 2267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.604757763721038e-06,
+      "loss": 0.8778,
+      "step": 2268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6044100101371195e-06,
+      "loss": 0.9242,
+      "step": 2269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.604062126870853e-06,
+      "loss": 0.7631,
+      "step": 2270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.603714113963087e-06,
+      "loss": 0.8356,
+      "step": 2271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.603365971454687e-06,
+      "loss": 1.0014,
+      "step": 2272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6030176993865315e-06,
+      "loss": 0.8695,
+      "step": 2273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6026692977995165e-06,
+      "loss": 1.0832,
+      "step": 2274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6023207667345514e-06,
+      "loss": 0.9918,
+      "step": 2275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6019721062325614e-06,
+      "loss": 0.7601,
+      "step": 2276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.601623316334488e-06,
+      "loss": 0.9116,
+      "step": 2277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6012743970812857e-06,
+      "loss": 1.1209,
+      "step": 2278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6009253485139262e-06,
+      "loss": 1.0155,
+      "step": 2279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6005761706733953e-06,
+      "loss": 1.0188,
+      "step": 2280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.600226863600694e-06,
+      "loss": 0.6458,
+      "step": 2281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.599877427336839e-06,
+      "loss": 0.8672,
+      "step": 2282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5995278619228623e-06,
+      "loss": 1.0948,
+      "step": 2283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5991781673998103e-06,
+      "loss": 0.8167,
+      "step": 2284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5988283438087445e-06,
+      "loss": 0.9881,
+      "step": 2285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5984783911907428e-06,
+      "loss": 1.2682,
+      "step": 2286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5981283095868967e-06,
+      "loss": 1.1878,
+      "step": 2287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.597778099038314e-06,
+      "loss": 1.2982,
+      "step": 2288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5974277595861175e-06,
+      "loss": 1.0164,
+      "step": 2289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5970772912714444e-06,
+      "loss": 0.6776,
+      "step": 2290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5967266941354475e-06,
+      "loss": 1.1601,
+      "step": 2291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5963759682192946e-06,
+      "loss": 0.8154,
+      "step": 2292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.596025113564169e-06,
+      "loss": 1.3286,
+      "step": 2293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5956741302112686e-06,
+      "loss": 0.6713,
+      "step": 2294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.595323018201807e-06,
+      "loss": 1.4138,
+      "step": 2295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.594971777577013e-06,
+      "loss": 0.6015,
+      "step": 2296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5946204083781287e-06,
+      "loss": 1.0331,
+      "step": 2297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.594268910646414e-06,
+      "loss": 0.8947,
+      "step": 2298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.593917284423142e-06,
+      "loss": 1.0055,
+      "step": 2299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5935655297496015e-06,
+      "loss": 0.9556,
+      "step": 2300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5932136466670962e-06,
+      "loss": 0.9529,
+      "step": 2301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.592861635216945e-06,
+      "loss": 0.9441,
+      "step": 2302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.592509495440483e-06,
+      "loss": 1.5322,
+      "step": 2303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5921572273790573e-06,
+      "loss": 1.0447,
+      "step": 2304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5918048310740334e-06,
+      "loss": 0.8469,
+      "step": 2305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5914523065667905e-06,
+      "loss": 1.1421,
+      "step": 2306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5910996538987223e-06,
+      "loss": 1.0144,
+      "step": 2307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5907468731112376e-06,
+      "loss": 0.9625,
+      "step": 2308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5903939642457623e-06,
+      "loss": 0.7254,
+      "step": 2309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.590040927343734e-06,
+      "loss": 0.9421,
+      "step": 2310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5896877624466085e-06,
+      "loss": 0.7976,
+      "step": 2311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.589334469595854e-06,
+      "loss": 1.1304,
+      "step": 2312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5889810488329565e-06,
+      "loss": 1.0357,
+      "step": 2313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.588627500199414e-06,
+      "loss": 1.0716,
+      "step": 2314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.588273823736742e-06,
+      "loss": 1.109,
+      "step": 2315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.587920019486469e-06,
+      "loss": 1.0031,
+      "step": 2316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.58756608749014e-06,
+      "loss": 1.1025,
+      "step": 2317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5872120277893154e-06,
+      "loss": 0.8572,
+      "step": 2318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.586857840425568e-06,
+      "loss": 0.9199,
+      "step": 2319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.586503525440489e-06,
+      "loss": 0.9582,
+      "step": 2320
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.2556439638137817,
+      "eval_runtime": 31.7958,
+      "eval_samples_per_second": 39.156,
+      "eval_steps_per_second": 19.594,
+      "step": 2320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5861490828756814e-06,
+      "loss": 0.925,
+      "step": 2321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.585794512772766e-06,
+      "loss": 1.1371,
+      "step": 2322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.585439815173376e-06,
+      "loss": 1.2926,
+      "step": 2323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5850849901191614e-06,
+      "loss": 1.0175,
+      "step": 2324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5847300376517864e-06,
+      "loss": 0.9989,
+      "step": 2325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5843749578129304e-06,
+      "loss": 1.0623,
+      "step": 2326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5840197506442884e-06,
+      "loss": 1.1541,
+      "step": 2327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5836644161875683e-06,
+      "loss": 0.6855,
+      "step": 2328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.583308954484495e-06,
+      "loss": 0.6819,
+      "step": 2329
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.582953365576808e-06,
+      "loss": 1.0005,
+      "step": 2330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5825976495062605e-06,
+      "loss": 1.2972,
+      "step": 2331
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.582241806314622e-06,
+      "loss": 0.9696,
+      "step": 2332
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5818858360436765e-06,
+      "loss": 1.0322,
+      "step": 2333
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5815297387352228e-06,
+      "loss": 1.1816,
+      "step": 2334
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5811735144310743e-06,
+      "loss": 1.5587,
+      "step": 2335
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5808171631730603e-06,
+      "loss": 1.1781,
+      "step": 2336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5804606850030236e-06,
+      "loss": 0.531,
+      "step": 2337
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5801040799628234e-06,
+      "loss": 1.0931,
+      "step": 2338
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5797473480943322e-06,
+      "loss": 0.9056,
+      "step": 2339
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5793904894394394e-06,
+      "loss": 0.6162,
+      "step": 2340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5790335040400475e-06,
+      "loss": 0.8722,
+      "step": 2341
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5786763919380746e-06,
+      "loss": 1.1654,
+      "step": 2342
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5783191531754534e-06,
+      "loss": 0.7685,
+      "step": 2343
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5779617877941325e-06,
+      "loss": 0.9146,
+      "step": 2344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5776042958360733e-06,
+      "loss": 0.9437,
+      "step": 2345
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5772466773432545e-06,
+      "loss": 0.8539,
+      "step": 2346
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.576888932357668e-06,
+      "loss": 1.015,
+      "step": 2347
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.576531060921321e-06,
+      "loss": 0.8624,
+      "step": 2348
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5761730630762357e-06,
+      "loss": 0.9613,
+      "step": 2349
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.575814938864449e-06,
+      "loss": 0.6335,
+      "step": 2350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5754566883280124e-06,
+      "loss": 0.8723,
+      "step": 2351
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5750983115089927e-06,
+      "loss": 1.1793,
+      "step": 2352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5747398084494718e-06,
+      "loss": 0.9308,
+      "step": 2353
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5743811791915445e-06,
+      "loss": 0.533,
+      "step": 2354
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.574022423777324e-06,
+      "loss": 0.8455,
+      "step": 2355
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.573663542248934e-06,
+      "loss": 0.7423,
+      "step": 2356
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.573304534648517e-06,
+      "loss": 1.0141,
+      "step": 2357
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5729454010182275e-06,
+      "loss": 1.2003,
+      "step": 2358
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.572586141400236e-06,
+      "loss": 1.0563,
+      "step": 2359
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5722267558367282e-06,
+      "loss": 1.0265,
+      "step": 2360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5718672443699023e-06,
+      "loss": 0.983,
+      "step": 2361
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5715076070419747e-06,
+      "loss": 0.9009,
+      "step": 2362
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5711478438951743e-06,
+      "loss": 1.075,
+      "step": 2363
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5707879549717447e-06,
+      "loss": 0.8962,
+      "step": 2364
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5704279403139457e-06,
+      "loss": 0.9643,
+      "step": 2365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5700677999640505e-06,
+      "loss": 0.6673,
+      "step": 2366
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5697075339643477e-06,
+      "loss": 1.583,
+      "step": 2367
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5693471423571407e-06,
+      "loss": 1.3425,
+      "step": 2368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.568986625184748e-06,
+      "loss": 1.0939,
+      "step": 2369
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5686259824895015e-06,
+      "loss": 1.3224,
+      "step": 2370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.568265214313749e-06,
+      "loss": 1.0354,
+      "step": 2371
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5679043206998525e-06,
+      "loss": 1.0086,
+      "step": 2372
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5675433016901894e-06,
+      "loss": 0.9747,
+      "step": 2373
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5671821573271514e-06,
+      "loss": 0.6542,
+      "step": 2374
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5668208876531443e-06,
+      "loss": 1.1192,
+      "step": 2375
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.56645949271059e-06,
+      "loss": 0.9265,
+      "step": 2376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.566097972541923e-06,
+      "loss": 1.0375,
+      "step": 2377
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5657363271895956e-06,
+      "loss": 0.4784,
+      "step": 2378
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5653745566960714e-06,
+      "loss": 1.0149,
+      "step": 2379
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.565012661103832e-06,
+      "loss": 0.902,
+      "step": 2380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5646506404553703e-06,
+      "loss": 1.3313,
+      "step": 2381
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.564288494793197e-06,
+      "loss": 0.7804,
+      "step": 2382
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.563926224159835e-06,
+      "loss": 1.0516,
+      "step": 2383
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5635638285978235e-06,
+      "loss": 0.871,
+      "step": 2384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5632013081497156e-06,
+      "loss": 1.0549,
+      "step": 2385
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5628386628580796e-06,
+      "loss": 1.1215,
+      "step": 2386
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.562475892765498e-06,
+      "loss": 1.1842,
+      "step": 2387
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5621129979145678e-06,
+      "loss": 1.0912,
+      "step": 2388
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5617499783479008e-06,
+      "loss": 0.6098,
+      "step": 2389
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5613868341081242e-06,
+      "loss": 0.9076,
+      "step": 2390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5610235652378794e-06,
+      "loss": 0.7464,
+      "step": 2391
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5606601717798212e-06,
+      "loss": 1.108,
+      "step": 2392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5602966537766215e-06,
+      "loss": 0.6375,
+      "step": 2393
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5599330112709636e-06,
+      "loss": 0.6843,
+      "step": 2394
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.559569244305549e-06,
+      "loss": 0.8292,
+      "step": 2395
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5592053529230913e-06,
+      "loss": 0.8574,
+      "step": 2396
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5588413371663195e-06,
+      "loss": 1.0824,
+      "step": 2397
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.558477197077977e-06,
+      "loss": 1.0934,
+      "step": 2398
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5581129327008215e-06,
+      "loss": 0.8951,
+      "step": 2399
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.557748544077627e-06,
+      "loss": 1.6814,
+      "step": 2400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.55738403125118e-06,
+      "loss": 0.9672,
+      "step": 2401
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.557019394264283e-06,
+      "loss": 1.0595,
+      "step": 2402
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5566546331597518e-06,
+      "loss": 0.9628,
+      "step": 2403
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.556289747980418e-06,
+      "loss": 1.2474,
+      "step": 2404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.555924738769127e-06,
+      "loss": 1.0117,
+      "step": 2405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.555559605568739e-06,
+      "loss": 7.1606,
+      "step": 2406
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5551943484221286e-06,
+      "loss": 0.8736,
+      "step": 2407
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.554828967372186e-06,
+      "loss": 1.132,
+      "step": 2408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5544634624618143e-06,
+      "loss": 0.7341,
+      "step": 2409
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.554097833733932e-06,
+      "loss": 0.7803,
+      "step": 2410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.553732081231472e-06,
+      "loss": 0.6861,
+      "step": 2411
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5533662049973823e-06,
+      "loss": 1.0436,
+      "step": 2412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5530002050746238e-06,
+      "loss": 0.9688,
+      "step": 2413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.552634081506175e-06,
+      "loss": 0.9996,
+      "step": 2414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5522678343350254e-06,
+      "loss": 0.9436,
+      "step": 2415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5519014636041805e-06,
+      "loss": 1.1553,
+      "step": 2416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.551534969356661e-06,
+      "loss": 0.951,
+      "step": 2417
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5511683516355016e-06,
+      "loss": 0.9266,
+      "step": 2418
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5508016104837514e-06,
+      "loss": 0.6679,
+      "step": 2419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.550434745944473e-06,
+      "loss": 0.986,
+      "step": 2420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.550067758060745e-06,
+      "loss": 0.869,
+      "step": 2421
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5497006468756613e-06,
+      "loss": 0.8697,
+      "step": 2422
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5493334124323273e-06,
+      "loss": 0.9364,
+      "step": 2423
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5489660547738645e-06,
+      "loss": 1.1088,
+      "step": 2424
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5485985739434095e-06,
+      "loss": 0.7782,
+      "step": 2425
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.548230969984113e-06,
+      "loss": 1.1305,
+      "step": 2426
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5478632429391394e-06,
+      "loss": 1.2706,
+      "step": 2427
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5474953928516683e-06,
+      "loss": 0.7652,
+      "step": 2428
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5471274197648934e-06,
+      "loss": 1.0144,
+      "step": 2429
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.546759323722023e-06,
+      "loss": 1.2083,
+      "step": 2430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.546391104766279e-06,
+      "loss": 1.1624,
+      "step": 2431
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5460227629408995e-06,
+      "loss": 0.7907,
+      "step": 2432
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5456542982891356e-06,
+      "loss": 1.1424,
+      "step": 2433
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.545285710854254e-06,
+      "loss": 0.9047,
+      "step": 2434
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5449170006795337e-06,
+      "loss": 0.9273,
+      "step": 2435
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.544548167808271e-06,
+      "loss": 1.0266,
+      "step": 2436
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.544179212283774e-06,
+      "loss": 1.3004,
+      "step": 2437
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.543810134149367e-06,
+      "loss": 0.8168,
+      "step": 2438
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5434409334483864e-06,
+      "loss": 0.7521,
+      "step": 2439
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.543071610224187e-06,
+      "loss": 0.9549,
+      "step": 2440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.542702164520134e-06,
+      "loss": 1.0244,
+      "step": 2441
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5423325963796095e-06,
+      "loss": 1.0518,
+      "step": 2442
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.541962905846008e-06,
+      "loss": 0.5628,
+      "step": 2443
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5415930929627405e-06,
+      "loss": 0.8749,
+      "step": 2444
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5412231577732305e-06,
+      "loss": 1.047,
+      "step": 2445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5408531003209164e-06,
+      "loss": 1.2149,
+      "step": 2446
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.540482920649252e-06,
+      "loss": 0.6893,
+      "step": 2447
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5401126188017043e-06,
+      "loss": 0.6614,
+      "step": 2448
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.539742194821755e-06,
+      "loss": 1.1236,
+      "step": 2449
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5393716487528995e-06,
+      "loss": 0.9625,
+      "step": 2450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5390009806386496e-06,
+      "loss": 1.0051,
+      "step": 2451
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.538630190522529e-06,
+      "loss": 0.9657,
+      "step": 2452
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5382592784480764e-06,
+      "loss": 0.848,
+      "step": 2453
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.537888244458846e-06,
+      "loss": 1.1419,
+      "step": 2454
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5375170885984046e-06,
+      "loss": 0.8439,
+      "step": 2455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5371458109103357e-06,
+      "loss": 0.6136,
+      "step": 2456
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5367744114382337e-06,
+      "loss": 1.1142,
+      "step": 2457
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.53640289022571e-06,
+      "loss": 0.7367,
+      "step": 2458
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5360312473163896e-06,
+      "loss": 0.929,
+      "step": 2459
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5356594827539117e-06,
+      "loss": 1.0352,
+      "step": 2460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5352875965819294e-06,
+      "loss": 1.0086,
+      "step": 2461
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.534915588844111e-06,
+      "loss": 1.1037,
+      "step": 2462
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5345434595841375e-06,
+      "loss": 0.8179,
+      "step": 2463
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.534171208845706e-06,
+      "loss": 1.0403,
+      "step": 2464
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.533798836672527e-06,
+      "loss": 0.9807,
+      "step": 2465
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.533426343108325e-06,
+      "loss": 0.8252,
+      "step": 2466
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5330537281968392e-06,
+      "loss": 0.8318,
+      "step": 2467
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5326809919818227e-06,
+      "loss": 0.8181,
+      "step": 2468
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.532308134507043e-06,
+      "loss": 0.8583,
+      "step": 2469
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5319351558162823e-06,
+      "loss": 0.8907,
+      "step": 2470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5315620559533366e-06,
+      "loss": 1.0339,
+      "step": 2471
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5311888349620155e-06,
+      "loss": 0.9013,
+      "step": 2472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5308154928861445e-06,
+      "loss": 1.0625,
+      "step": 2473
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.530442029769561e-06,
+      "loss": 1.2366,
+      "step": 2474
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.530068445656119e-06,
+      "loss": 0.7907,
+      "step": 2475
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5296947405896845e-06,
+      "loss": 1.0453,
+      "step": 2476
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.52932091461414e-06,
+      "loss": 0.8629,
+      "step": 2477
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5289469677733803e-06,
+      "loss": 0.9315,
+      "step": 2478
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5285729001113147e-06,
+      "loss": 1.0497,
+      "step": 2479
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5281987116718685e-06,
+      "loss": 1.1773,
+      "step": 2480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5278244024989786e-06,
+      "loss": 0.7074,
+      "step": 2481
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5274499726365975e-06,
+      "loss": 1.0942,
+      "step": 2482
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5270754221286913e-06,
+      "loss": 0.8741,
+      "step": 2483
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.526700751019241e-06,
+      "loss": 1.0115,
+      "step": 2484
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5263259593522417e-06,
+      "loss": 0.9186,
+      "step": 2485
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5259510471717016e-06,
+      "loss": 1.0414,
+      "step": 2486
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5255760145216436e-06,
+      "loss": 0.8089,
+      "step": 2487
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.525200861446106e-06,
+      "loss": 0.9579,
+      "step": 2488
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.5248255879891385e-06,
+      "loss": 1.0916,
+      "step": 2489
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5244501941948084e-06,
+      "loss": 0.7339,
+      "step": 2490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.524074680107194e-06,
+      "loss": 0.7612,
+      "step": 2491
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.523699045770389e-06,
+      "loss": 0.7962,
+      "step": 2492
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5233232912285023e-06,
+      "loss": 0.7948,
+      "step": 2493
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5229474165256547e-06,
+      "loss": 1.2014,
+      "step": 2494
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.522571421705983e-06,
+      "loss": 1.0086,
+      "step": 2495
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5221953068136373e-06,
+      "loss": 0.9105,
+      "step": 2496
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5218190718927816e-06,
+      "loss": 1.1624,
+      "step": 2497
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.521442716987594e-06,
+      "loss": 0.8083,
+      "step": 2498
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5210662421422677e-06,
+      "loss": 0.7885,
+      "step": 2499
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5206896474010088e-06,
+      "loss": 1.2886,
+      "step": 2500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.520312932808038e-06,
+      "loss": 0.7225,
+      "step": 2501
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5199360984075894e-06,
+      "loss": 0.8698,
+      "step": 2502
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.519559144243913e-06,
+      "loss": 0.9908,
+      "step": 2503
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5191820703612705e-06,
+      "loss": 1.0439,
+      "step": 2504
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.51880487680394e-06,
+      "loss": 1.1282,
+      "step": 2505
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5184275636162106e-06,
+      "loss": 1.0074,
+      "step": 2506
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.518050130842389e-06,
+      "loss": 1.111,
+      "step": 2507
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.517672578526793e-06,
+      "loss": 1.0498,
+      "step": 2508
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5172949067137574e-06,
+      "loss": 0.9295,
+      "step": 2509
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5169171154476263e-06,
+      "loss": 0.9655,
+      "step": 2510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5165392047727637e-06,
+      "loss": 0.5978,
+      "step": 2511
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.5161611747335434e-06,
+      "loss": 1.4487,
+      "step": 2512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.515783025374355e-06,
+      "loss": 1.0124,
+      "step": 2513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5154047567396016e-06,
+      "loss": 1.1097,
+      "step": 2514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5150263688737e-06,
+      "loss": 0.9237,
+      "step": 2515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5146478618210815e-06,
+      "loss": 0.9277,
+      "step": 2516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5142692356261912e-06,
+      "loss": 0.7699,
+      "step": 2517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5138904903334885e-06,
+      "loss": 0.9681,
+      "step": 2518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5135116259874467e-06,
+      "loss": 0.9092,
+      "step": 2519
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5131326426325525e-06,
+      "loss": 1.0778,
+      "step": 2520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.512753540313307e-06,
+      "loss": 1.1998,
+      "step": 2521
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5123743190742255e-06,
+      "loss": 0.9571,
+      "step": 2522
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.511994978959837e-06,
+      "loss": 1.0997,
+      "step": 2523
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5116155200146848e-06,
+      "loss": 1.336,
+      "step": 2524
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.511235942283325e-06,
+      "loss": 1.1284,
+      "step": 2525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5108562458103284e-06,
+      "loss": 0.8306,
+      "step": 2526
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5104764306402813e-06,
+      "loss": 1.1278,
+      "step": 2527
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.510096496817781e-06,
+      "loss": 0.8534,
+      "step": 2528
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.509716444387441e-06,
+      "loss": 0.6925,
+      "step": 2529
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.509336273393888e-06,
+      "loss": 1.2175,
+      "step": 2530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.508955983881762e-06,
+      "loss": 0.7948,
+      "step": 2531
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.508575575895717e-06,
+      "loss": 1.1404,
+      "step": 2532
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.508195049480423e-06,
+      "loss": 1.2131,
+      "step": 2533
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5078144046805606e-06,
+      "loss": 0.8778,
+      "step": 2534
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.5074336415408266e-06,
+      "loss": 1.2591,
+      "step": 2535
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.507052760105932e-06,
+      "loss": 0.9651,
+      "step": 2536
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5066717604205994e-06,
+      "loss": 0.9182,
+      "step": 2537
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5062906425295673e-06,
+      "loss": 0.9418,
+      "step": 2538
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5059094064775873e-06,
+      "loss": 0.9968,
+      "step": 2539
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5055280523094255e-06,
+      "loss": 0.8205,
+      "step": 2540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5051465800698603e-06,
+      "loss": 0.73,
+      "step": 2541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.504764989803686e-06,
+      "loss": 0.9783,
+      "step": 2542
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.504383281555709e-06,
+      "loss": 1.1105,
+      "step": 2543
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5040014553707514e-06,
+      "loss": 1.2167,
+      "step": 2544
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5036195112936475e-06,
+      "loss": 0.8922,
+      "step": 2545
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5032374493692456e-06,
+      "loss": 0.8815,
+      "step": 2546
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5028552696424097e-06,
+      "loss": 0.8245,
+      "step": 2547
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5024729721580144e-06,
+      "loss": 0.7858,
+      "step": 2548
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.502090556960951e-06,
+      "loss": 1.3013,
+      "step": 2549
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.501708024096124e-06,
+      "loss": 0.9579,
+      "step": 2550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.50132537360845e-06,
+      "loss": 0.9012,
+      "step": 2551
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.5009426055428616e-06,
+      "loss": 1.1126,
+      "step": 2552
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.500559719944304e-06,
+      "loss": 0.9676,
+      "step": 2553
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.500176716857736e-06,
+      "loss": 1.0251,
+      "step": 2554
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4997935963281323e-06,
+      "loss": 1.061,
+      "step": 2555
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4994103584004787e-06,
+      "loss": 1.2763,
+      "step": 2556
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4990270031197755e-06,
+      "loss": 0.5624,
+      "step": 2557
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.498643530531038e-06,
+      "loss": 0.9289,
+      "step": 2558
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4982599406792936e-06,
+      "loss": 0.9317,
+      "step": 2559
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4978762336095845e-06,
+      "loss": 0.9323,
+      "step": 2560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4974924093669674e-06,
+      "loss": 0.9674,
+      "step": 2561
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4971084679965106e-06,
+      "loss": 0.9061,
+      "step": 2562
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.496724409543298e-06,
+      "loss": 1.0783,
+      "step": 2563
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4963402340524264e-06,
+      "loss": 0.4885,
+      "step": 2564
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4959559415690067e-06,
+      "loss": 0.7584,
+      "step": 2565
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4955715321381633e-06,
+      "loss": 1.0364,
+      "step": 2566
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.495187005805035e-06,
+      "loss": 1.3522,
+      "step": 2567
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.494802362614773e-06,
+      "loss": 0.9979,
+      "step": 2568
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4944176026125428e-06,
+      "loss": 0.9545,
+      "step": 2569
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4940327258435243e-06,
+      "loss": 1.0676,
+      "step": 2570
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.493647732352911e-06,
+      "loss": 0.711,
+      "step": 2571
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.493262622185909e-06,
+      "loss": 1.0011,
+      "step": 2572
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4928773953877392e-06,
+      "loss": 0.9036,
+      "step": 2573
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4924920520036357e-06,
+      "loss": 0.9596,
+      "step": 2574
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.492106592078846e-06,
+      "loss": 0.9664,
+      "step": 2575
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.491721015658633e-06,
+      "loss": 0.7917,
+      "step": 2576
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.49133532278827e-06,
+      "loss": 0.7859,
+      "step": 2577
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.490949513513047e-06,
+      "loss": 0.8622,
+      "step": 2578
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4905635878782674e-06,
+      "loss": 0.8306,
+      "step": 2579
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.490177545929246e-06,
+      "loss": 0.6728,
+      "step": 2580
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4897913877113138e-06,
+      "loss": 0.936,
+      "step": 2581
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.489405113269814e-06,
+      "loss": 0.8633,
+      "step": 2582
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4890187226501035e-06,
+      "loss": 0.87,
+      "step": 2583
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4886322158975532e-06,
+      "loss": 1.1496,
+      "step": 2584
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4882455930575485e-06,
+      "loss": 1.1804,
+      "step": 2585
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4878588541754866e-06,
+      "loss": 1.0818,
+      "step": 2586
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4874719992967797e-06,
+      "loss": 0.9902,
+      "step": 2587
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4870850284668526e-06,
+      "loss": 0.9424,
+      "step": 2588
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.486697941731145e-06,
+      "loss": 1.0025,
+      "step": 2589
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4863107391351097e-06,
+      "loss": 0.9878,
+      "step": 2590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4859234207242124e-06,
+      "loss": 0.7819,
+      "step": 2591
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.485535986543933e-06,
+      "loss": 0.8938,
+      "step": 2592
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4851484366397645e-06,
+      "loss": 1.2243,
+      "step": 2593
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4847607710572145e-06,
+      "loss": 1.2802,
+      "step": 2594
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.484372989841803e-06,
+      "loss": 0.8045,
+      "step": 2595
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4839850930390655e-06,
+      "loss": 1.003,
+      "step": 2596
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4835970806945478e-06,
+      "loss": 0.7899,
+      "step": 2597
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4832089528538127e-06,
+      "loss": 0.9639,
+      "step": 2598
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.482820709562434e-06,
+      "loss": 0.9513,
+      "step": 2599
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4824323508660014e-06,
+      "loss": 1.0523,
+      "step": 2600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4820438768101155e-06,
+      "loss": 1.0782,
+      "step": 2601
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4816552874403923e-06,
+      "loss": 0.9752,
+      "step": 2602
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4812665828024616e-06,
+      "loss": 1.2415,
+      "step": 2603
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.480877762941965e-06,
+      "loss": 0.6793,
+      "step": 2604
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4804888279045584e-06,
+      "loss": 0.956,
+      "step": 2605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.480099777735913e-06,
+      "loss": 0.9999,
+      "step": 2606
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4797106124817103e-06,
+      "loss": 1.1231,
+      "step": 2607
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.479321332187648e-06,
+      "loss": 0.9158,
+      "step": 2608
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.478931936899436e-06,
+      "loss": 1.0454,
+      "step": 2609
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4785424266627974e-06,
+      "loss": 0.9577,
+      "step": 2610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4781528015234702e-06,
+      "loss": 1.2663,
+      "step": 2611
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4777630615272047e-06,
+      "loss": 1.0804,
+      "step": 2612
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.477373206719765e-06,
+      "loss": 0.6738,
+      "step": 2613
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.476983237146929e-06,
+      "loss": 1.1237,
+      "step": 2614
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.476593152854488e-06,
+      "loss": 0.8463,
+      "step": 2615
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.476202953888246e-06,
+      "loss": 0.5317,
+      "step": 2616
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4758126402940214e-06,
+      "loss": 0.9565,
+      "step": 2617
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4754222121176454e-06,
+      "loss": 0.9107,
+      "step": 2618
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.475031669404964e-06,
+      "loss": 0.825,
+      "step": 2619
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.474641012201834e-06,
+      "loss": 1.0717,
+      "step": 2620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.474250240554128e-06,
+      "loss": 1.0429,
+      "step": 2621
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.473859354507732e-06,
+      "loss": 0.8601,
+      "step": 2622
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4734683541085438e-06,
+      "loss": 1.1099,
+      "step": 2623
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.473077239402476e-06,
+      "loss": 0.8243,
+      "step": 2624
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.472686010435454e-06,
+      "loss": 1.1395,
+      "step": 2625
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.472294667253417e-06,
+      "loss": 0.7941,
+      "step": 2626
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.471903209902317e-06,
+      "loss": 1.2336,
+      "step": 2627
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4715116384281206e-06,
+      "loss": 0.9558,
+      "step": 2628
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4711199528768056e-06,
+      "loss": 1.038,
+      "step": 2629
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4707281532943667e-06,
+      "loss": 1.0425,
+      "step": 2630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4703362397268078e-06,
+      "loss": 0.9683,
+      "step": 2631
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.46994421222015e-06,
+      "loss": 1.024,
+      "step": 2632
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4695520708204247e-06,
+      "loss": 1.0468,
+      "step": 2633
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.469159815573679e-06,
+      "loss": 0.8857,
+      "step": 2634
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4687674465259717e-06,
+      "loss": 1.2018,
+      "step": 2635
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.468374963723376e-06,
+      "loss": 0.9468,
+      "step": 2636
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4679823672119785e-06,
+      "loss": 1.2098,
+      "step": 2637
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4675896570378784e-06,
+      "loss": 0.8503,
+      "step": 2638
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.467196833247188e-06,
+      "loss": 0.8843,
+      "step": 2639
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.466803895886035e-06,
+      "loss": 1.3717,
+      "step": 2640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.466410845000558e-06,
+      "loss": 0.8819,
+      "step": 2641
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4660176806369103e-06,
+      "loss": 1.1798,
+      "step": 2642
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4656244028412574e-06,
+      "loss": 0.759,
+      "step": 2643
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.46523101165978e-06,
+      "loss": 1.075,
+      "step": 2644
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4648375071386698e-06,
+      "loss": 0.8983,
+      "step": 2645
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.464443889324135e-06,
+      "loss": 1.2313,
+      "step": 2646
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4640501582623923e-06,
+      "loss": 0.891,
+      "step": 2647
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.463656313999677e-06,
+      "loss": 1.0126,
+      "step": 2648
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4632623565822335e-06,
+      "loss": 1.2146,
+      "step": 2649
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.462868286056322e-06,
+      "loss": 0.7453,
+      "step": 2650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4624741024682153e-06,
+      "loss": 0.6798,
+      "step": 2651
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.462079805864199e-06,
+      "loss": 1.1141,
+      "step": 2652
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.461685396290572e-06,
+      "loss": 1.0828,
+      "step": 2653
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4612908737936477e-06,
+      "loss": 0.5539,
+      "step": 2654
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4608962384197507e-06,
+      "loss": 0.8562,
+      "step": 2655
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4605014902152214e-06,
+      "loss": 1.0911,
+      "step": 2656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.460106629226411e-06,
+      "loss": 1.1502,
+      "step": 2657
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.459711655499684e-06,
+      "loss": 1.1744,
+      "step": 2658
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4593165690814212e-06,
+      "loss": 0.8058,
+      "step": 2659
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4589213700180135e-06,
+      "loss": 0.9511,
+      "step": 2660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4585260583558666e-06,
+      "loss": 1.2131,
+      "step": 2661
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.458130634141398e-06,
+      "loss": 0.6333,
+      "step": 2662
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.45773509742104e-06,
+      "loss": 0.9699,
+      "step": 2663
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4573394482412376e-06,
+      "loss": 0.8509,
+      "step": 2664
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.456943686648448e-06,
+      "loss": 1.3955,
+      "step": 2665
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4565478126891434e-06,
+      "loss": 1.0224,
+      "step": 2666
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.456151826409807e-06,
+      "loss": 0.945,
+      "step": 2667
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4557557278569376e-06,
+      "loss": 0.8569,
+      "step": 2668
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.455359517077046e-06,
+      "loss": 0.9466,
+      "step": 2669
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.454963194116655e-06,
+      "loss": 0.865,
+      "step": 2670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.454566759022303e-06,
+      "loss": 1.0599,
+      "step": 2671
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4541702118405394e-06,
+      "loss": 0.8234,
+      "step": 2672
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4537735526179282e-06,
+      "loss": 1.3785,
+      "step": 2673
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.453376781401046e-06,
+      "loss": 0.9222,
+      "step": 2674
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4529798982364825e-06,
+      "loss": 1.1764,
+      "step": 2675
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4525829031708406e-06,
+      "loss": 0.9599,
+      "step": 2676
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4521857962507363e-06,
+      "loss": 0.9531,
+      "step": 2677
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4517885775227993e-06,
+      "loss": 0.9274,
+      "step": 2678
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4513912470336713e-06,
+      "loss": 0.9977,
+      "step": 2679
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4509938048300075e-06,
+      "loss": 1.1954,
+      "step": 2680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4505962509584775e-06,
+      "loss": 3.1061,
+      "step": 2681
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4501985854657624e-06,
+      "loss": 1.0261,
+      "step": 2682
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4498008083985567e-06,
+      "loss": 0.8466,
+      "step": 2683
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.449402919803569e-06,
+      "loss": 0.8899,
+      "step": 2684
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4490049197275195e-06,
+      "loss": 1.0001,
+      "step": 2685
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4486068082171427e-06,
+      "loss": 1.0564,
+      "step": 2686
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4482085853191864e-06,
+      "loss": 3.2739,
+      "step": 2687
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4478102510804094e-06,
+      "loss": 0.9619,
+      "step": 2688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4474118055475865e-06,
+      "loss": 0.9056,
+      "step": 2689
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4470132487675027e-06,
+      "loss": 0.9096,
+      "step": 2690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4466145807869588e-06,
+      "loss": 1.0023,
+      "step": 2691
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.446215801652766e-06,
+      "loss": 1.0781,
+      "step": 2692
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.445816911411751e-06,
+      "loss": 0.5919,
+      "step": 2693
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4454179101107515e-06,
+      "loss": 1.2899,
+      "step": 2694
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4450187977966207e-06,
+      "loss": 1.0621,
+      "step": 2695
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4446195745162203e-06,
+      "loss": 1.0867,
+      "step": 2696
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.444220240316431e-06,
+      "loss": 0.8543,
+      "step": 2697
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4438207952441424e-06,
+      "loss": 0.7192,
+      "step": 2698
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.443421239346258e-06,
+      "loss": 0.8982,
+      "step": 2699
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.443021572669695e-06,
+      "loss": 1.2484,
+      "step": 2700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4426217952613826e-06,
+      "loss": 0.9262,
+      "step": 2701
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4422219071682644e-06,
+      "loss": 1.0028,
+      "step": 2702
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.441821908437295e-06,
+      "loss": 1.0324,
+      "step": 2703
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4414217991154443e-06,
+      "loss": 1.0685,
+      "step": 2704
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4410215792496943e-06,
+      "loss": 0.8754,
+      "step": 2705
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4406212488870383e-06,
+      "loss": 1.2566,
+      "step": 2706
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4402208080744847e-06,
+      "loss": 1.0405,
+      "step": 2707
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4398202568590546e-06,
+      "loss": 0.8868,
+      "step": 2708
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.439419595287781e-06,
+      "loss": 0.9321,
+      "step": 2709
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4390188234077113e-06,
+      "loss": 0.9362,
+      "step": 2710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4386179412659044e-06,
+      "loss": 0.8097,
+      "step": 2711
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4382169489094325e-06,
+      "loss": 0.7423,
+      "step": 2712
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.437815846385382e-06,
+      "loss": 0.9435,
+      "step": 2713
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4374146337408504e-06,
+      "loss": 1.1807,
+      "step": 2714
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4370133110229493e-06,
+      "loss": 1.0028,
+      "step": 2715
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4366118782788032e-06,
+      "loss": 1.0537,
+      "step": 2716
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4362103355555494e-06,
+      "loss": 1.0122,
+      "step": 2717
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.435808682900337e-06,
+      "loss": 1.1843,
+      "step": 2718
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4354069203603296e-06,
+      "loss": 0.9758,
+      "step": 2719
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4350050479827037e-06,
+      "loss": 0.8752,
+      "step": 2720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.434603065814647e-06,
+      "loss": 0.8919,
+      "step": 2721
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.434200973903362e-06,
+      "loss": 1.0056,
+      "step": 2722
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.433798772296062e-06,
+      "loss": 0.909,
+      "step": 2723
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.433396461039976e-06,
+      "loss": 1.2045,
+      "step": 2724
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.432994040182343e-06,
+      "loss": 0.8423,
+      "step": 2725
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4325915097704174e-06,
+      "loss": 1.0705,
+      "step": 2726
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.432188869851464e-06,
+      "loss": 0.8814,
+      "step": 2727
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4317861204727628e-06,
+      "loss": 1.0182,
+      "step": 2728
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4313832616816047e-06,
+      "loss": 0.6807,
+      "step": 2729
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.430980293525295e-06,
+      "loss": 0.9818,
+      "step": 2730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4305772160511506e-06,
+      "loss": 0.7988,
+      "step": 2731
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4301740293065024e-06,
+      "loss": 1.1167,
+      "step": 2732
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4297707333386925e-06,
+      "loss": 0.8922,
+      "step": 2733
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4293673281950772e-06,
+      "loss": 1.1808,
+      "step": 2734
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.428963813923026e-06,
+      "loss": 0.7922,
+      "step": 2735
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4285601905699203e-06,
+      "loss": 0.7821,
+      "step": 2736
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4281564581831536e-06,
+      "loss": 0.5233,
+      "step": 2737
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4277526168101336e-06,
+      "loss": 1.1693,
+      "step": 2738
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4273486664982804e-06,
+      "loss": 1.1423,
+      "step": 2739
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.426944607295027e-06,
+      "loss": 0.9868,
+      "step": 2740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.426540439247818e-06,
+      "loss": 0.8488,
+      "step": 2741
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4261361624041127e-06,
+      "loss": 0.9562,
+      "step": 2742
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4257317768113817e-06,
+      "loss": 1.1164,
+      "step": 2743
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.425327282517109e-06,
+      "loss": 0.8914,
+      "step": 2744
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.424922679568791e-06,
+      "loss": 0.6564,
+      "step": 2745
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.424517968013938e-06,
+      "loss": 1.1139,
+      "step": 2746
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4241131479000713e-06,
+      "loss": 0.9337,
+      "step": 2747
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.423708219274726e-06,
+      "loss": 0.8733,
+      "step": 2748
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.42330318218545e-06,
+      "loss": 0.9905,
+      "step": 2749
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4228980366798034e-06,
+      "loss": 0.6101,
+      "step": 2750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4224927828053593e-06,
+      "loss": 0.7379,
+      "step": 2751
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.422087420609703e-06,
+      "loss": 0.9978,
+      "step": 2752
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4216819501404347e-06,
+      "loss": 0.881,
+      "step": 2753
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4212763714451636e-06,
+      "loss": 0.92,
+      "step": 2754
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4208706845715155e-06,
+      "loss": 0.8042,
+      "step": 2755
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.420464889567126e-06,
+      "loss": 1.0109,
+      "step": 2756
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.420058986479645e-06,
+      "loss": 1.2223,
+      "step": 2757
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.419652975356734e-06,
+      "loss": 0.9562,
+      "step": 2758
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4192468562460684e-06,
+      "loss": 0.9775,
+      "step": 2759
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.418840629195335e-06,
+      "loss": 0.8919,
+      "step": 2760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4184342942522347e-06,
+      "loss": 1.0268,
+      "step": 2761
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4180278514644797e-06,
+      "loss": 1.0008,
+      "step": 2762
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4176213008797955e-06,
+      "loss": 1.0289,
+      "step": 2763
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4172146425459202e-06,
+      "loss": 0.9083,
+      "step": 2764
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4168078765106047e-06,
+      "loss": 0.8734,
+      "step": 2765
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.416401002821612e-06,
+      "loss": 0.761,
+      "step": 2766
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4159940215267186e-06,
+      "loss": 1.0514,
+      "step": 2767
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4155869326737133e-06,
+      "loss": 0.7746,
+      "step": 2768
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.415179736310397e-06,
+      "loss": 0.8407,
+      "step": 2769
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4147724324845834e-06,
+      "loss": 1.0831,
+      "step": 2770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4143650212441003e-06,
+      "loss": 0.9816,
+      "step": 2771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.413957502636785e-06,
+      "loss": 0.9427,
+      "step": 2772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4135498767104907e-06,
+      "loss": 0.8084,
+      "step": 2773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4131421435130812e-06,
+      "loss": 1.02,
+      "step": 2774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4127343030924336e-06,
+      "loss": 0.9371,
+      "step": 2775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4123263554964373e-06,
+      "loss": 0.833,
+      "step": 2776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4119183007729947e-06,
+      "loss": 1.1449,
+      "step": 2777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.41151013897002e-06,
+      "loss": 0.9315,
+      "step": 2778
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4111018701354413e-06,
+      "loss": 0.7589,
+      "step": 2779
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4106934943171977e-06,
+      "loss": 0.829,
+      "step": 2780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.410285011563242e-06,
+      "loss": 0.937,
+      "step": 2781
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4098764219215396e-06,
+      "loss": 1.1268,
+      "step": 2782
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.409467725440067e-06,
+      "loss": 0.8948,
+      "step": 2783
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.409058922166815e-06,
+      "loss": 0.8466,
+      "step": 2784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.408650012149786e-06,
+      "loss": 1.0918,
+      "step": 2785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.408240995436995e-06,
+      "loss": 1.0549,
+      "step": 2786
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.40783187207647e-06,
+      "loss": 0.9655,
+      "step": 2787
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4074226421162517e-06,
+      "loss": 1.1514,
+      "step": 2788
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4070133056043913e-06,
+      "loss": 0.8514,
+      "step": 2789
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4066038625889553e-06,
+      "loss": 1.0648,
+      "step": 2790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4061943131180217e-06,
+      "loss": 1.0354,
+      "step": 2791
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4057846572396786e-06,
+      "loss": 1.1063,
+      "step": 2792
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.405374895002031e-06,
+      "loss": 1.1498,
+      "step": 2793
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4049650264531938e-06,
+      "loss": 0.8577,
+      "step": 2794
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.404555051641294e-06,
+      "loss": 1.2796,
+      "step": 2795
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.404144970614472e-06,
+      "loss": 0.9405,
+      "step": 2796
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.40373478342088e-06,
+      "loss": 0.8857,
+      "step": 2797
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.403324490108684e-06,
+      "loss": 0.9809,
+      "step": 2798
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4029140907260614e-06,
+      "loss": 1.1944,
+      "step": 2799
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4025035853212013e-06,
+      "loss": 1.0692,
+      "step": 2800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4020929739423073e-06,
+      "loss": 0.98,
+      "step": 2801
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.401682256637594e-06,
+      "loss": 1.3334,
+      "step": 2802
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4012714334552885e-06,
+      "loss": 0.8398,
+      "step": 2803
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4008605044436306e-06,
+      "loss": 0.9892,
+      "step": 2804
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4004494696508734e-06,
+      "loss": 1.056,
+      "step": 2805
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.40003832912528e-06,
+      "loss": 1.0492,
+      "step": 2806
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3996270829151285e-06,
+      "loss": 1.1347,
+      "step": 2807
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3992157310687086e-06,
+      "loss": 0.7096,
+      "step": 2808
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3988042736343213e-06,
+      "loss": 1.0886,
+      "step": 2809
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3983927106602815e-06,
+      "loss": 0.9588,
+      "step": 2810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3979810421949156e-06,
+      "loss": 1.0585,
+      "step": 2811
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.397569268286563e-06,
+      "loss": 0.9141,
+      "step": 2812
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3971573889835743e-06,
+      "loss": 0.8462,
+      "step": 2813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.396745404334315e-06,
+      "loss": 0.9231,
+      "step": 2814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.396333314387159e-06,
+      "loss": 1.0558,
+      "step": 2815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3959211191904963e-06,
+      "loss": 1.1094,
+      "step": 2816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.395508818792728e-06,
+      "loss": 1.1278,
+      "step": 2817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3950964132422664e-06,
+      "loss": 1.0669,
+      "step": 2818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3946839025875376e-06,
+      "loss": 1.1374,
+      "step": 2819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3942712868769797e-06,
+      "loss": 1.0434,
+      "step": 2820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3938585661590426e-06,
+      "loss": 0.9158,
+      "step": 2821
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.393445740482189e-06,
+      "loss": 0.9039,
+      "step": 2822
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.393032809894894e-06,
+      "loss": 0.9629,
+      "step": 2823
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3926197744456453e-06,
+      "loss": 0.5956,
+      "step": 2824
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.392206634182941e-06,
+      "loss": 1.319,
+      "step": 2825
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3917933891552948e-06,
+      "loss": 1.0015,
+      "step": 2826
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3913800394112296e-06,
+      "loss": 0.9611,
+      "step": 2827
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3909665849992824e-06,
+      "loss": 1.0829,
+      "step": 2828
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3905530259680017e-06,
+      "loss": 1.3606,
+      "step": 2829
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3901393623659488e-06,
+      "loss": 0.8973,
+      "step": 2830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.389725594241697e-06,
+      "loss": 0.901,
+      "step": 2831
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3893117216438316e-06,
+      "loss": 1.1439,
+      "step": 2832
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3888977446209505e-06,
+      "loss": 1.0576,
+      "step": 2833
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3884836632216645e-06,
+      "loss": 1.1535,
+      "step": 2834
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3880694774945954e-06,
+      "loss": 0.7992,
+      "step": 2835
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.387655187488378e-06,
+      "loss": 0.9871,
+      "step": 2836
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3872407932516584e-06,
+      "loss": 0.9733,
+      "step": 2837
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.386826294833097e-06,
+      "loss": 1.023,
+      "step": 2838
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3864116922813646e-06,
+      "loss": 0.964,
+      "step": 2839
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.385996985645145e-06,
+      "loss": 0.6308,
+      "step": 2840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.385582174973133e-06,
+      "loss": 0.9484,
+      "step": 2841
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3851672603140373e-06,
+      "loss": 0.4798,
+      "step": 2842
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.384752241716579e-06,
+      "loss": 0.918,
+      "step": 2843
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3843371192294894e-06,
+      "loss": 1.0833,
+      "step": 2844
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3839218929015136e-06,
+      "loss": 0.8086,
+      "step": 2845
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3835065627814082e-06,
+      "loss": 1.6204,
+      "step": 2846
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.383091128917942e-06,
+      "loss": 0.6789,
+      "step": 2847
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.382675591359897e-06,
+      "loss": 0.7261,
+      "step": 2848
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3822599501560656e-06,
+      "loss": 0.6745,
+      "step": 2849
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.381844205355254e-06,
+      "loss": 0.8865,
+      "step": 2850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.38142835700628e-06,
+      "loss": 1.1586,
+      "step": 2851
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3810124051579727e-06,
+      "loss": 1.0325,
+      "step": 2852
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3805963498591746e-06,
+      "loss": 0.7227,
+      "step": 2853
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3801801911587402e-06,
+      "loss": 0.7488,
+      "step": 2854
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3797639291055356e-06,
+      "loss": 1.1725,
+      "step": 2855
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.379347563748439e-06,
+      "loss": 1.06,
+      "step": 2856
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3789310951363406e-06,
+      "loss": 1.0399,
+      "step": 2857
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.378514523318144e-06,
+      "loss": 0.9466,
+      "step": 2858
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3780978483427637e-06,
+      "loss": 0.6784,
+      "step": 2859
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.377681070259126e-06,
+      "loss": 1.1496,
+      "step": 2860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3772641891161707e-06,
+      "loss": 1.0703,
+      "step": 2861
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3768472049628486e-06,
+      "loss": 0.7102,
+      "step": 2862
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.376430117848123e-06,
+      "loss": 1.3137,
+      "step": 2863
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.37601292782097e-06,
+      "loss": 0.634,
+      "step": 2864
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3755956349303753e-06,
+      "loss": 1.1246,
+      "step": 2865
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.37517823922534e-06,
+      "loss": 0.8312,
+      "step": 2866
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.374760740754874e-06,
+      "loss": 0.7662,
+      "step": 2867
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3743431395680027e-06,
+      "loss": 1.0413,
+      "step": 2868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3739254357137615e-06,
+      "loss": 1.0735,
+      "step": 2869
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3735076292411967e-06,
+      "loss": 1.0198,
+      "step": 2870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3730897201993693e-06,
+      "loss": 1.1502,
+      "step": 2871
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3726717086373512e-06,
+      "loss": 1.0561,
+      "step": 2872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.372253594604226e-06,
+      "loss": 1.2673,
+      "step": 2873
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.371835378149089e-06,
+      "loss": 0.9224,
+      "step": 2874
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3714170593210484e-06,
+      "loss": 0.9505,
+      "step": 2875
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.370998638169225e-06,
+      "loss": 0.8496,
+      "step": 2876
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3705801147427508e-06,
+      "loss": 0.9709,
+      "step": 2877
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.370161489090768e-06,
+      "loss": 1.0023,
+      "step": 2878
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3697427612624345e-06,
+      "loss": 0.7389,
+      "step": 2879
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3693239313069177e-06,
+      "loss": 0.8942,
+      "step": 2880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3689049992733967e-06,
+      "loss": 1.0081,
+      "step": 2881
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3684859652110646e-06,
+      "loss": 1.0597,
+      "step": 2882
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.368066829169125e-06,
+      "loss": 0.8788,
+      "step": 2883
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3676475911967937e-06,
+      "loss": 1.0701,
+      "step": 2884
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3672282513432987e-06,
+      "loss": 1.055,
+      "step": 2885
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.366808809657879e-06,
+      "loss": 0.9614,
+      "step": 2886
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.366389266189788e-06,
+      "loss": 0.8948,
+      "step": 2887
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.365969620988288e-06,
+      "loss": 1.2991,
+      "step": 2888
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3655498741026557e-06,
+      "loss": 1.0858,
+      "step": 2889
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3651300255821784e-06,
+      "loss": 1.0444,
+      "step": 2890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3647100754761556e-06,
+      "loss": 1.0012,
+      "step": 2891
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3642900238338987e-06,
+      "loss": 0.9467,
+      "step": 2892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3638698707047307e-06,
+      "loss": 0.7468,
+      "step": 2893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3634496161379885e-06,
+      "loss": 0.9432,
+      "step": 2894
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3630292601830175e-06,
+      "loss": 0.8355,
+      "step": 2895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.362608802889178e-06,
+      "loss": 1.2786,
+      "step": 2896
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.362188244305841e-06,
+      "loss": 1.1297,
+      "step": 2897
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3617675844823888e-06,
+      "loss": 0.9534,
+      "step": 2898
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.361346823468217e-06,
+      "loss": 0.5672,
+      "step": 2899
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3609259613127315e-06,
+      "loss": 1.0009,
+      "step": 2900
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 1.25594961643219,
+      "eval_runtime": 32.2074,
+      "eval_samples_per_second": 38.656,
+      "eval_steps_per_second": 19.343,
+      "step": 2900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3605049980653514e-06,
+      "loss": 1.1159,
+      "step": 2901
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3600839337755075e-06,
+      "loss": 0.9947,
+      "step": 2902
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3596627684926416e-06,
+      "loss": 1.0054,
+      "step": 2903
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.359241502266208e-06,
+      "loss": 1.0574,
+      "step": 2904
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3588201351456724e-06,
+      "loss": 0.9696,
+      "step": 2905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.358398667180514e-06,
+      "loss": 0.9578,
+      "step": 2906
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.357977098420221e-06,
+      "loss": 1.3099,
+      "step": 2907
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.357555428914295e-06,
+      "loss": 1.2034,
+      "step": 2908
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.357133658712251e-06,
+      "loss": 0.9202,
+      "step": 2909
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3567117878636127e-06,
+      "loss": 1.1987,
+      "step": 2910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3562898164179177e-06,
+      "loss": 0.5331,
+      "step": 2911
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.355867744424715e-06,
+      "loss": 0.7917,
+      "step": 2912
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3554455719335647e-06,
+      "loss": 0.7829,
+      "step": 2913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.35502329899404e-06,
+      "loss": 0.967,
+      "step": 2914
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.354600925655724e-06,
+      "loss": 0.892,
+      "step": 2915
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3541784519682143e-06,
+      "loss": 0.8996,
+      "step": 2916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3537558779811176e-06,
+      "loss": 1.1004,
+      "step": 2917
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.353333203744054e-06,
+      "loss": 0.7529,
+      "step": 2918
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3529104293066543e-06,
+      "loss": 0.7379,
+      "step": 2919
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.352487554718562e-06,
+      "loss": 0.7765,
+      "step": 2920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3520645800294324e-06,
+      "loss": 0.8713,
+      "step": 2921
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.351641505288932e-06,
+      "loss": 1.0416,
+      "step": 2922
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.351218330546739e-06,
+      "loss": 0.9741,
+      "step": 2923
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.350795055852543e-06,
+      "loss": 0.9232,
+      "step": 2924
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3503716812560467e-06,
+      "loss": 1.1406,
+      "step": 2925
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3499482068069635e-06,
+      "loss": 0.8872,
+      "step": 2926
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.349524632555019e-06,
+      "loss": 0.9954,
+      "step": 2927
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3491009585499492e-06,
+      "loss": 0.916,
+      "step": 2928
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3486771848415042e-06,
+      "loss": 1.3307,
+      "step": 2929
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3482533114794437e-06,
+      "loss": 1.0902,
+      "step": 2930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3478293385135397e-06,
+      "loss": 0.9234,
+      "step": 2931
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3474052659935774e-06,
+      "loss": 0.8208,
+      "step": 2932
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3469810939693505e-06,
+      "loss": 1.141,
+      "step": 2933
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3465568224906675e-06,
+      "loss": 0.9672,
+      "step": 2934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3461324516073467e-06,
+      "loss": 1.0003,
+      "step": 2935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3457079813692194e-06,
+      "loss": 0.8266,
+      "step": 2936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.345283411826127e-06,
+      "loss": 1.0491,
+      "step": 2937
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3448587430279243e-06,
+      "loss": 1.2875,
+      "step": 2938
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.344433975024476e-06,
+      "loss": 0.8332,
+      "step": 2939
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3440091078656605e-06,
+      "loss": 0.9092,
+      "step": 2940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3435841416013653e-06,
+      "loss": 0.847,
+      "step": 2941
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.343159076281492e-06,
+      "loss": 0.9053,
+      "step": 2942
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3427339119559524e-06,
+      "loss": 0.7659,
+      "step": 2943
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.34230864867467e-06,
+      "loss": 0.8783,
+      "step": 2944
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.34188328648758e-06,
+      "loss": 1.0908,
+      "step": 2945
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3414578254446298e-06,
+      "loss": 1.061,
+      "step": 2946
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3410322655957786e-06,
+      "loss": 0.892,
+      "step": 2947
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3406066069909952e-06,
+      "loss": 0.9572,
+      "step": 2948
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3401808496802627e-06,
+      "loss": 0.9294,
+      "step": 2949
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3397549937135737e-06,
+      "loss": 0.9175,
+      "step": 2950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3393290391409338e-06,
+      "loss": 1.285,
+      "step": 2951
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.338902986012359e-06,
+      "loss": 0.8222,
+      "step": 2952
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3384768343778777e-06,
+      "loss": 1.0977,
+      "step": 2953
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3380505842875296e-06,
+      "loss": 0.9696,
+      "step": 2954
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3376242357913657e-06,
+      "loss": 0.9841,
+      "step": 2955
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3371977889394496e-06,
+      "loss": 0.7337,
+      "step": 2956
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3367712437818544e-06,
+      "loss": 0.8417,
+      "step": 2957
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3363446003686677e-06,
+      "loss": 0.8971,
+      "step": 2958
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3359178587499847e-06,
+      "loss": 0.8332,
+      "step": 2959
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3354910189759164e-06,
+      "loss": 0.9011,
+      "step": 2960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3350640810965824e-06,
+      "loss": 1.1907,
+      "step": 2961
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3346370451621146e-06,
+      "loss": 0.958,
+      "step": 2962
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3342099112226575e-06,
+      "loss": 1.1906,
+      "step": 2963
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3337826793283646e-06,
+      "loss": 1.2538,
+      "step": 2964
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3333553495294033e-06,
+      "loss": 1.2149,
+      "step": 2965
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3329279218759524e-06,
+      "loss": 1.0179,
+      "step": 2966
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3325003964182e-06,
+      "loss": 0.9796,
+      "step": 2967
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3320727732063477e-06,
+      "loss": 0.8586,
+      "step": 2968
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3316450522906087e-06,
+      "loss": 0.6425,
+      "step": 2969
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3312172337212057e-06,
+      "loss": 1.1595,
+      "step": 2970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.330789317548375e-06,
+      "loss": 1.3341,
+      "step": 2971
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.330361303822363e-06,
+      "loss": 0.925,
+      "step": 2972
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3299331925934284e-06,
+      "loss": 1.059,
+      "step": 2973
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.329504983911841e-06,
+      "loss": 0.6899,
+      "step": 2974
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3290766778278827e-06,
+      "loss": 0.8483,
+      "step": 2975
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3286482743918447e-06,
+      "loss": 1.09,
+      "step": 2976
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3282197736540315e-06,
+      "loss": 0.8289,
+      "step": 2977
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3277911756647598e-06,
+      "loss": 0.7575,
+      "step": 2978
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.327362480474355e-06,
+      "loss": 1.2306,
+      "step": 2979
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3269336881331565e-06,
+      "loss": 0.8603,
+      "step": 2980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.326504798691514e-06,
+      "loss": 0.984,
+      "step": 2981
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.326075812199788e-06,
+      "loss": 0.769,
+      "step": 2982
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.325646728708352e-06,
+      "loss": 0.8422,
+      "step": 2983
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.325217548267589e-06,
+      "loss": 0.7701,
+      "step": 2984
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3247882709278956e-06,
+      "loss": 1.311,
+      "step": 2985
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.324358896739677e-06,
+      "loss": 0.6014,
+      "step": 2986
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3239294257533526e-06,
+      "loss": 1.0488,
+      "step": 2987
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.323499858019351e-06,
+      "loss": 1.1331,
+      "step": 2988
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.323070193588113e-06,
+      "loss": 1.1757,
+      "step": 2989
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3226404325100923e-06,
+      "loss": 1.0591,
+      "step": 2990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3222105748357505e-06,
+      "loss": 1.022,
+      "step": 2991
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3217806206155638e-06,
+      "loss": 1.1803,
+      "step": 2992
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.321350569900017e-06,
+      "loss": 0.799,
+      "step": 2993
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.320920422739609e-06,
+      "loss": 1.0947,
+      "step": 2994
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3204901791848477e-06,
+      "loss": 0.8649,
+      "step": 2995
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3200598392862545e-06,
+      "loss": 0.9351,
+      "step": 2996
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3196294030943593e-06,
+      "loss": 1.0317,
+      "step": 2997
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3191988706597066e-06,
+      "loss": 0.8338,
+      "step": 2998
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.318768242032849e-06,
+      "loss": 0.9996,
+      "step": 2999
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3183375172643526e-06,
+      "loss": 0.8244,
+      "step": 3000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.317906696404794e-06,
+      "loss": 1.0023,
+      "step": 3001
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.317475779504761e-06,
+      "loss": 0.9649,
+      "step": 3002
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3170447666148536e-06,
+      "loss": 0.9936,
+      "step": 3003
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.316613657785682e-06,
+      "loss": 1.0419,
+      "step": 3004
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.316182453067867e-06,
+      "loss": 0.8273,
+      "step": 3005
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3157511525120426e-06,
+      "loss": 0.9323,
+      "step": 3006
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3153197561688532e-06,
+      "loss": 0.8961,
+      "step": 3007
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.314888264088953e-06,
+      "loss": 0.8547,
+      "step": 3008
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.314456676323011e-06,
+      "loss": 0.9892,
+      "step": 3009
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3140249929217034e-06,
+      "loss": 0.8362,
+      "step": 3010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3135932139357207e-06,
+      "loss": 1.0033,
+      "step": 3011
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3131613394157617e-06,
+      "loss": 1.3095,
+      "step": 3012
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.31272936941254e-06,
+      "loss": 0.7612,
+      "step": 3013
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.312297303976777e-06,
+      "loss": 0.9912,
+      "step": 3014
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.311865143159208e-06,
+      "loss": 0.9927,
+      "step": 3015
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.311432887010578e-06,
+      "loss": 1.0486,
+      "step": 3016
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.311000535581643e-06,
+      "loss": 1.1285,
+      "step": 3017
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.310568088923171e-06,
+      "loss": 0.6829,
+      "step": 3018
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.310135547085941e-06,
+      "loss": 0.67,
+      "step": 3019
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3097029101207427e-06,
+      "loss": 0.9994,
+      "step": 3020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3092701780783773e-06,
+      "loss": 0.8873,
+      "step": 3021
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3088373510096578e-06,
+      "loss": 1.1709,
+      "step": 3022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3084044289654073e-06,
+      "loss": 1.0295,
+      "step": 3023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.30797141199646e-06,
+      "loss": 0.6688,
+      "step": 3024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.307538300153663e-06,
+      "loss": 0.9388,
+      "step": 3025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3071050934878724e-06,
+      "loss": 0.8683,
+      "step": 3026
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3066717920499563e-06,
+      "loss": 0.9259,
+      "step": 3027
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.306238395890794e-06,
+      "loss": 0.9722,
+      "step": 3028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.305804905061277e-06,
+      "loss": 1.0382,
+      "step": 3029
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.305371319612305e-06,
+      "loss": 0.8018,
+      "step": 3030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3049376395947908e-06,
+      "loss": 0.8077,
+      "step": 3031
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3045038650596596e-06,
+      "loss": 0.8918,
+      "step": 3032
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3040699960578453e-06,
+      "loss": 0.5507,
+      "step": 3033
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3036360326402935e-06,
+      "loss": 0.8343,
+      "step": 3034
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3032019748579614e-06,
+      "loss": 1.2425,
+      "step": 3035
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3027678227618174e-06,
+      "loss": 0.9894,
+      "step": 3036
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.30233357640284e-06,
+      "loss": 1.2346,
+      "step": 3037
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.30189923583202e-06,
+      "loss": 1.2687,
+      "step": 3038
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.301464801100358e-06,
+      "loss": 0.9123,
+      "step": 3039
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3010302722588674e-06,
+      "loss": 1.0863,
+      "step": 3040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.300595649358571e-06,
+      "loss": 1.1948,
+      "step": 3041
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.300160932450502e-06,
+      "loss": 0.7198,
+      "step": 3042
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.2997261215857076e-06,
+      "loss": 1.2074,
+      "step": 3043
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.299291216815244e-06,
+      "loss": 1.1621,
+      "step": 3044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.298856218190178e-06,
+      "loss": 0.8367,
+      "step": 3045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2984211257615882e-06,
+      "loss": 1.028,
+      "step": 3046
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2979859395805646e-06,
+      "loss": 0.9555,
+      "step": 3047
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.297550659698208e-06,
+      "loss": 0.6628,
+      "step": 3048
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.297115286165629e-06,
+      "loss": 0.9246,
+      "step": 3049
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2966798190339513e-06,
+      "loss": 1.0204,
+      "step": 3050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.296244258354308e-06,
+      "loss": 0.6101,
+      "step": 3051
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2958086041778426e-06,
+      "loss": 0.8934,
+      "step": 3052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.295372856555712e-06,
+      "loss": 0.7568,
+      "step": 3053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.294937015539082e-06,
+      "loss": 0.6779,
+      "step": 3054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2945010811791303e-06,
+      "loss": 1.107,
+      "step": 3055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2940650535270454e-06,
+      "loss": 0.9597,
+      "step": 3056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.293628932634026e-06,
+      "loss": 1.0839,
+      "step": 3057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2931927185512837e-06,
+      "loss": 0.9173,
+      "step": 3058
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2927564113300386e-06,
+      "loss": 0.9196,
+      "step": 3059
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.292320011021523e-06,
+      "loss": 1.2273,
+      "step": 3060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2918835176769803e-06,
+      "loss": 1.0475,
+      "step": 3061
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.291446931347665e-06,
+      "loss": 1.0722,
+      "step": 3062
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2910102520848415e-06,
+      "loss": 0.7268,
+      "step": 3063
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2905734799397854e-06,
+      "loss": 1.0647,
+      "step": 3064
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.290136614963784e-06,
+      "loss": 1.1908,
+      "step": 3065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2896996572081353e-06,
+      "loss": 0.9202,
+      "step": 3066
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2892626067241466e-06,
+      "loss": 1.0727,
+      "step": 3067
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2888254635631386e-06,
+      "loss": 1.0912,
+      "step": 3068
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2883882277764414e-06,
+      "loss": 1.0252,
+      "step": 3069
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.287950899415396e-06,
+      "loss": 1.2209,
+      "step": 3070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2875134785313545e-06,
+      "loss": 0.7629,
+      "step": 3071
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2870759651756798e-06,
+      "loss": 1.116,
+      "step": 3072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.286638359399746e-06,
+      "loss": 1.0799,
+      "step": 3073
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2862006612549374e-06,
+      "loss": 0.7917,
+      "step": 3074
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.28576287079265e-06,
+      "loss": 0.8531,
+      "step": 3075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2853249880642898e-06,
+      "loss": 0.9548,
+      "step": 3076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2848870131212737e-06,
+      "loss": 1.1134,
+      "step": 3077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2844489460150306e-06,
+      "loss": 0.9525,
+      "step": 3078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2840107867969987e-06,
+      "loss": 0.865,
+      "step": 3079
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2835725355186277e-06,
+      "loss": 0.903,
+      "step": 3080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.283134192231378e-06,
+      "loss": 0.7751,
+      "step": 3081
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2826957569867214e-06,
+      "loss": 1.4039,
+      "step": 3082
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.282257229836139e-06,
+      "loss": 1.1211,
+      "step": 3083
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2818186108311247e-06,
+      "loss": 0.8848,
+      "step": 3084
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.281379900023182e-06,
+      "loss": 0.841,
+      "step": 3085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2809410974638246e-06,
+      "loss": 0.9771,
+      "step": 3086
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.280502203204578e-06,
+      "loss": 0.9607,
+      "step": 3087
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2800632172969785e-06,
+      "loss": 0.9655,
+      "step": 3088
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2796241397925727e-06,
+      "loss": 1.2565,
+      "step": 3089
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.279184970742918e-06,
+      "loss": 0.7441,
+      "step": 3090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2787457101995825e-06,
+      "loss": 0.5813,
+      "step": 3091
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2783063582141454e-06,
+      "loss": 0.7168,
+      "step": 3092
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.277866914838196e-06,
+      "loss": 1.2366,
+      "step": 3093
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2774273801233354e-06,
+      "loss": 0.9876,
+      "step": 3094
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2769877541211746e-06,
+      "loss": 0.8084,
+      "step": 3095
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2765480368833354e-06,
+      "loss": 0.9772,
+      "step": 3096
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.27610822846145e-06,
+      "loss": 0.7533,
+      "step": 3097
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.275668328907162e-06,
+      "loss": 0.8046,
+      "step": 3098
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2752283382721254e-06,
+      "loss": 0.927,
+      "step": 3099
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2747882566080057e-06,
+      "loss": 0.6188,
+      "step": 3100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2743480839664765e-06,
+      "loss": 1.0142,
+      "step": 3101
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.273907820399226e-06,
+      "loss": 0.9636,
+      "step": 3102
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2734674659579493e-06,
+      "loss": 0.9135,
+      "step": 3103
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2730270206943547e-06,
+      "loss": 0.7435,
+      "step": 3104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2725864846601594e-06,
+      "loss": 0.9903,
+      "step": 3105
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2721458579070935e-06,
+      "loss": 1.2405,
+      "step": 3106
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2717051404868955e-06,
+      "loss": 0.8921,
+      "step": 3107
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2712643324513156e-06,
+      "loss": 0.994,
+      "step": 3108
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2708234338521144e-06,
+      "loss": 0.7448,
+      "step": 3109
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2703824447410634e-06,
+      "loss": 0.8288,
+      "step": 3110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2699413651699445e-06,
+      "loss": 0.9893,
+      "step": 3111
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.26950019519055e-06,
+      "loss": 1.0783,
+      "step": 3112
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2690589348546837e-06,
+      "loss": 0.8984,
+      "step": 3113
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2686175842141587e-06,
+      "loss": 0.7797,
+      "step": 3114
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2681761433208e-06,
+      "loss": 0.8078,
+      "step": 3115
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2677346122264417e-06,
+      "loss": 0.9046,
+      "step": 3116
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2672929909829306e-06,
+      "loss": 1.0483,
+      "step": 3117
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.266851279642122e-06,
+      "loss": 0.8725,
+      "step": 3118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.266409478255884e-06,
+      "loss": 1.1602,
+      "step": 3119
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.265967586876091e-06,
+      "loss": 1.1254,
+      "step": 3120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2655256055546344e-06,
+      "loss": 1.1054,
+      "step": 3121
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2650835343434104e-06,
+      "loss": 0.9248,
+      "step": 3122
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2646413732943282e-06,
+      "loss": 1.1161,
+      "step": 3123
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2641991224593085e-06,
+      "loss": 0.7349,
+      "step": 3124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2637567818902804e-06,
+      "loss": 0.8353,
+      "step": 3125
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.263314351639185e-06,
+      "loss": 0.8598,
+      "step": 3126
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.262871831757973e-06,
+      "loss": 1.0044,
+      "step": 3127
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2624292222986067e-06,
+      "loss": 0.8585,
+      "step": 3128
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.261986523313058e-06,
+      "loss": 1.1172,
+      "step": 3129
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2615437348533098e-06,
+      "loss": 0.9341,
+      "step": 3130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.261100856971355e-06,
+      "loss": 1.049,
+      "step": 3131
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2606578897191973e-06,
+      "loss": 0.8722,
+      "step": 3132
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2602148331488515e-06,
+      "loss": 1.0746,
+      "step": 3133
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.259771687312342e-06,
+      "loss": 1.1996,
+      "step": 3134
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2593284522617035e-06,
+      "loss": 1.181,
+      "step": 3135
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2588851280489826e-06,
+      "loss": 0.9254,
+      "step": 3136
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.258441714726235e-06,
+      "loss": 1.0153,
+      "step": 3137
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.257998212345527e-06,
+      "loss": 1.0286,
+      "step": 3138
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2575546209589365e-06,
+      "loss": 0.8045,
+      "step": 3139
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.25711094061855e-06,
+      "loss": 1.245,
+      "step": 3140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.256667171376466e-06,
+      "loss": 0.8488,
+      "step": 3141
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2562233132847922e-06,
+      "loss": 1.2715,
+      "step": 3142
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.255779366395649e-06,
+      "loss": 0.841,
+      "step": 3143
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.255335330761164e-06,
+      "loss": 1.244,
+      "step": 3144
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2548912064334787e-06,
+      "loss": 1.2514,
+      "step": 3145
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.254446993464741e-06,
+      "loss": 1.0925,
+      "step": 3146
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2540026919071126e-06,
+      "loss": 0.7785,
+      "step": 3147
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2535583018127647e-06,
+      "loss": 0.9193,
+      "step": 3148
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2531138232338784e-06,
+      "loss": 1.0248,
+      "step": 3149
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2526692562226446e-06,
+      "loss": 2.0007,
+      "step": 3150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.252224600831266e-06,
+      "loss": 0.8571,
+      "step": 3151
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.251779857111955e-06,
+      "loss": 1.0312,
+      "step": 3152
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.251335025116935e-06,
+      "loss": 0.9908,
+      "step": 3153
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.250890104898438e-06,
+      "loss": 0.6948,
+      "step": 3154
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2504450965087087e-06,
+      "loss": 0.9749,
+      "step": 3155
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.25e-06,
+      "loss": 0.8249,
+      "step": 3156
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2495548154245767e-06,
+      "loss": 1.0998,
+      "step": 3157
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2491095428347136e-06,
+      "loss": 0.9554,
+      "step": 3158
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2486641822826956e-06,
+      "loss": 0.8907,
+      "step": 3159
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2482187338208173e-06,
+      "loss": 1.3291,
+      "step": 3160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2477731975013846e-06,
+      "loss": 1.0687,
+      "step": 3161
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2473275733767137e-06,
+      "loss": 0.7591,
+      "step": 3162
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.24688186149913e-06,
+      "loss": 0.8987,
+      "step": 3163
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2464360619209713e-06,
+      "loss": 1.1798,
+      "step": 3164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2459901746945834e-06,
+      "loss": 1.0544,
+      "step": 3165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2455441998723235e-06,
+      "loss": 1.041,
+      "step": 3166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2450981375065594e-06,
+      "loss": 0.9923,
+      "step": 3167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2446519876496686e-06,
+      "loss": 1.0707,
+      "step": 3168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2442057503540385e-06,
+      "loss": 0.5406,
+      "step": 3169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2437594256720687e-06,
+      "loss": 1.1916,
+      "step": 3170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.243313013656166e-06,
+      "loss": 0.6468,
+      "step": 3171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.24286651435875e-06,
+      "loss": 1.2067,
+      "step": 3172
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2424199278322496e-06,
+      "loss": 0.7675,
+      "step": 3173
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.241973254129104e-06,
+      "loss": 1.0011,
+      "step": 3174
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2415264933017626e-06,
+      "loss": 0.6777,
+      "step": 3175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2410796454026847e-06,
+      "loss": 1.1471,
+      "step": 3176
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.240632710484341e-06,
+      "loss": 1.1318,
+      "step": 3177
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2401856885992105e-06,
+      "loss": 1.2874,
+      "step": 3178
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.239738579799785e-06,
+      "loss": 0.8189,
+      "step": 3179
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.239291384138564e-06,
+      "loss": 1.1578,
+      "step": 3180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2388441016680585e-06,
+      "loss": 1.0535,
+      "step": 3181
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.238396732440789e-06,
+      "loss": 1.058,
+      "step": 3182
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.2379492765092873e-06,
+      "loss": 1.0997,
+      "step": 3183
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.237501733926095e-06,
+      "loss": 0.7044,
+      "step": 3184
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.237054104743762e-06,
+      "loss": 0.857,
+      "step": 3185
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2366063890148515e-06,
+      "loss": 0.8356,
+      "step": 3186
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.236158586791935e-06,
+      "loss": 1.0593,
+      "step": 3187
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2357106981275938e-06,
+      "loss": 1.029,
+      "step": 3188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2352627230744203e-06,
+      "loss": 1.0948,
+      "step": 3189
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2348146616850175e-06,
+      "loss": 0.9512,
+      "step": 3190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2343665140119966e-06,
+      "loss": 1.0313,
+      "step": 3191
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2339182801079808e-06,
+      "loss": 0.8912,
+      "step": 3192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.233469960025603e-06,
+      "loss": 0.6939,
+      "step": 3193
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.233021553817505e-06,
+      "loss": 0.7215,
+      "step": 3194
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2325730615363404e-06,
+      "loss": 1.1967,
+      "step": 3195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2321244832347727e-06,
+      "loss": 0.8794,
+      "step": 3196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.231675818965474e-06,
+      "loss": 0.906,
+      "step": 3197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2312270687811277e-06,
+      "loss": 1.1438,
+      "step": 3198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2307782327344277e-06,
+      "loss": 1.0889,
+      "step": 3199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.230329310878076e-06,
+      "loss": 0.6664,
+      "step": 3200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2298803032647883e-06,
+      "loss": 0.8639,
+      "step": 3201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2294312099472855e-06,
+      "loss": 0.8491,
+      "step": 3202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2289820309783027e-06,
+      "loss": 0.9977,
+      "step": 3203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2285327664105835e-06,
+      "loss": 0.9592,
+      "step": 3204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.228083416296881e-06,
+      "loss": 1.0523,
+      "step": 3205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2276339806899597e-06,
+      "loss": 0.8888,
+      "step": 3206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2271844596425923e-06,
+      "loss": 1.0424,
+      "step": 3207
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.226734853207563e-06,
+      "loss": 0.9843,
+      "step": 3208
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.226285161437666e-06,
+      "loss": 0.9943,
+      "step": 3209
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2258353843857053e-06,
+      "loss": 0.7879,
+      "step": 3210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.225385522104494e-06,
+      "loss": 1.2037,
+      "step": 3211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2249355746468564e-06,
+      "loss": 1.247,
+      "step": 3212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.224485542065626e-06,
+      "loss": 1.0096,
+      "step": 3213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.224035424413647e-06,
+      "loss": 0.9435,
+      "step": 3214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.223585221743773e-06,
+      "loss": 0.8407,
+      "step": 3215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2231349341088685e-06,
+      "loss": 1.1747,
+      "step": 3216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.222684561561806e-06,
+      "loss": 0.8505,
+      "step": 3217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.222234104155471e-06,
+      "loss": 1.1108,
+      "step": 3218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2217835619427558e-06,
+      "loss": 0.7335,
+      "step": 3219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2213329349765642e-06,
+      "loss": 0.9187,
+      "step": 3220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2208822233098105e-06,
+      "loss": 0.6428,
+      "step": 3221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2204314269954185e-06,
+      "loss": 0.6475,
+      "step": 3222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.219980546086321e-06,
+      "loss": 0.6583,
+      "step": 3223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2195295806354617e-06,
+      "loss": 0.6895,
+      "step": 3224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.219078530695794e-06,
+      "loss": 0.9842,
+      "step": 3225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.218627396320281e-06,
+      "loss": 0.4529,
+      "step": 3226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2181761775618968e-06,
+      "loss": 0.9506,
+      "step": 3227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2177248744736233e-06,
+      "loss": 1.3852,
+      "step": 3228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.217273487108455e-06,
+      "loss": 1.25,
+      "step": 3229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2168220155193934e-06,
+      "loss": 0.677,
+      "step": 3230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2163704597594516e-06,
+      "loss": 0.9382,
+      "step": 3231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.215918819881653e-06,
+      "loss": 0.8199,
+      "step": 3232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.21546709593903e-06,
+      "loss": 0.899,
+      "step": 3233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.215015287984625e-06,
+      "loss": 0.8132,
+      "step": 3234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2145633960714897e-06,
+      "loss": 0.941,
+      "step": 3235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.214111420252687e-06,
+      "loss": 1.0979,
+      "step": 3236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2136593605812887e-06,
+      "loss": 0.8068,
+      "step": 3237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2132072171103765e-06,
+      "loss": 0.9362,
+      "step": 3238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2127549898930424e-06,
+      "loss": 0.8687,
+      "step": 3239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.212302678982388e-06,
+      "loss": 0.7137,
+      "step": 3240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2118502844315248e-06,
+      "loss": 0.7887,
+      "step": 3241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2113978062935736e-06,
+      "loss": 0.9396,
+      "step": 3242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.210945244621665e-06,
+      "loss": 1.1603,
+      "step": 3243
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.210492599468941e-06,
+      "loss": 0.8271,
+      "step": 3244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2100398708885516e-06,
+      "loss": 1.0228,
+      "step": 3245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2095870589336567e-06,
+      "loss": 1.1088,
+      "step": 3246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2091341636574277e-06,
+      "loss": 0.9239,
+      "step": 3247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2086811851130437e-06,
+      "loss": 1.256,
+      "step": 3248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2082281233536946e-06,
+      "loss": 1.1271,
+      "step": 3249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.20777497843258e-06,
+      "loss": 1.0979,
+      "step": 3250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.20732175040291e-06,
+      "loss": 1.1538,
+      "step": 3251
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2068684393179024e-06,
+      "loss": 1.1245,
+      "step": 3252
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.206415045230787e-06,
+      "loss": 1.0046,
+      "step": 3253
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.205961568194801e-06,
+      "loss": 0.9606,
+      "step": 3254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2055080082631945e-06,
+      "loss": 0.8746,
+      "step": 3255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.205054365489225e-06,
+      "loss": 1.0819,
+      "step": 3256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2046006399261593e-06,
+      "loss": 0.6138,
+      "step": 3257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2041468316272757e-06,
+      "loss": 0.9061,
+      "step": 3258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2036929406458613e-06,
+      "loss": 0.8142,
+      "step": 3259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.203238967035213e-06,
+      "loss": 1.1935,
+      "step": 3260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2027849108486377e-06,
+      "loss": 1.1475,
+      "step": 3261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.202330772139451e-06,
+      "loss": 0.9043,
+      "step": 3262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2018765509609804e-06,
+      "loss": 0.9634,
+      "step": 3263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2014222473665596e-06,
+      "loss": 0.9134,
+      "step": 3264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.200967861409535e-06,
+      "loss": 0.9466,
+      "step": 3265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2005133931432614e-06,
+      "loss": 0.6518,
+      "step": 3266
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.2000588426211037e-06,
+      "loss": 1.0202,
+      "step": 3267
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1996042098964366e-06,
+      "loss": 0.6723,
+      "step": 3268
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1991494950226436e-06,
+      "loss": 1.1041,
+      "step": 3269
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1986946980531185e-06,
+      "loss": 1.0805,
+      "step": 3270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1982398190412646e-06,
+      "loss": 0.9252,
+      "step": 3271
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1977848580404944e-06,
+      "loss": 1.03,
+      "step": 3272
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.197329815104231e-06,
+      "loss": 1.1671,
+      "step": 3273
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1968746902859066e-06,
+      "loss": 1.2739,
+      "step": 3274
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.196419483638963e-06,
+      "loss": 1.0133,
+      "step": 3275
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1959641952168515e-06,
+      "loss": 0.8917,
+      "step": 3276
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1955088250730326e-06,
+      "loss": 0.6023,
+      "step": 3277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.195053373260978e-06,
+      "loss": 0.6886,
+      "step": 3278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.194597839834167e-06,
+      "loss": 0.9861,
+      "step": 3279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.194142224846089e-06,
+      "loss": 0.8166,
+      "step": 3280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.193686528350245e-06,
+      "loss": 0.7682,
+      "step": 3281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1932307504001426e-06,
+      "loss": 0.9061,
+      "step": 3282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1927748910493004e-06,
+      "loss": 0.7433,
+      "step": 3283
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1923189503512464e-06,
+      "loss": 0.7482,
+      "step": 3284
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.191862928359519e-06,
+      "loss": 1.2112,
+      "step": 3285
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1914068251276644e-06,
+      "loss": 1.2907,
+      "step": 3286
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1909506407092397e-06,
+      "loss": 0.5191,
+      "step": 3287
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.190494375157811e-06,
+      "loss": 0.9416,
+      "step": 3288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.190038028526955e-06,
+      "loss": 0.7501,
+      "step": 3289
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.189581600870255e-06,
+      "loss": 0.5926,
+      "step": 3290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1891250922413077e-06,
+      "loss": 1.0254,
+      "step": 3291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.188668502693716e-06,
+      "loss": 1.0403,
+      "step": 3292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.188211832281095e-06,
+      "loss": 0.863,
+      "step": 3293
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1877550810570664e-06,
+      "loss": 0.9017,
+      "step": 3294
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.187298249075264e-06,
+      "loss": 0.7408,
+      "step": 3295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1868413363893302e-06,
+      "loss": 0.871,
+      "step": 3296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1863843430529163e-06,
+      "loss": 0.8362,
+      "step": 3297
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.185927269119684e-06,
+      "loss": 0.7266,
+      "step": 3298
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.1854701146433028e-06,
+      "loss": 1.1024,
+      "step": 3299
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.185012879677454e-06,
+      "loss": 0.9354,
+      "step": 3300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1845555642758265e-06,
+      "loss": 0.8416,
+      "step": 3301
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.18409816849212e-06,
+      "loss": 0.8314,
+      "step": 3302
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1836406923800425e-06,
+      "loss": 0.7754,
+      "step": 3303
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1831831359933118e-06,
+      "loss": 1.045,
+      "step": 3304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1827254993856555e-06,
+      "loss": 0.9573,
+      "step": 3305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.18226778261081e-06,
+      "loss": 0.9065,
+      "step": 3306
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1818099857225214e-06,
+      "loss": 1.1355,
+      "step": 3307
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.181352108774546e-06,
+      "loss": 0.7486,
+      "step": 3308
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.180894151820648e-06,
+      "loss": 0.9617,
+      "step": 3309
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.180436114914602e-06,
+      "loss": 0.7408,
+      "step": 3310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.179977998110192e-06,
+      "loss": 0.9454,
+      "step": 3311
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1795198014612105e-06,
+      "loss": 0.7289,
+      "step": 3312
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1790615250214605e-06,
+      "loss": 0.8744,
+      "step": 3313
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.178603168844754e-06,
+      "loss": 0.8713,
+      "step": 3314
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1781447329849117e-06,
+      "loss": 0.9182,
+      "step": 3315
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1776862174957648e-06,
+      "loss": 0.7701,
+      "step": 3316
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1772276224311526e-06,
+      "loss": 1.2508,
+      "step": 3317
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.176768947844925e-06,
+      "loss": 0.9591,
+      "step": 3318
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1763101937909397e-06,
+      "loss": 1.0209,
+      "step": 3319
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.175851360323066e-06,
+      "loss": 1.1636,
+      "step": 3320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1753924474951795e-06,
+      "loss": 1.0135,
+      "step": 3321
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1749334553611685e-06,
+      "loss": 0.9497,
+      "step": 3322
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1744743839749278e-06,
+      "loss": 1.0029,
+      "step": 3323
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1740152333903627e-06,
+      "loss": 1.0443,
+      "step": 3324
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1735560036613883e-06,
+      "loss": 1.0517,
+      "step": 3325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.173096694841928e-06,
+      "loss": 0.845,
+      "step": 3326
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.172637306985915e-06,
+      "loss": 0.7999,
+      "step": 3327
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1721778401472916e-06,
+      "loss": 0.8691,
+      "step": 3328
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1717182943800094e-06,
+      "loss": 1.0544,
+      "step": 3329
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1712586697380293e-06,
+      "loss": 0.967,
+      "step": 3330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1707989662753216e-06,
+      "loss": 1.0399,
+      "step": 3331
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.170339184045866e-06,
+      "loss": 0.9871,
+      "step": 3332
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1698793231036504e-06,
+      "loss": 1.5088,
+      "step": 3333
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1694193835026734e-06,
+      "loss": 0.957,
+      "step": 3334
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.168959365296942e-06,
+      "loss": 1.0246,
+      "step": 3335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.168499268540472e-06,
+      "loss": 1.3515,
+      "step": 3336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1680390932872904e-06,
+      "loss": 1.2265,
+      "step": 3337
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.167578839591431e-06,
+      "loss": 1.332,
+      "step": 3338
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1671185075069372e-06,
+      "loss": 1.0292,
+      "step": 3339
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.166658097087864e-06,
+      "loss": 1.2651,
+      "step": 3340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.166197608388272e-06,
+      "loss": 1.0351,
+      "step": 3341
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1657370414622346e-06,
+      "loss": 0.8858,
+      "step": 3342
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1652763963638313e-06,
+      "loss": 0.8591,
+      "step": 3343
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.164815673147153e-06,
+      "loss": 1.1305,
+      "step": 3344
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.164354871866298e-06,
+      "loss": 0.8707,
+      "step": 3345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1638939925753755e-06,
+      "loss": 1.3922,
+      "step": 3346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.163433035328502e-06,
+      "loss": 0.6778,
+      "step": 3347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.162972000179805e-06,
+      "loss": 0.7604,
+      "step": 3348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.16251088718342e-06,
+      "loss": 0.7646,
+      "step": 3349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.162049696393492e-06,
+      "loss": 0.9823,
+      "step": 3350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.161588427864175e-06,
+      "loss": 0.8479,
+      "step": 3351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1611270816496326e-06,
+      "loss": 1.0139,
+      "step": 3352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1606656578040364e-06,
+      "loss": 0.9188,
+      "step": 3353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1602041563815684e-06,
+      "loss": 0.7026,
+      "step": 3354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1597425774364193e-06,
+      "loss": 0.7501,
+      "step": 3355
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.159280921022788e-06,
+      "loss": 0.791,
+      "step": 3356
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.158819187194884e-06,
+      "loss": 1.1841,
+      "step": 3357
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1583573760069244e-06,
+      "loss": 0.9439,
+      "step": 3358
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1578954875131367e-06,
+      "loss": 1.0603,
+      "step": 3359
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.157433521767757e-06,
+      "loss": 0.9203,
+      "step": 3360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.15697147882503e-06,
+      "loss": 1.0065,
+      "step": 3361
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1565093587392098e-06,
+      "loss": 1.0733,
+      "step": 3362
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1560471615645602e-06,
+      "loss": 0.7426,
+      "step": 3363
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1555848873553524e-06,
+      "loss": 0.7463,
+      "step": 3364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1551225361658685e-06,
+      "loss": 1.0222,
+      "step": 3365
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.154660108050399e-06,
+      "loss": 0.9972,
+      "step": 3366
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.154197603063243e-06,
+      "loss": 1.0106,
+      "step": 3367
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.1537350212587085e-06,
+      "loss": 0.8653,
+      "step": 3368
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.153272362691113e-06,
+      "loss": 0.8868,
+      "step": 3369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.152809627414783e-06,
+      "loss": 1.1274,
+      "step": 3370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.152346815484055e-06,
+      "loss": 1.2307,
+      "step": 3371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.151883926953272e-06,
+      "loss": 0.8535,
+      "step": 3372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.151420961876787e-06,
+      "loss": 1.0054,
+      "step": 3373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1509579203089646e-06,
+      "loss": 1.0496,
+      "step": 3374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.150494802304175e-06,
+      "loss": 0.7666,
+      "step": 3375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.150031607916798e-06,
+      "loss": 0.9457,
+      "step": 3376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1495683372012236e-06,
+      "loss": 1.3713,
+      "step": 3377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.14910499021185e-06,
+      "loss": 1.1571,
+      "step": 3378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1486415670030847e-06,
+      "loss": 1.033,
+      "step": 3379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.148178067629343e-06,
+      "loss": 0.8902,
+      "step": 3380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1477144921450513e-06,
+      "loss": 0.8007,
+      "step": 3381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1472508406046428e-06,
+      "loss": 1.0163,
+      "step": 3382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.146787113062561e-06,
+      "loss": 0.9275,
+      "step": 3383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1463233095732578e-06,
+      "loss": 1.1013,
+      "step": 3384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1458594301911936e-06,
+      "loss": 0.9118,
+      "step": 3385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.145395474970839e-06,
+      "loss": 1.1859,
+      "step": 3386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.144931443966672e-06,
+      "loss": 1.3471,
+      "step": 3387
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1444673372331797e-06,
+      "loss": 0.9522,
+      "step": 3388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1440031548248605e-06,
+      "loss": 0.9282,
+      "step": 3389
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1435388967962177e-06,
+      "loss": 0.9898,
+      "step": 3390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1430745632017667e-06,
+      "loss": 1.137,
+      "step": 3391
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.14261015409603e-06,
+      "loss": 1.1428,
+      "step": 3392
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.1421456695335403e-06,
+      "loss": 1.0014,
+      "step": 3393
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1416811095688374e-06,
+      "loss": 0.9535,
+      "step": 3394
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.141216474256472e-06,
+      "loss": 1.3685,
+      "step": 3395
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1407517636510024e-06,
+      "loss": 0.9256,
+      "step": 3396
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1402869778069956e-06,
+      "loss": 1.0484,
+      "step": 3397
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1398221167790276e-06,
+      "loss": 0.9605,
+      "step": 3398
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.139357180621684e-06,
+      "loss": 1.2883,
+      "step": 3399
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1388921693895588e-06,
+      "loss": 0.8885,
+      "step": 3400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1384270831372547e-06,
+      "loss": 0.8501,
+      "step": 3401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1379619219193826e-06,
+      "loss": 0.919,
+      "step": 3402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.137496685790563e-06,
+      "loss": 0.7253,
+      "step": 3403
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.137031374805425e-06,
+      "loss": 0.9868,
+      "step": 3404
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.136565989018607e-06,
+      "loss": 1.102,
+      "step": 3405
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1361005284847544e-06,
+      "loss": 1.2095,
+      "step": 3406
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1356349932585243e-06,
+      "loss": 0.9872,
+      "step": 3407
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.13516938339458e-06,
+      "loss": 0.8888,
+      "step": 3408
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1347036989475945e-06,
+      "loss": 1.2411,
+      "step": 3409
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1342379399722493e-06,
+      "loss": 0.9965,
+      "step": 3410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1337721065232356e-06,
+      "loss": 0.8477,
+      "step": 3411
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.133306198655252e-06,
+      "loss": 0.9053,
+      "step": 3412
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.132840216423007e-06,
+      "loss": 1.1067,
+      "step": 3413
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.132374159881216e-06,
+      "loss": 1.0612,
+      "step": 3414
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1319080290846064e-06,
+      "loss": 0.9129,
+      "step": 3415
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.1314418240879107e-06,
+      "loss": 1.209,
+      "step": 3416
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1309755449458726e-06,
+      "loss": 0.8395,
+      "step": 3417
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1305091917132434e-06,
+      "loss": 0.5709,
+      "step": 3418
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.130042764444784e-06,
+      "loss": 0.9937,
+      "step": 3419
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.129576263195262e-06,
+      "loss": 0.9892,
+      "step": 3420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.129109688019456e-06,
+      "loss": 0.7849,
+      "step": 3421
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1286430389721523e-06,
+      "loss": 0.9343,
+      "step": 3422
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1281763161081456e-06,
+      "loss": 1.038,
+      "step": 3423
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.12770951948224e-06,
+      "loss": 1.0558,
+      "step": 3424
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1272426491492474e-06,
+      "loss": 1.334,
+      "step": 3425
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1267757051639894e-06,
+      "loss": 1.2139,
+      "step": 3426
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.126308687581294e-06,
+      "loss": 1.0162,
+      "step": 3427
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.125841596456001e-06,
+      "loss": 1.0943,
+      "step": 3428
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.125374431842958e-06,
+      "loss": 0.9962,
+      "step": 3429
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1249071937970185e-06,
+      "loss": 1.0939,
+      "step": 3430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.124439882373048e-06,
+      "loss": 1.0588,
+      "step": 3431
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1239724976259184e-06,
+      "loss": 0.9338,
+      "step": 3432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.123505039610512e-06,
+      "loss": 1.1362,
+      "step": 3433
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.123037508381718e-06,
+      "loss": 1.2055,
+      "step": 3434
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1225699039944357e-06,
+      "loss": 0.9658,
+      "step": 3435
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1221022265035715e-06,
+      "loss": 0.8397,
+      "step": 3436
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.121634475964042e-06,
+      "loss": 1.5609,
+      "step": 3437
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.1211666524307716e-06,
+      "loss": 0.6663,
+      "step": 3438
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.120698755958692e-06,
+      "loss": 0.9195,
+      "step": 3439
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1202307866027454e-06,
+      "loss": 1.1912,
+      "step": 3440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1197627444178824e-06,
+      "loss": 0.9285,
+      "step": 3441
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1192946294590606e-06,
+      "loss": 1.0135,
+      "step": 3442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1188264417812477e-06,
+      "loss": 0.9355,
+      "step": 3443
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.118358181439419e-06,
+      "loss": 0.9542,
+      "step": 3444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.117889848488559e-06,
+      "loss": 1.0196,
+      "step": 3445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1174214429836605e-06,
+      "loss": 1.1495,
+      "step": 3446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1169529649797248e-06,
+      "loss": 0.9865,
+      "step": 3447
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.116484414531761e-06,
+      "loss": 1.118,
+      "step": 3448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1160157916947885e-06,
+      "loss": 1.4312,
+      "step": 3449
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.115547096523833e-06,
+      "loss": 1.0783,
+      "step": 3450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.11507832907393e-06,
+      "loss": 0.7274,
+      "step": 3451
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.114609489400123e-06,
+      "loss": 0.7253,
+      "step": 3452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1141405775574653e-06,
+      "loss": 1.218,
+      "step": 3453
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.113671593601017e-06,
+      "loss": 1.1505,
+      "step": 3454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1132025375858467e-06,
+      "loss": 0.8591,
+      "step": 3455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1127334095670327e-06,
+      "loss": 1.02,
+      "step": 3456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.112264209599661e-06,
+      "loss": 1.0158,
+      "step": 3457
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.111794937738826e-06,
+      "loss": 0.8646,
+      "step": 3458
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.111325594039631e-06,
+      "loss": 1.0452,
+      "step": 3459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.110856178557187e-06,
+      "loss": 1.0788,
+      "step": 3460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1103866913466133e-06,
+      "loss": 1.223,
+      "step": 3461
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.1099171324630393e-06,
+      "loss": 0.8839,
+      "step": 3462
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1094475019616013e-06,
+      "loss": 1.1844,
+      "step": 3463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1089777998974438e-06,
+      "loss": 1.0687,
+      "step": 3464
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1085080263257218e-06,
+      "loss": 1.0343,
+      "step": 3465
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1080381813015953e-06,
+      "loss": 0.9698,
+      "step": 3466
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1075682648802354e-06,
+      "loss": 0.9962,
+      "step": 3467
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.107098277116821e-06,
+      "loss": 1.0555,
+      "step": 3468
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1066282180665395e-06,
+      "loss": 0.6337,
+      "step": 3469
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1061580877845845e-06,
+      "loss": 1.1165,
+      "step": 3470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.105687886326162e-06,
+      "loss": 1.0191,
+      "step": 3471
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1052176137464825e-06,
+      "loss": 0.8836,
+      "step": 3472
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.104747270100768e-06,
+      "loss": 0.7901,
+      "step": 3473
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1042768554442454e-06,
+      "loss": 0.9752,
+      "step": 3474
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1038063698321538e-06,
+      "loss": 0.9118,
+      "step": 3475
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1033358133197376e-06,
+      "loss": 1.0307,
+      "step": 3476
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1028651859622502e-06,
+      "loss": 1.0669,
+      "step": 3477
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1023944878149554e-06,
+      "loss": 1.1364,
+      "step": 3478
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.101923718933122e-06,
+      "loss": 0.9276,
+      "step": 3479
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1014528793720296e-06,
+      "loss": 0.7816,
+      "step": 3480
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 1.2555830478668213,
+      "eval_runtime": 32.1093,
+      "eval_samples_per_second": 38.774,
+      "eval_steps_per_second": 19.403,
+      "step": 3480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1009819691869647e-06,
+      "loss": 0.9084,
+      "step": 3481
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1005109884332236e-06,
+      "loss": 1.1883,
+      "step": 3482
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.1000399371661092e-06,
+      "loss": 0.6838,
+      "step": 3483
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.0995688154409334e-06,
+      "loss": 0.8309,
+      "step": 3484
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.0990976233130163e-06,
+      "loss": 0.771,
+      "step": 3485
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.098626360837687e-06,
+      "loss": 0.5415,
+      "step": 3486
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0981550280702813e-06,
+      "loss": 1.2632,
+      "step": 3487
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0976836250661453e-06,
+      "loss": 0.9579,
+      "step": 3488
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.097212151880631e-06,
+      "loss": 1.1008,
+      "step": 3489
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.096740608569101e-06,
+      "loss": 1.0163,
+      "step": 3490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.096268995186924e-06,
+      "loss": 1.5051,
+      "step": 3491
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0957973117894788e-06,
+      "loss": 0.6761,
+      "step": 3492
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.095325558432151e-06,
+      "loss": 0.8352,
+      "step": 3493
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.094853735170335e-06,
+      "loss": 0.8289,
+      "step": 3494
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0943818420594336e-06,
+      "loss": 1.0935,
+      "step": 3495
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0939098791548566e-06,
+      "loss": 0.6221,
+      "step": 3496
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.093437846512025e-06,
+      "loss": 0.5272,
+      "step": 3497
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.092965744186364e-06,
+      "loss": 0.9987,
+      "step": 3498
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.09249357223331e-06,
+      "loss": 1.1002,
+      "step": 3499
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.092021330708306e-06,
+      "loss": 0.8967,
+      "step": 3500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.091549019666804e-06,
+      "loss": 0.6201,
+      "step": 3501
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0910766391642632e-06,
+      "loss": 1.1211,
+      "step": 3502
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0906041892561523e-06,
+      "loss": 0.9487,
+      "step": 3503
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0901316699979477e-06,
+      "loss": 0.9628,
+      "step": 3504
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0896590814451325e-06,
+      "loss": 0.8315,
+      "step": 3505
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0891864236532007e-06,
+      "loss": 1.1357,
+      "step": 3506
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.088713696677651e-06,
+      "loss": 0.7443,
+      "step": 3507
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.0882409005739943e-06,
+      "loss": 0.9679,
+      "step": 3508
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0877680353977454e-06,
+      "loss": 0.8472,
+      "step": 3509
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0872951012044304e-06,
+      "loss": 0.5642,
+      "step": 3510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0868220980495815e-06,
+      "loss": 1.0703,
+      "step": 3511
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.086349025988741e-06,
+      "loss": 0.9932,
+      "step": 3512
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.085875885077457e-06,
+      "loss": 1.2928,
+      "step": 3513
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0854026753712863e-06,
+      "loss": 0.8754,
+      "step": 3514
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.084929396925796e-06,
+      "loss": 1.1539,
+      "step": 3515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0844560497965585e-06,
+      "loss": 0.9104,
+      "step": 3516
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.083982634039155e-06,
+      "loss": 1.2779,
+      "step": 3517
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.083509149709176e-06,
+      "loss": 0.9091,
+      "step": 3518
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0830355968622183e-06,
+      "loss": 0.83,
+      "step": 3519
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.082561975553888e-06,
+      "loss": 0.8403,
+      "step": 3520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0820882858397985e-06,
+      "loss": 0.594,
+      "step": 3521
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.081614527775572e-06,
+      "loss": 0.7707,
+      "step": 3522
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.081140701416838e-06,
+      "loss": 0.965,
+      "step": 3523
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.080666806819234e-06,
+      "loss": 0.9338,
+      "step": 3524
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0801928440384056e-06,
+      "loss": 0.883,
+      "step": 3525
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.079718813130008e-06,
+      "loss": 1.0065,
+      "step": 3526
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.0792447141497014e-06,
+      "loss": 0.7347,
+      "step": 3527
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.078770547153156e-06,
+      "loss": 0.8012,
+      "step": 3528
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.07829631219605e-06,
+      "loss": 1.2052,
+      "step": 3529
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.077822009334069e-06,
+      "loss": 0.9172,
+      "step": 3530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.077347638622906e-06,
+      "loss": 0.949,
+      "step": 3531
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.076873200118264e-06,
+      "loss": 1.0354,
+      "step": 3532
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0763986938758515e-06,
+      "loss": 1.0538,
+      "step": 3533
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0759241199513872e-06,
+      "loss": 1.1268,
+      "step": 3534
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.075449478400596e-06,
+      "loss": 0.7609,
+      "step": 3535
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.074974769279211e-06,
+      "loss": 0.923,
+      "step": 3536
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.074499992642974e-06,
+      "loss": 1.1472,
+      "step": 3537
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.074025148547635e-06,
+      "loss": 0.7846,
+      "step": 3538
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.07355023704895e-06,
+      "loss": 1.0487,
+      "step": 3539
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.073075258202685e-06,
+      "loss": 1.0827,
+      "step": 3540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0726002120646136e-06,
+      "loss": 0.6781,
+      "step": 3541
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.072125098690516e-06,
+      "loss": 1.0726,
+      "step": 3542
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.071649918136181e-06,
+      "loss": 0.9987,
+      "step": 3543
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.071174670457406e-06,
+      "loss": 0.8616,
+      "step": 3544
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.070699355709995e-06,
+      "loss": 1.2795,
+      "step": 3545
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0702239739497612e-06,
+      "loss": 0.9547,
+      "step": 3546
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.069748525232525e-06,
+      "loss": 0.8258,
+      "step": 3547
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0692730096141138e-06,
+      "loss": 1.163,
+      "step": 3548
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0687974271503645e-06,
+      "loss": 1.1324,
+      "step": 3549
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0683217778971213e-06,
+      "loss": 1.35,
+      "step": 3550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0678460619102348e-06,
+      "loss": 1.0576,
+      "step": 3551
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.067370279245566e-06,
+      "loss": 0.8916,
+      "step": 3552
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.066894429958982e-06,
+      "loss": 0.9805,
+      "step": 3553
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0664185141063575e-06,
+      "loss": 0.9688,
+      "step": 3554
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0659425317435766e-06,
+      "loss": 0.9113,
+      "step": 3555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0654664829265297e-06,
+      "loss": 0.8617,
+      "step": 3556
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.064990367711116e-06,
+      "loss": 1.0067,
+      "step": 3557
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.064514186153241e-06,
+      "loss": 0.7755,
+      "step": 3558
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0640379383088196e-06,
+      "loss": 1.1914,
+      "step": 3559
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.063561624233774e-06,
+      "loss": 1.1308,
+      "step": 3560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0630852439840343e-06,
+      "loss": 0.9927,
+      "step": 3561
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0626087976155376e-06,
+      "loss": 0.9418,
+      "step": 3562
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.06213228518423e-06,
+      "loss": 0.8246,
+      "step": 3563
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0616557067460642e-06,
+      "loss": 1.2416,
+      "step": 3564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.061179062357001e-06,
+      "loss": 0.7887,
+      "step": 3565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0607023520730092e-06,
+      "loss": 1.169,
+      "step": 3566
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.060225575950066e-06,
+      "loss": 0.8004,
+      "step": 3567
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0597487340441545e-06,
+      "loss": 0.9793,
+      "step": 3568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.059271826411267e-06,
+      "loss": 0.8564,
+      "step": 3569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.058794853107403e-06,
+      "loss": 1.0621,
+      "step": 3570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0583178141885693e-06,
+      "loss": 1.0057,
+      "step": 3571
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0578407097107824e-06,
+      "loss": 0.7074,
+      "step": 3572
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0573635397300636e-06,
+      "loss": 1.0954,
+      "step": 3573
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0568863043024433e-06,
+      "loss": 0.8141,
+      "step": 3574
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.0564090034839604e-06,
+      "loss": 0.9034,
+      "step": 3575
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.05593163733066e-06,
+      "loss": 0.9328,
+      "step": 3576
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.055454205898596e-06,
+      "loss": 1.1355,
+      "step": 3577
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.054976709243829e-06,
+      "loss": 0.5724,
+      "step": 3578
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.054499147422428e-06,
+      "loss": 0.8618,
+      "step": 3579
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0540215204904696e-06,
+      "loss": 1.1201,
+      "step": 3580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0535438285040376e-06,
+      "loss": 0.6755,
+      "step": 3581
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0530660715192237e-06,
+      "loss": 0.54,
+      "step": 3582
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.052588249592127e-06,
+      "loss": 0.7017,
+      "step": 3583
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.052110362778855e-06,
+      "loss": 1.1076,
+      "step": 3584
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0516324111355222e-06,
+      "loss": 0.7498,
+      "step": 3585
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0511543947182504e-06,
+      "loss": 0.8829,
+      "step": 3586
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0506763135831695e-06,
+      "loss": 0.5821,
+      "step": 3587
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0501981677864164e-06,
+      "loss": 0.7813,
+      "step": 3588
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0497199573841366e-06,
+      "loss": 0.8382,
+      "step": 3589
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.049241682432483e-06,
+      "loss": 1.0323,
+      "step": 3590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0487633429876154e-06,
+      "loss": 0.9613,
+      "step": 3591
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0482849391057013e-06,
+      "loss": 1.0322,
+      "step": 3592
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.047806470842916e-06,
+      "loss": 1.0238,
+      "step": 3593
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.047327938255443e-06,
+      "loss": 1.0794,
+      "step": 3594
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.046849341399472e-06,
+      "loss": 0.9089,
+      "step": 3595
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0463706803312004e-06,
+      "loss": 0.8161,
+      "step": 3596
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.045891955106835e-06,
+      "loss": 0.8495,
+      "step": 3597
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0454131657825883e-06,
+      "loss": 1.0967,
+      "step": 3598
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.0449343124146807e-06,
+      "loss": 0.946,
+      "step": 3599
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.04445539505934e-06,
+      "loss": 1.2055,
+      "step": 3600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.043976413772802e-06,
+      "loss": 0.8372,
+      "step": 3601
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0434973686113102e-06,
+      "loss": 0.8591,
+      "step": 3602
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0430182596311145e-06,
+      "loss": 1.0872,
+      "step": 3603
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0425390868884737e-06,
+      "loss": 1.1383,
+      "step": 3604
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0420598504396526e-06,
+      "loss": 1.0215,
+      "step": 3605
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.041580550340925e-06,
+      "loss": 0.969,
+      "step": 3606
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0411011866485707e-06,
+      "loss": 1.293,
+      "step": 3607
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.040621759418878e-06,
+      "loss": 0.9651,
+      "step": 3608
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.040142268708142e-06,
+      "loss": 1.0998,
+      "step": 3609
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0396627145726665e-06,
+      "loss": 0.6426,
+      "step": 3610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0391830970687607e-06,
+      "loss": 0.7922,
+      "step": 3611
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0387034162527434e-06,
+      "loss": 1.1775,
+      "step": 3612
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.038223672180939e-06,
+      "loss": 0.8794,
+      "step": 3613
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.037743864909681e-06,
+      "loss": 1.1183,
+      "step": 3614
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0372639944953084e-06,
+      "loss": 0.5915,
+      "step": 3615
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0367840609941697e-06,
+      "loss": 1.0006,
+      "step": 3616
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0363040644626196e-06,
+      "loss": 0.6532,
+      "step": 3617
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0358240049570196e-06,
+      "loss": 1.0311,
+      "step": 3618
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.03534388253374e-06,
+      "loss": 0.7406,
+      "step": 3619
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.034863697249158e-06,
+      "loss": 0.9305,
+      "step": 3620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.034383449159658e-06,
+      "loss": 1.0239,
+      "step": 3621
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.033903138321631e-06,
+      "loss": 0.6138,
+      "step": 3622
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.0334227647914776e-06,
+      "loss": 1.0625,
+      "step": 3623
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.032942328625603e-06,
+      "loss": 0.9581,
+      "step": 3624
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.032461829880422e-06,
+      "loss": 1.1433,
+      "step": 3625
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0319812686123557e-06,
+      "loss": 1.0484,
+      "step": 3626
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0315006448778325e-06,
+      "loss": 0.9077,
+      "step": 3627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.031019958733289e-06,
+      "loss": 0.705,
+      "step": 3628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0305392102351673e-06,
+      "loss": 0.8236,
+      "step": 3629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0300583994399187e-06,
+      "loss": 1.1857,
+      "step": 3630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.029577526404001e-06,
+      "loss": 0.7853,
+      "step": 3631
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.029096591183879e-06,
+      "loss": 1.3357,
+      "step": 3632
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0286155938360262e-06,
+      "loss": 1.173,
+      "step": 3633
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.028134534416922e-06,
+      "loss": 0.8147,
+      "step": 3634
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.027653412983053e-06,
+      "loss": 0.7953,
+      "step": 3635
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.027172229590914e-06,
+      "loss": 1.0691,
+      "step": 3636
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0266909842970065e-06,
+      "loss": 1.1124,
+      "step": 3637
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0262096771578395e-06,
+      "loss": 0.8838,
+      "step": 3638
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0257283082299293e-06,
+      "loss": 1.0838,
+      "step": 3639
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0252468775697993e-06,
+      "loss": 1.1096,
+      "step": 3640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0247653852339795e-06,
+      "loss": 0.948,
+      "step": 3641
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.024283831279009e-06,
+      "loss": 0.9074,
+      "step": 3642
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0238022157614323e-06,
+      "loss": 0.9699,
+      "step": 3643
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.023320538737802e-06,
+      "loss": 0.7436,
+      "step": 3644
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.022838800264677e-06,
+      "loss": 0.8278,
+      "step": 3645
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0223570003986253e-06,
+      "loss": 1.1281,
+      "step": 3646
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.02187513919622e-06,
+      "loss": 1.181,
+      "step": 3647
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0213932167140435e-06,
+      "loss": 1.2477,
+      "step": 3648
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0209112330086833e-06,
+      "loss": 0.9957,
+      "step": 3649
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.020429188136734e-06,
+      "loss": 0.6052,
+      "step": 3650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.019947082154801e-06,
+      "loss": 0.8805,
+      "step": 3651
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0194649151194924e-06,
+      "loss": 0.6393,
+      "step": 3652
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.018982687087426e-06,
+      "loss": 0.3233,
+      "step": 3653
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0185003981152255e-06,
+      "loss": 1.0226,
+      "step": 3654
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0180180482595235e-06,
+      "loss": 0.8804,
+      "step": 3655
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0175356375769575e-06,
+      "loss": 1.2015,
+      "step": 3656
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.017053166124174e-06,
+      "loss": 0.5624,
+      "step": 3657
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0165706339578257e-06,
+      "loss": 0.6916,
+      "step": 3658
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0160880411345717e-06,
+      "loss": 0.9994,
+      "step": 3659
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0156053877110807e-06,
+      "loss": 1.1047,
+      "step": 3660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.015122673744026e-06,
+      "loss": 1.1546,
+      "step": 3661
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.01463989929009e-06,
+      "loss": 6.212,
+      "step": 3662
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.01415706440596e-06,
+      "loss": 1.0233,
+      "step": 3663
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0136741691483317e-06,
+      "loss": 0.7629,
+      "step": 3664
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0131912135739083e-06,
+      "loss": 0.8331,
+      "step": 3665
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0127081977393994e-06,
+      "loss": 1.0535,
+      "step": 3666
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.012225121701522e-06,
+      "loss": 0.9929,
+      "step": 3667
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0117419855169996e-06,
+      "loss": 1.237,
+      "step": 3668
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.011258789242563e-06,
+      "loss": 0.7752,
+      "step": 3669
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0107755329349508e-06,
+      "loss": 1.1141,
+      "step": 3670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.0102922166509073e-06,
+      "loss": 0.8553,
+      "step": 3671
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0098088404471856e-06,
+      "loss": 0.9774,
+      "step": 3672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0093254043805446e-06,
+      "loss": 0.6756,
+      "step": 3673
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0088419085077494e-06,
+      "loss": 0.9508,
+      "step": 3674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.008358352885574e-06,
+      "loss": 1.0905,
+      "step": 3675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.007874737570799e-06,
+      "loss": 0.9693,
+      "step": 3676
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.007391062620211e-06,
+      "loss": 1.2101,
+      "step": 3677
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0069073280906047e-06,
+      "loss": 0.8022,
+      "step": 3678
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.006423534038781e-06,
+      "loss": 1.0651,
+      "step": 3679
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.005939680521548e-06,
+      "loss": 0.6834,
+      "step": 3680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.005455767595721e-06,
+      "loss": 1.1031,
+      "step": 3681
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0049717953181224e-06,
+      "loss": 0.9753,
+      "step": 3682
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0044877637455806e-06,
+      "loss": 1.0437,
+      "step": 3683
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0040036729349325e-06,
+      "loss": 1.2557,
+      "step": 3684
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.003519522943021e-06,
+      "loss": 0.9242,
+      "step": 3685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.003035313826696e-06,
+      "loss": 0.7665,
+      "step": 3686
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0025510456428144e-06,
+      "loss": 1.1757,
+      "step": 3687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0020667184482397e-06,
+      "loss": 0.7617,
+      "step": 3688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.001582332299843e-06,
+      "loss": 0.9334,
+      "step": 3689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0010978872545028e-06,
+      "loss": 0.911,
+      "step": 3690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0006133833691026e-06,
+      "loss": 1.1607,
+      "step": 3691
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.0001288207005344e-06,
+      "loss": 0.8018,
+      "step": 3692
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9996441993056965e-06,
+      "loss": 0.6397,
+      "step": 3693
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9991595192414947e-06,
+      "loss": 0.9178,
+      "step": 3694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.998674780564841e-06,
+      "loss": 1.0392,
+      "step": 3695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9981899833326544e-06,
+      "loss": 0.7797,
+      "step": 3696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.997705127601861e-06,
+      "loss": 0.8085,
+      "step": 3697
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9972202134293937e-06,
+      "loss": 0.7719,
+      "step": 3698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.996735240872192e-06,
+      "loss": 0.8109,
+      "step": 3699
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9962502099872025e-06,
+      "loss": 1.117,
+      "step": 3700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.995765120831379e-06,
+      "loss": 0.5963,
+      "step": 3701
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.995279973461682e-06,
+      "loss": 0.8511,
+      "step": 3702
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9947947679350775e-06,
+      "loss": 0.8789,
+      "step": 3703
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9943095043085405e-06,
+      "loss": 0.8314,
+      "step": 3704
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9938241826390513e-06,
+      "loss": 1.0667,
+      "step": 3705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.993338802983598e-06,
+      "loss": 0.7124,
+      "step": 3706
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.992853365399174e-06,
+      "loss": 1.0088,
+      "step": 3707
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.992367869942781e-06,
+      "loss": 0.9897,
+      "step": 3708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.991882316671428e-06,
+      "loss": 0.8505,
+      "step": 3709
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9913967056421275e-06,
+      "loss": 1.0574,
+      "step": 3710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9909110369119035e-06,
+      "loss": 1.1487,
+      "step": 3711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9904253105377826e-06,
+      "loss": 0.7737,
+      "step": 3712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.989939526576801e-06,
+      "loss": 1.0541,
+      "step": 3713
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9894536850859995e-06,
+      "loss": 0.8254,
+      "step": 3714
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9889677861224274e-06,
+      "loss": 0.9963,
+      "step": 3715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9884818297431403e-06,
+      "loss": 1.1164,
+      "step": 3716
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9879958160051994e-06,
+      "loss": 1.0414,
+      "step": 3717
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987509744965675e-06,
+      "loss": 1.1998,
+      "step": 3718
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987023616681641e-06,
+      "loss": 1.0703,
+      "step": 3719
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.986537431210181e-06,
+      "loss": 1.0229,
+      "step": 3720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.986051188608383e-06,
+      "loss": 0.8075,
+      "step": 3721
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9855648889333437e-06,
+      "loss": 0.9134,
+      "step": 3722
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9850785322421645e-06,
+      "loss": 0.8964,
+      "step": 3723
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9845921185919553e-06,
+      "loss": 0.9517,
+      "step": 3724
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9841056480398317e-06,
+      "loss": 0.7829,
+      "step": 3725
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.983619120642916e-06,
+      "loss": 1.0471,
+      "step": 3726
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9831325364583376e-06,
+      "loss": 1.0482,
+      "step": 3727
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9826458955432323e-06,
+      "loss": 1.0009,
+      "step": 3728
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9821591979547425e-06,
+      "loss": 1.0485,
+      "step": 3729
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9816724437500173e-06,
+      "loss": 1.0226,
+      "step": 3730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9811856329862125e-06,
+      "loss": 0.587,
+      "step": 3731
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9806987657204905e-06,
+      "loss": 1.0098,
+      "step": 3732
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.980211842010021e-06,
+      "loss": 1.0456,
+      "step": 3733
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9797248619119785e-06,
+      "loss": 1.191,
+      "step": 3734
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.979237825483546e-06,
+      "loss": 0.6575,
+      "step": 3735
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9787507327819122e-06,
+      "loss": 1.0519,
+      "step": 3736
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9782635838642734e-06,
+      "loss": 0.9837,
+      "step": 3737
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.977776378787831e-06,
+      "loss": 0.8503,
+      "step": 3738
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9772891176097933e-06,
+      "loss": 1.0276,
+      "step": 3739
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.976801800387377e-06,
+      "loss": 1.2172,
+      "step": 3740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9763144271778028e-06,
+      "loss": 1.3177,
+      "step": 3741
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9758269980383e-06,
+      "loss": 0.9368,
+      "step": 3742
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9753395130261026e-06,
+      "loss": 0.9919,
+      "step": 3743
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9748519721984537e-06,
+      "loss": 0.9293,
+      "step": 3744
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9743643756126002e-06,
+      "loss": 1.2667,
+      "step": 3745
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.973876723325797e-06,
+      "loss": 1.046,
+      "step": 3746
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.973389015395306e-06,
+      "loss": 0.7783,
+      "step": 3747
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.972901251878394e-06,
+      "loss": 0.8489,
+      "step": 3748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.972413432832336e-06,
+      "loss": 0.8679,
+      "step": 3749
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.971925558314413e-06,
+      "loss": 1.2325,
+      "step": 3750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.971437628381912e-06,
+      "loss": 1.0832,
+      "step": 3751
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.970949643092126e-06,
+      "loss": 0.6132,
+      "step": 3752
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.970461602502357e-06,
+      "loss": 1.0717,
+      "step": 3753
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.969973506669911e-06,
+      "loss": 0.7751,
+      "step": 3754
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9694853556521013e-06,
+      "loss": 1.2649,
+      "step": 3755
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9689971495062477e-06,
+      "loss": 0.8671,
+      "step": 3756
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9685088882896767e-06,
+      "loss": 1.3585,
+      "step": 3757
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.968020572059721e-06,
+      "loss": 1.021,
+      "step": 3758
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9675322008737193e-06,
+      "loss": 0.9693,
+      "step": 3759
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9670437747890177e-06,
+      "loss": 1.1635,
+      "step": 3760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.966555293862969e-06,
+      "loss": 1.2262,
+      "step": 3761
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9660667581529305e-06,
+      "loss": 0.8256,
+      "step": 3762
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.965578167716268e-06,
+      "loss": 0.8971,
+      "step": 3763
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9650895226103524e-06,
+      "loss": 0.8774,
+      "step": 3764
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9646008228925614e-06,
+      "loss": 0.8061,
+      "step": 3765
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9641120686202802e-06,
+      "loss": 0.6023,
+      "step": 3766
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9636232598508987e-06,
+      "loss": 0.8475,
+      "step": 3767
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.963134396641814e-06,
+      "loss": 1.3484,
+      "step": 3768
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9626454790504298e-06,
+      "loss": 1.1658,
+      "step": 3769
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9621565071341555e-06,
+      "loss": 0.872,
+      "step": 3770
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9616674809504074e-06,
+      "loss": 0.8731,
+      "step": 3771
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9611784005566086e-06,
+      "loss": 1.0056,
+      "step": 3772
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9606892660101876e-06,
+      "loss": 1.0212,
+      "step": 3773
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9602000773685797e-06,
+      "loss": 0.9768,
+      "step": 3774
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9597108346892267e-06,
+      "loss": 0.9167,
+      "step": 3775
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.959221538029576e-06,
+      "loss": 0.8795,
+      "step": 3776
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9587321874470826e-06,
+      "loss": 0.924,
+      "step": 3777
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.958242782999207e-06,
+      "loss": 0.9034,
+      "step": 3778
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9577533247434165e-06,
+      "loss": 1.0946,
+      "step": 3779
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.957263812737184e-06,
+      "loss": 0.7527,
+      "step": 3780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.956774247037989e-06,
+      "loss": 0.9879,
+      "step": 3781
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9562846277033173e-06,
+      "loss": 1.1524,
+      "step": 3782
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.955794954790662e-06,
+      "loss": 0.9572,
+      "step": 3783
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9553052283575212e-06,
+      "loss": 0.9542,
+      "step": 3784
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.954815448461399e-06,
+      "loss": 0.8424,
+      "step": 3785
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.954325615159807e-06,
+      "loss": 0.7492,
+      "step": 3786
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9538357285102635e-06,
+      "loss": 1.0836,
+      "step": 3787
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9533457885702904e-06,
+      "loss": 0.9569,
+      "step": 3788
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9528557953974184e-06,
+      "loss": 0.9461,
+      "step": 3789
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9523657490491837e-06,
+      "loss": 1.1165,
+      "step": 3790
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.951875649583129e-06,
+      "loss": 0.9871,
+      "step": 3791
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9513854970568015e-06,
+      "loss": 0.9832,
+      "step": 3792
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9508952915277572e-06,
+      "loss": 1.0177,
+      "step": 3793
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.950405033053557e-06,
+      "loss": 0.9314,
+      "step": 3794
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9499147216917685e-06,
+      "loss": 0.9784,
+      "step": 3795
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9494243574999647e-06,
+      "loss": 1.1013,
+      "step": 3796
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.948933940535725e-06,
+      "loss": 1.0641,
+      "step": 3797
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9484434708566355e-06,
+      "loss": 0.968,
+      "step": 3798
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.947952948520289e-06,
+      "loss": 0.659,
+      "step": 3799
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.947462373584283e-06,
+      "loss": 0.7804,
+      "step": 3800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9469717461062217e-06,
+      "loss": 0.9855,
+      "step": 3801
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9464810661437166e-06,
+      "loss": 0.5953,
+      "step": 3802
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459903337543837e-06,
+      "loss": 1.0511,
+      "step": 3803
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945499548995846e-06,
+      "loss": 0.8757,
+      "step": 3804
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9450087119257324e-06,
+      "loss": 1.2438,
+      "step": 3805
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.944517822601679e-06,
+      "loss": 0.6625,
+      "step": 3806
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.944026881081327e-06,
+      "loss": 1.0422,
+      "step": 3807
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.943535887422323e-06,
+      "loss": 0.9332,
+      "step": 3808
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9430448416823212e-06,
+      "loss": 1.0661,
+      "step": 3809
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.942553743918981e-06,
+      "loss": 0.6279,
+      "step": 3810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9420625941899684e-06,
+      "loss": 0.7642,
+      "step": 3811
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.941571392552956e-06,
+      "loss": 1.1113,
+      "step": 3812
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9410801390656205e-06,
+      "loss": 1.2471,
+      "step": 3813
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9405888337856465e-06,
+      "loss": 0.8589,
+      "step": 3814
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.940097476770725e-06,
+      "loss": 0.7441,
+      "step": 3815
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.939606068078551e-06,
+      "loss": 1.0914,
+      "step": 3816
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9391146077668278e-06,
+      "loss": 0.782,
+      "step": 3817
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.938623095893263e-06,
+      "loss": 0.8867,
+      "step": 3818
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9381315325155718e-06,
+      "loss": 1.4521,
+      "step": 3819
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9376399176914747e-06,
+      "loss": 0.8598,
+      "step": 3820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.937148251478698e-06,
+      "loss": 0.9866,
+      "step": 3821
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9366565339349734e-06,
+      "loss": 1.1058,
+      "step": 3822
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9361647651180406e-06,
+      "loss": 0.9413,
+      "step": 3823
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.935672945085644e-06,
+      "loss": 0.7221,
+      "step": 3824
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9351810738955344e-06,
+      "loss": 0.9373,
+      "step": 3825
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9346891516054676e-06,
+      "loss": 1.0704,
+      "step": 3826
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9341971782732073e-06,
+      "loss": 0.9318,
+      "step": 3827
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9337051539565214e-06,
+      "loss": 0.6031,
+      "step": 3828
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.933213078713185e-06,
+      "loss": 1.1402,
+      "step": 3829
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9327209526009787e-06,
+      "loss": 0.7805,
+      "step": 3830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.932228775677688e-06,
+      "loss": 1.1216,
+      "step": 3831
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.931736548001107e-06,
+      "loss": 0.913,
+      "step": 3832
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.931244269629033e-06,
+      "loss": 1.0468,
+      "step": 3833
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9307519406192712e-06,
+      "loss": 0.7141,
+      "step": 3834
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.930259561029632e-06,
+      "loss": 1.009,
+      "step": 3835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9297671309179314e-06,
+      "loss": 0.8915,
+      "step": 3836
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9292746503419913e-06,
+      "loss": 0.769,
+      "step": 3837
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.928782119359641e-06,
+      "loss": 1.1328,
+      "step": 3838
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9282895380287133e-06,
+      "loss": 0.8981,
+      "step": 3839
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9277969064070493e-06,
+      "loss": 1.0296,
+      "step": 3840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.927304224552494e-06,
+      "loss": 0.5588,
+      "step": 3841
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9268114925229002e-06,
+      "loss": 0.8919,
+      "step": 3842
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.926318710376125e-06,
+      "loss": 0.788,
+      "step": 3843
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9258258781700325e-06,
+      "loss": 0.9891,
+      "step": 3844
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9253329959624915e-06,
+      "loss": 1.0116,
+      "step": 3845
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.924840063811378e-06,
+      "loss": 1.1816,
+      "step": 3846
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.924347081774572e-06,
+      "loss": 0.9898,
+      "step": 3847
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.923854049909963e-06,
+      "loss": 0.9966,
+      "step": 3848
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.923360968275441e-06,
+      "loss": 0.9681,
+      "step": 3849
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.922867836928907e-06,
+      "loss": 1.0951,
+      "step": 3850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9223746559282648e-06,
+      "loss": 0.7897,
+      "step": 3851
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.921881425331425e-06,
+      "loss": 0.65,
+      "step": 3852
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.921388145196303e-06,
+      "loss": 0.8379,
+      "step": 3853
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.920894815580822e-06,
+      "loss": 0.7449,
+      "step": 3854
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9204014365429096e-06,
+      "loss": 1.066,
+      "step": 3855
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9199080081404993e-06,
+      "loss": 1.1991,
+      "step": 3856
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9194145304315304e-06,
+      "loss": 0.9625,
+      "step": 3857
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.918921003473949e-06,
+      "loss": 0.9177,
+      "step": 3858
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9184274273257055e-06,
+      "loss": 0.5295,
+      "step": 3859
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9179338020447567e-06,
+      "loss": 1.276,
+      "step": 3860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.917440127689065e-06,
+      "loss": 0.6612,
+      "step": 3861
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9169464043166e-06,
+      "loss": 0.9496,
+      "step": 3862
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9164526319853344e-06,
+      "loss": 0.9059,
+      "step": 3863
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.915958810753249e-06,
+      "loss": 0.8796,
+      "step": 3864
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9154649406783285e-06,
+      "loss": 0.9022,
+      "step": 3865
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.914971021818565e-06,
+      "loss": 1.0116,
+      "step": 3866
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.914477054231956e-06,
+      "loss": 0.9321,
+      "step": 3867
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9139830379765028e-06,
+      "loss": 1.1785,
+      "step": 3868
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9134889731102156e-06,
+      "loss": 1.0388,
+      "step": 3869
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9129948596911076e-06,
+      "loss": 1.0015,
+      "step": 3870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.912500697777199e-06,
+      "loss": 0.8871,
+      "step": 3871
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9120064874265154e-06,
+      "loss": 0.9375,
+      "step": 3872
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.911512228697088e-06,
+      "loss": 1.004,
+      "step": 3873
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9110179216469537e-06,
+      "loss": 0.7029,
+      "step": 3874
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9105235663341557e-06,
+      "loss": 1.3208,
+      "step": 3875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.910029162816742e-06,
+      "loss": 0.495,
+      "step": 3876
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9095347111527665e-06,
+      "loss": 1.1176,
+      "step": 3877
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.909040211400289e-06,
+      "loss": 1.2138,
+      "step": 3878
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9085456636173745e-06,
+      "loss": 1.054,
+      "step": 3879
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.908051067862094e-06,
+      "loss": 1.2954,
+      "step": 3880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9075564241925243e-06,
+      "loss": 1.0459,
+      "step": 3881
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.907061732666748e-06,
+      "loss": 1.211,
+      "step": 3882
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9065669933428517e-06,
+      "loss": 1.4657,
+      "step": 3883
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9060722062789299e-06,
+      "loss": 0.7121,
+      "step": 3884
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9055773715330814e-06,
+      "loss": 0.948,
+      "step": 3885
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9050824891634103e-06,
+      "loss": 1.2614,
+      "step": 3886
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9045875592280274e-06,
+      "loss": 0.8514,
+      "step": 3887
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9040925817850488e-06,
+      "loss": 0.9778,
+      "step": 3888
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9035975568925952e-06,
+      "loss": 1.3539,
+      "step": 3889
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.903102484608793e-06,
+      "loss": 1.2406,
+      "step": 3890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.902607364991776e-06,
+      "loss": 1.0692,
+      "step": 3891
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.902112198099682e-06,
+      "loss": 0.7777,
+      "step": 3892
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9016169839906544e-06,
+      "loss": 0.7751,
+      "step": 3893
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9011217227228417e-06,
+      "loss": 1.0149,
+      "step": 3894
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9006264143543996e-06,
+      "loss": 1.162,
+      "step": 3895
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9001310589434884e-06,
+      "loss": 0.7233,
+      "step": 3896
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8996356565482734e-06,
+      "loss": 0.7783,
+      "step": 3897
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8991402072269252e-06,
+      "loss": 0.8931,
+      "step": 3898
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.898644711037622e-06,
+      "loss": 0.7538,
+      "step": 3899
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8981491680385454e-06,
+      "loss": 0.8271,
+      "step": 3900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8976535782878828e-06,
+      "loss": 1.0168,
+      "step": 3901
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8971579418438282e-06,
+      "loss": 0.9686,
+      "step": 3902
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.89666225876458e-06,
+      "loss": 0.9416,
+      "step": 3903
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8961665291083427e-06,
+      "loss": 1.2917,
+      "step": 3904
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.895670752933326e-06,
+      "loss": 0.7497,
+      "step": 3905
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8951749302977448e-06,
+      "loss": 0.9727,
+      "step": 3906
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8946790612598195e-06,
+      "loss": 1.1413,
+      "step": 3907
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8941831458777768e-06,
+      "loss": 1.056,
+      "step": 3908
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8936871842098476e-06,
+      "loss": 0.9438,
+      "step": 3909
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8931911763142692e-06,
+      "loss": 0.9751,
+      "step": 3910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8926951222492843e-06,
+      "loss": 1.0148,
+      "step": 3911
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.89219902207314e-06,
+      "loss": 0.7107,
+      "step": 3912
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8917028758440901e-06,
+      "loss": 0.8653,
+      "step": 3913
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.891206683620393e-06,
+      "loss": 0.8249,
+      "step": 3914
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8907104454603127e-06,
+      "loss": 0.8138,
+      "step": 3915
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8902141614221184e-06,
+      "loss": 0.8261,
+      "step": 3916
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8897178315640855e-06,
+      "loss": 1.1043,
+      "step": 3917
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8892214559444937e-06,
+      "loss": 0.8076,
+      "step": 3918
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8887250346216286e-06,
+      "loss": 1.2215,
+      "step": 3919
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.888228567653781e-06,
+      "loss": 0.7522,
+      "step": 3920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.887732055099248e-06,
+      "loss": 0.8633,
+      "step": 3921
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8872354970163304e-06,
+      "loss": 0.9708,
+      "step": 3922
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8867388934633353e-06,
+      "loss": 1.1559,
+      "step": 3923
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.886242244498575e-06,
+      "loss": 0.7983,
+      "step": 3924
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8857455501803677e-06,
+      "loss": 0.7618,
+      "step": 3925
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8852488105670358e-06,
+      "loss": 1.0308,
+      "step": 3926
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8847520257169076e-06,
+      "loss": 0.4644,
+      "step": 3927
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8842551956883173e-06,
+      "loss": 0.9558,
+      "step": 3928
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8837583205396035e-06,
+      "loss": 1.1041,
+      "step": 3929
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8832614003291102e-06,
+      "loss": 1.1363,
+      "step": 3930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8827644351151868e-06,
+      "loss": 0.8301,
+      "step": 3931
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8822674249561881e-06,
+      "loss": 1.1744,
+      "step": 3932
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8817703699104754e-06,
+      "loss": 0.9769,
+      "step": 3933
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8812732700364122e-06,
+      "loss": 0.9638,
+      "step": 3934
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8807761253923706e-06,
+      "loss": 0.9527,
+      "step": 3935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.880278936036725e-06,
+      "loss": 1.2264,
+      "step": 3936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.879781702027858e-06,
+      "loss": 1.0725,
+      "step": 3937
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.879284423424155e-06,
+      "loss": 0.8487,
+      "step": 3938
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.878787100284008e-06,
+      "loss": 0.7324,
+      "step": 3939
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8782897326658137e-06,
+      "loss": 1.0692,
+      "step": 3940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.877792320627974e-06,
+      "loss": 0.6354,
+      "step": 3941
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.877294864228896e-06,
+      "loss": 0.9755,
+      "step": 3942
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8767973635269932e-06,
+      "loss": 0.6466,
+      "step": 3943
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8762998185806823e-06,
+      "loss": 0.9536,
+      "step": 3944
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8758022294483862e-06,
+      "loss": 1.2702,
+      "step": 3945
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8753045961885328e-06,
+      "loss": 1.0724,
+      "step": 3946
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8748069188595563e-06,
+      "loss": 1.1068,
+      "step": 3947
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8743091975198942e-06,
+      "loss": 0.91,
+      "step": 3948
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8738114322279905e-06,
+      "loss": 1.1529,
+      "step": 3949
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8733136230422935e-06,
+      "loss": 1.033,
+      "step": 3950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.872815770021258e-06,
+      "loss": 0.9835,
+      "step": 3951
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.872317873223342e-06,
+      "loss": 1.1287,
+      "step": 3952
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.87181993270701e-06,
+      "loss": 0.8721,
+      "step": 3953
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8713219485307318e-06,
+      "loss": 1.0096,
+      "step": 3954
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8708239207529813e-06,
+      "loss": 0.7555,
+      "step": 3955
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8703258494322383e-06,
+      "loss": 0.8461,
+      "step": 3956
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.869827734626987e-06,
+      "loss": 0.8503,
+      "step": 3957
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8693295763957183e-06,
+      "loss": 0.8479,
+      "step": 3958
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.868831374796926e-06,
+      "loss": 0.8383,
+      "step": 3959
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8683331298891105e-06,
+      "loss": 0.8392,
+      "step": 3960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8678348417307764e-06,
+      "loss": 1.1184,
+      "step": 3961
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8673365103804349e-06,
+      "loss": 1.0383,
+      "step": 3962
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8668381358966003e-06,
+      "loss": 0.9835,
+      "step": 3963
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.866339718337793e-06,
+      "loss": 1.0964,
+      "step": 3964
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.865841257762538e-06,
+      "loss": 1.057,
+      "step": 3965
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.865342754229367e-06,
+      "loss": 1.0893,
+      "step": 3966
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8648442077968141e-06,
+      "loss": 1.1496,
+      "step": 3967
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8643456185234204e-06,
+      "loss": 0.7905,
+      "step": 3968
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.863846986467731e-06,
+      "loss": 1.6522,
+      "step": 3969
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8633483116882965e-06,
+      "loss": 0.9443,
+      "step": 3970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8628495942436733e-06,
+      "loss": 0.8659,
+      "step": 3971
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.86235083419242e-06,
+      "loss": 0.7819,
+      "step": 3972
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.861852031593104e-06,
+      "loss": 1.2499,
+      "step": 3973
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8613531865042954e-06,
+      "loss": 0.6485,
+      "step": 3974
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.860854298984569e-06,
+      "loss": 1.0466,
+      "step": 3975
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.860355369092506e-06,
+      "loss": 0.5067,
+      "step": 3976
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8598563968866916e-06,
+      "loss": 0.965,
+      "step": 3977
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8593573824257166e-06,
+      "loss": 0.8946,
+      "step": 3978
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8588583257681757e-06,
+      "loss": 0.9367,
+      "step": 3979
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8583592269726701e-06,
+      "loss": 0.7022,
+      "step": 3980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8578600860978053e-06,
+      "loss": 0.8725,
+      "step": 3981
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8573609032021905e-06,
+      "loss": 1.0167,
+      "step": 3982
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8568616783444411e-06,
+      "loss": 1.0937,
+      "step": 3983
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.856362411583178e-06,
+      "loss": 1.0278,
+      "step": 3984
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.855863102977026e-06,
+      "loss": 0.9047,
+      "step": 3985
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8553637525846146e-06,
+      "loss": 1.1005,
+      "step": 3986
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8548643604645788e-06,
+      "loss": 1.5267,
+      "step": 3987
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8543649266755586e-06,
+      "loss": 0.8525,
+      "step": 3988
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8538654512761985e-06,
+      "loss": 0.9413,
+      "step": 3989
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8533659343251484e-06,
+      "loss": 1.1188,
+      "step": 3990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.852866375881062e-06,
+      "loss": 1.0688,
+      "step": 3991
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8523667760025993e-06,
+      "loss": 0.7245,
+      "step": 3992
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8518671347484241e-06,
+      "loss": 0.7457,
+      "step": 3993
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.851367452177205e-06,
+      "loss": 0.9309,
+      "step": 3994
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.850867728347617e-06,
+      "loss": 0.958,
+      "step": 3995
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8503679633183383e-06,
+      "loss": 0.399,
+      "step": 3996
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.849868157148052e-06,
+      "loss": 1.1309,
+      "step": 3997
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.849368309895447e-06,
+      "loss": 0.9287,
+      "step": 3998
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8488684216192163e-06,
+      "loss": 1.1171,
+      "step": 3999
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8483684923780575e-06,
+      "loss": 0.9323,
+      "step": 4000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8478685222306743e-06,
+      "loss": 0.6778,
+      "step": 4001
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8473685112357736e-06,
+      "loss": 1.1661,
+      "step": 4002
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8468684594520684e-06,
+      "loss": 0.9382,
+      "step": 4003
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8463683669382758e-06,
+      "loss": 0.9245,
+      "step": 4004
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8458682337531174e-06,
+      "loss": 0.7419,
+      "step": 4005
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8453680599553202e-06,
+      "loss": 1.0524,
+      "step": 4006
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8448678456036163e-06,
+      "loss": 1.1127,
+      "step": 4007
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.844367590756741e-06,
+      "loss": 0.9596,
+      "step": 4008
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8438672954734358e-06,
+      "loss": 0.823,
+      "step": 4009
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.843366959812447e-06,
+      "loss": 1.0975,
+      "step": 4010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8428665838325247e-06,
+      "loss": 1.0843,
+      "step": 4011
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8423661675924245e-06,
+      "loss": 0.9589,
+      "step": 4012
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8418657111509056e-06,
+      "loss": 1.0023,
+      "step": 4013
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8413652145667335e-06,
+      "loss": 0.7686,
+      "step": 4014
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8408646778986777e-06,
+      "loss": 1.0191,
+      "step": 4015
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8403641012055121e-06,
+      "loss": 0.6547,
+      "step": 4016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.839863484546015e-06,
+      "loss": 1.0418,
+      "step": 4017
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8393628279789713e-06,
+      "loss": 0.8445,
+      "step": 4018
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8388621315631683e-06,
+      "loss": 0.8321,
+      "step": 4019
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8383613953573989e-06,
+      "loss": 0.9842,
+      "step": 4020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8378606194204608e-06,
+      "loss": 1.0719,
+      "step": 4021
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8373598038111568e-06,
+      "loss": 0.8317,
+      "step": 4022
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8368589485882929e-06,
+      "loss": 0.9443,
+      "step": 4023
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.836358053810681e-06,
+      "loss": 0.8076,
+      "step": 4024
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.835857119537138e-06,
+      "loss": 1.0995,
+      "step": 4025
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8353561458264838e-06,
+      "loss": 0.968,
+      "step": 4026
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8348551327375445e-06,
+      "loss": 0.9052,
+      "step": 4027
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8343540803291494e-06,
+      "loss": 1.0964,
+      "step": 4028
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8338529886601343e-06,
+      "loss": 1.1334,
+      "step": 4029
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.833351857789338e-06,
+      "loss": 0.921,
+      "step": 4030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8328506877756046e-06,
+      "loss": 0.8256,
+      "step": 4031
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8323494786777823e-06,
+      "loss": 1.0398,
+      "step": 4032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8318482305547242e-06,
+      "loss": 1.0878,
+      "step": 4033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8313469434652886e-06,
+      "loss": 0.9127,
+      "step": 4034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8308456174683367e-06,
+      "loss": 0.7472,
+      "step": 4035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8303442526227366e-06,
+      "loss": 0.3431,
+      "step": 4036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8298428489873586e-06,
+      "loss": 0.8434,
+      "step": 4037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8293414066210796e-06,
+      "loss": 1.1774,
+      "step": 4038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8288399255827793e-06,
+      "loss": 0.9821,
+      "step": 4039
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8283384059313434e-06,
+      "loss": 1.1427,
+      "step": 4040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8278368477256612e-06,
+      "loss": 1.3832,
+      "step": 4041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.827335251024627e-06,
+      "loss": 1.0508,
+      "step": 4042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.826833615887139e-06,
+      "loss": 1.1298,
+      "step": 4043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8263319423721008e-06,
+      "loss": 0.9539,
+      "step": 4044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8258302305384196e-06,
+      "loss": 1.3553,
+      "step": 4045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.825328480445008e-06,
+      "loss": 0.7547,
+      "step": 4046
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8248266921507822e-06,
+      "loss": 1.3125,
+      "step": 4047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8243248657146639e-06,
+      "loss": 0.8311,
+      "step": 4048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8238230011955785e-06,
+      "loss": 0.9792,
+      "step": 4049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8233210986524551e-06,
+      "loss": 0.6943,
+      "step": 4050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8228191581442302e-06,
+      "loss": 1.2449,
+      "step": 4051
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8223171797298413e-06,
+      "loss": 1.0147,
+      "step": 4052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8218151634682323e-06,
+      "loss": 0.8861,
+      "step": 4053
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8213131094183509e-06,
+      "loss": 0.6279,
+      "step": 4054
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8208110176391497e-06,
+      "loss": 1.0425,
+      "step": 4055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8203088881895855e-06,
+      "loss": 0.9051,
+      "step": 4056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.819806721128619e-06,
+      "loss": 1.0019,
+      "step": 4057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.819304516515216e-06,
+      "loss": 0.9829,
+      "step": 4058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8188022744083472e-06,
+      "loss": 1.0986,
+      "step": 4059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8182999948669863e-06,
+      "loss": 0.9843,
+      "step": 4060
+    },
+    {
+      "epoch": 1.73,
+      "eval_loss": 1.2551723718643188,
+      "eval_runtime": 32.1869,
+      "eval_samples_per_second": 38.68,
+      "eval_steps_per_second": 19.356,
+      "step": 4060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8177976779501125e-06,
+      "loss": 1.0784,
+      "step": 4061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8172953237167085e-06,
+      "loss": 0.847,
+      "step": 4062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8167929322257623e-06,
+      "loss": 0.9712,
+      "step": 4063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8162905035362657e-06,
+      "loss": 1.0022,
+      "step": 4064
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8157880377072153e-06,
+      "loss": 0.9416,
+      "step": 4065
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8152855347976114e-06,
+      "loss": 0.8662,
+      "step": 4066
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8147829948664591e-06,
+      "loss": 0.8409,
+      "step": 4067
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.814280417972768e-06,
+      "loss": 1.0053,
+      "step": 4068
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8137778041755515e-06,
+      "loss": 1.181,
+      "step": 4069
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.813275153533828e-06,
+      "loss": 0.9689,
+      "step": 4070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8127724661066195e-06,
+      "loss": 0.8559,
+      "step": 4071
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.812269741952953e-06,
+      "loss": 1.7472,
+      "step": 4072
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.81176698113186e-06,
+      "loss": 0.7472,
+      "step": 4073
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.811264183702375e-06,
+      "loss": 0.5997,
+      "step": 4074
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8107613497235378e-06,
+      "loss": 1.0756,
+      "step": 4075
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8102584792543925e-06,
+      "loss": 0.695,
+      "step": 4076
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8097555723539876e-06,
+      "loss": 0.9222,
+      "step": 4077
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8092526290813749e-06,
+      "loss": 0.921,
+      "step": 4078
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8087496494956114e-06,
+      "loss": 0.6828,
+      "step": 4079
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.808246633655758e-06,
+      "loss": 0.9029,
+      "step": 4080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8077435816208805e-06,
+      "loss": 1.1254,
+      "step": 4081
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.807240493450048e-06,
+      "loss": 0.702,
+      "step": 4082
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8067373692023345e-06,
+      "loss": 0.7908,
+      "step": 4083
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8062342089368176e-06,
+      "loss": 0.8894,
+      "step": 4084
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8057310127125796e-06,
+      "loss": 0.9233,
+      "step": 4085
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8052277805887075e-06,
+      "loss": 1.4311,
+      "step": 4086
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8047245126242913e-06,
+      "loss": 0.9002,
+      "step": 4087
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8042212088784264e-06,
+      "loss": 0.8057,
+      "step": 4088
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8037178694102114e-06,
+      "loss": 1.2138,
+      "step": 4089
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.80321449427875e-06,
+      "loss": 1.0924,
+      "step": 4090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8027110835431493e-06,
+      "loss": 0.7547,
+      "step": 4091
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8022076372625215e-06,
+      "loss": 0.9919,
+      "step": 4092
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8017041554959818e-06,
+      "loss": 1.0311,
+      "step": 4093
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8012006383026502e-06,
+      "loss": 0.8519,
+      "step": 4094
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8006970857416513e-06,
+      "loss": 0.8682,
+      "step": 4095
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8001934978721131e-06,
+      "loss": 0.9108,
+      "step": 4096
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.799689874753168e-06,
+      "loss": 1.0693,
+      "step": 4097
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7991862164439525e-06,
+      "loss": 0.8723,
+      "step": 4098
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7986825230036078e-06,
+      "loss": 0.9631,
+      "step": 4099
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7981787944912778e-06,
+      "loss": 1.0018,
+      "step": 4100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7976750309661127e-06,
+      "loss": 0.8259,
+      "step": 4101
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7971712324872643e-06,
+      "loss": 0.689,
+      "step": 4102
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7966673991138907e-06,
+      "loss": 1.0346,
+      "step": 4103
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.796163530905153e-06,
+      "loss": 0.8931,
+      "step": 4104
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7956596279202162e-06,
+      "loss": 0.9735,
+      "step": 4105
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7951556902182499e-06,
+      "loss": 1.0288,
+      "step": 4106
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7946517178584277e-06,
+      "loss": 0.8892,
+      "step": 4107
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7941477108999274e-06,
+      "loss": 1.2085,
+      "step": 4108
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7936436694019302e-06,
+      "loss": 1.299,
+      "step": 4109
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7931395934236217e-06,
+      "loss": 1.0556,
+      "step": 4110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7926354830241926e-06,
+      "loss": 0.841,
+      "step": 4111
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7921313382628355e-06,
+      "loss": 1.0767,
+      "step": 4112
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.791627159198749e-06,
+      "loss": 0.7847,
+      "step": 4113
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7911229458911348e-06,
+      "loss": 0.7628,
+      "step": 4114
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7906186983991987e-06,
+      "loss": 0.809,
+      "step": 4115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7901144167821509e-06,
+      "loss": 0.7485,
+      "step": 4116
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7896101010992042e-06,
+      "loss": 0.9664,
+      "step": 4117
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7891057514095779e-06,
+      "loss": 1.1681,
+      "step": 4118
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.788601367772493e-06,
+      "loss": 0.638,
+      "step": 4119
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7880969502471763e-06,
+      "loss": 1.3513,
+      "step": 4120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7875924988928563e-06,
+      "loss": 0.8658,
+      "step": 4121
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7870880137687682e-06,
+      "loss": 1.1213,
+      "step": 4122
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7865834949341488e-06,
+      "loss": 0.644,
+      "step": 4123
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7860789424482405e-06,
+      "loss": 0.4287,
+      "step": 4124
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7855743563702881e-06,
+      "loss": 0.9111,
+      "step": 4125
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7850697367595427e-06,
+      "loss": 0.9724,
+      "step": 4126
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7845650836752569e-06,
+      "loss": 1.0075,
+      "step": 4127
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.784060397176688e-06,
+      "loss": 1.0967,
+      "step": 4128
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7835556773230982e-06,
+      "loss": 1.2478,
+      "step": 4129
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7830509241737523e-06,
+      "loss": 1.3544,
+      "step": 4130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.78254613778792e-06,
+      "loss": 0.7636,
+      "step": 4131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.782041318224874e-06,
+      "loss": 0.7162,
+      "step": 4132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7815364655438915e-06,
+      "loss": 1.0199,
+      "step": 4133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.781031579804254e-06,
+      "loss": 0.6936,
+      "step": 4134
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7805266610652457e-06,
+      "loss": 0.9383,
+      "step": 4135
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7800217093861553e-06,
+      "loss": 1.3372,
+      "step": 4136
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7795167248262755e-06,
+      "loss": 0.7236,
+      "step": 4137
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7790117074449033e-06,
+      "loss": 0.9176,
+      "step": 4138
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7785066573013378e-06,
+      "loss": 0.6828,
+      "step": 4139
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7780015744548846e-06,
+      "loss": 0.9276,
+      "step": 4140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7774964589648508e-06,
+      "loss": 0.9058,
+      "step": 4141
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7769913108905483e-06,
+      "loss": 0.9648,
+      "step": 4142
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7764861302912927e-06,
+      "loss": 0.7603,
+      "step": 4143
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7759809172264036e-06,
+      "loss": 0.9921,
+      "step": 4144
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7754756717552043e-06,
+      "loss": 0.8591,
+      "step": 4145
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.774970393937022e-06,
+      "loss": 0.7732,
+      "step": 4146
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7744650838311869e-06,
+      "loss": 0.9587,
+      "step": 4147
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7739597414970345e-06,
+      "loss": 0.9832,
+      "step": 4148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.773454366993903e-06,
+      "loss": 0.9446,
+      "step": 4149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7729489603811347e-06,
+      "loss": 0.85,
+      "step": 4150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7724435217180747e-06,
+      "loss": 1.0807,
+      "step": 4151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7719380510640744e-06,
+      "loss": 0.963,
+      "step": 4152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.771432548478486e-06,
+      "loss": 0.8644,
+      "step": 4153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.770927014020667e-06,
+      "loss": 0.9484,
+      "step": 4154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7704214477499789e-06,
+      "loss": 1.4628,
+      "step": 4155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.769915849725786e-06,
+      "loss": 0.7443,
+      "step": 4156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7694102200074572e-06,
+      "loss": 1.0412,
+      "step": 4157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7689045586543644e-06,
+      "loss": 0.615,
+      "step": 4158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7683988657258838e-06,
+      "loss": 1.0354,
+      "step": 4159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7678931412813946e-06,
+      "loss": 0.9812,
+      "step": 4160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7673873853802798e-06,
+      "loss": 0.9586,
+      "step": 4161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7668815980819269e-06,
+      "loss": 0.7448,
+      "step": 4162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7663757794457272e-06,
+      "loss": 0.75,
+      "step": 4163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7658699295310742e-06,
+      "loss": 0.7127,
+      "step": 4164
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.765364048397366e-06,
+      "loss": 0.6851,
+      "step": 4165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7648581361040046e-06,
+      "loss": 0.5688,
+      "step": 4166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7643521927103957e-06,
+      "loss": 0.945,
+      "step": 4167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.763846218275948e-06,
+      "loss": 0.8032,
+      "step": 4168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7633402128600735e-06,
+      "loss": 1.0722,
+      "step": 4169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7628341765221896e-06,
+      "loss": 1.159,
+      "step": 4170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.762328109321716e-06,
+      "loss": 0.8739,
+      "step": 4171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7618220113180758e-06,
+      "loss": 0.7725,
+      "step": 4172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7613158825706963e-06,
+      "loss": 0.8106,
+      "step": 4173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7608097231390086e-06,
+      "loss": 0.9554,
+      "step": 4174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.760303533082447e-06,
+      "loss": 1.0577,
+      "step": 4175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7597973124604495e-06,
+      "loss": 0.7409,
+      "step": 4176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7592910613324574e-06,
+      "loss": 0.9291,
+      "step": 4177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7587847797579161e-06,
+      "loss": 0.7995,
+      "step": 4178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7582784677962742e-06,
+      "loss": 1.0438,
+      "step": 4179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.757772125506984e-06,
+      "loss": 5.606,
+      "step": 4180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7572657529495023e-06,
+      "loss": 0.975,
+      "step": 4181
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7567593501832871e-06,
+      "loss": 1.074,
+      "step": 4182
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7562529172678022e-06,
+      "loss": 0.897,
+      "step": 4183
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7557464542625137e-06,
+      "loss": 0.8277,
+      "step": 4184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7552399612268917e-06,
+      "loss": 0.9661,
+      "step": 4185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7547334382204105e-06,
+      "loss": 1.0473,
+      "step": 4186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7542268853025463e-06,
+      "loss": 1.0779,
+      "step": 4187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7537203025327797e-06,
+      "loss": 1.0993,
+      "step": 4188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7532136899705956e-06,
+      "loss": 0.8315,
+      "step": 4189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.752707047675481e-06,
+      "loss": 0.8988,
+      "step": 4190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.752200375706927e-06,
+      "loss": 1.2029,
+      "step": 4191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7516936741244285e-06,
+      "loss": 0.9883,
+      "step": 4192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7511869429874832e-06,
+      "loss": 0.9012,
+      "step": 4193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7506801823555928e-06,
+      "loss": 0.8125,
+      "step": 4194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7501733922882623e-06,
+      "loss": 1.0666,
+      "step": 4195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.749666572845e-06,
+      "loss": 1.1345,
+      "step": 4196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7491597240853182e-06,
+      "loss": 0.8877,
+      "step": 4197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7486528460687316e-06,
+      "loss": 1.2818,
+      "step": 4198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7481459388547596e-06,
+      "loss": 1.1329,
+      "step": 4199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7476390025029241e-06,
+      "loss": 1.2801,
+      "step": 4200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7471320370727508e-06,
+      "loss": 0.6542,
+      "step": 4201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7466250426237685e-06,
+      "loss": 1.0605,
+      "step": 4202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7461180192155099e-06,
+      "loss": 0.8461,
+      "step": 4203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7456109669075107e-06,
+      "loss": 1.142,
+      "step": 4204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7451038857593107e-06,
+      "loss": 1.0122,
+      "step": 4205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7445967758304515e-06,
+      "loss": 0.5962,
+      "step": 4206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.74408963718048e-06,
+      "loss": 1.0282,
+      "step": 4207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7435824698689454e-06,
+      "loss": 1.0362,
+      "step": 4208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7430752739554e-06,
+      "loss": 1.1064,
+      "step": 4209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7425680494994003e-06,
+      "loss": 1.06,
+      "step": 4210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7420607965605057e-06,
+      "loss": 0.8269,
+      "step": 4211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7415535151982792e-06,
+      "loss": 1.2008,
+      "step": 4212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7410462054722864e-06,
+      "loss": 1.2998,
+      "step": 4213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.740538867442097e-06,
+      "loss": 1.0696,
+      "step": 4214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.740031501167284e-06,
+      "loss": 0.8833,
+      "step": 4215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7395241067074235e-06,
+      "loss": 0.9471,
+      "step": 4216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7390166841220947e-06,
+      "loss": 0.6905,
+      "step": 4217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.73850923347088e-06,
+      "loss": 1.1248,
+      "step": 4218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7380017548133665e-06,
+      "loss": 1.1448,
+      "step": 4219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7374942482091424e-06,
+      "loss": 0.9388,
+      "step": 4220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7369867137178007e-06,
+      "loss": 1.0587,
+      "step": 4221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7364791513989378e-06,
+      "loss": 1.1138,
+      "step": 4222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7359715613121514e-06,
+      "loss": 0.876,
+      "step": 4223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7354639435170453e-06,
+      "loss": 0.868,
+      "step": 4224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7349562980732247e-06,
+      "loss": 0.8703,
+      "step": 4225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7344486250402984e-06,
+      "loss": 1.028,
+      "step": 4226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7339409244778785e-06,
+      "loss": 0.8437,
+      "step": 4227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7334331964455803e-06,
+      "loss": 0.8613,
+      "step": 4228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7329254410030222e-06,
+      "loss": 0.7388,
+      "step": 4229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7324176582098268e-06,
+      "loss": 1.1759,
+      "step": 4230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7319098481256188e-06,
+      "loss": 0.6181,
+      "step": 4231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7314020108100256e-06,
+      "loss": 1.0902,
+      "step": 4232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7308941463226795e-06,
+      "loss": 0.716,
+      "step": 4233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7303862547232148e-06,
+      "loss": 0.669,
+      "step": 4234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7298783360712697e-06,
+      "loss": 0.8854,
+      "step": 4235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7293703904264843e-06,
+      "loss": 0.7894,
+      "step": 4236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7288624178485036e-06,
+      "loss": 0.8872,
+      "step": 4237
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7283544183969748e-06,
+      "loss": 1.089,
+      "step": 4238
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.727846392131548e-06,
+      "loss": 1.2127,
+      "step": 4239
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7273383391118761e-06,
+      "loss": 1.2395,
+      "step": 4240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7268302593976177e-06,
+      "loss": 1.1704,
+      "step": 4241
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7263221530484315e-06,
+      "loss": 1.0975,
+      "step": 4242
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7258140201239803e-06,
+      "loss": 0.5836,
+      "step": 4243
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7253058606839307e-06,
+      "loss": 1.0631,
+      "step": 4244
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7247976747879518e-06,
+      "loss": 0.9105,
+      "step": 4245
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7242894624957161e-06,
+      "loss": 0.8582,
+      "step": 4246
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7237812238668987e-06,
+      "loss": 1.1452,
+      "step": 4247
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7232729589611787e-06,
+      "loss": 1.1019,
+      "step": 4248
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7227646678382375e-06,
+      "loss": 0.9041,
+      "step": 4249
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7222563505577596e-06,
+      "loss": 1.1084,
+      "step": 4250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7217480071794324e-06,
+      "loss": 0.7841,
+      "step": 4251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7212396377629476e-06,
+      "loss": 1.4062,
+      "step": 4252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7207312423679991e-06,
+      "loss": 1.1609,
+      "step": 4253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7202228210542831e-06,
+      "loss": 0.74,
+      "step": 4254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7197143738815002e-06,
+      "loss": 0.7777,
+      "step": 4255
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.719205900909353e-06,
+      "loss": 0.8766,
+      "step": 4256
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7186974021975478e-06,
+      "loss": 1.2109,
+      "step": 4257
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7181888778057934e-06,
+      "loss": 1.169,
+      "step": 4258
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7176803277938024e-06,
+      "loss": 1.0662,
+      "step": 4259
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7171717522212895e-06,
+      "loss": 1.3139,
+      "step": 4260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7166631511479735e-06,
+      "loss": 0.7293,
+      "step": 4261
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.716154524633574e-06,
+      "loss": 1.0084,
+      "step": 4262
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7156458727378166e-06,
+      "loss": 0.9367,
+      "step": 4263
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.715137195520428e-06,
+      "loss": 0.9464,
+      "step": 4264
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7146284930411378e-06,
+      "loss": 0.7869,
+      "step": 4265
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.714119765359679e-06,
+      "loss": 0.7626,
+      "step": 4266
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.713611012535788e-06,
+      "loss": 0.9181,
+      "step": 4267
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7131022346292038e-06,
+      "loss": 1.1021,
+      "step": 4268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.712593431699668e-06,
+      "loss": 1.3851,
+      "step": 4269
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.712084603806925e-06,
+      "loss": 1.0465,
+      "step": 4270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.711575751010723e-06,
+      "loss": 0.9506,
+      "step": 4271
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.711066873370813e-06,
+      "loss": 0.5819,
+      "step": 4272
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.710557970946948e-06,
+      "loss": 1.0545,
+      "step": 4273
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7100490437988847e-06,
+      "loss": 1.1117,
+      "step": 4274
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7095400919863823e-06,
+      "loss": 0.8039,
+      "step": 4275
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7090311155692036e-06,
+      "loss": 0.883,
+      "step": 4276
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.708522114607113e-06,
+      "loss": 0.6236,
+      "step": 4277
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.708013089159879e-06,
+      "loss": 1.1774,
+      "step": 4278
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7075040392872728e-06,
+      "loss": 0.8046,
+      "step": 4279
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7069949650490677e-06,
+      "loss": 1.1113,
+      "step": 4280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7064858665050405e-06,
+      "loss": 0.8141,
+      "step": 4281
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7059767437149708e-06,
+      "loss": 0.7885,
+      "step": 4282
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7054675967386413e-06,
+      "loss": 0.7219,
+      "step": 4283
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7049584256358368e-06,
+      "loss": 0.9993,
+      "step": 4284
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.704449230466345e-06,
+      "loss": 1.2357,
+      "step": 4285
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.703940011289957e-06,
+      "loss": 0.7177,
+      "step": 4286
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7034307681664673e-06,
+      "loss": 0.9152,
+      "step": 4287
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7029215011556713e-06,
+      "loss": 0.9229,
+      "step": 4288
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7024122103173689e-06,
+      "loss": 0.991,
+      "step": 4289
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7019028957113618e-06,
+      "loss": 0.5208,
+      "step": 4290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7013935573974553e-06,
+      "loss": 1.5638,
+      "step": 4291
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7008841954354565e-06,
+      "loss": 1.0497,
+      "step": 4292
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7003748098851766e-06,
+      "loss": 1.0456,
+      "step": 4293
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6998654008064282e-06,
+      "loss": 1.1277,
+      "step": 4294
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6993559682590277e-06,
+      "loss": 1.2106,
+      "step": 4295
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.698846512302793e-06,
+      "loss": 1.208,
+      "step": 4296
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.698337032997547e-06,
+      "loss": 1.0626,
+      "step": 4297
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6978275304031127e-06,
+      "loss": 0.932,
+      "step": 4298
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6973180045793175e-06,
+      "loss": 0.8442,
+      "step": 4299
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6968084555859914e-06,
+      "loss": 0.7317,
+      "step": 4300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6962988834829662e-06,
+      "loss": 0.9194,
+      "step": 4301
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6957892883300778e-06,
+      "loss": 1.1869,
+      "step": 4302
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6952796701871628e-06,
+      "loss": 1.0347,
+      "step": 4303
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6947700291140628e-06,
+      "loss": 1.1392,
+      "step": 4304
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6942603651706207e-06,
+      "loss": 1.0005,
+      "step": 4305
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6937506784166828e-06,
+      "loss": 0.9922,
+      "step": 4306
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6932409689120966e-06,
+      "loss": 0.9452,
+      "step": 4307
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6927312367167143e-06,
+      "loss": 0.8859,
+      "step": 4308
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6922214818903898e-06,
+      "loss": 1.0088,
+      "step": 4309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6917117044929795e-06,
+      "loss": 1.0862,
+      "step": 4310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.691201904584342e-06,
+      "loss": 0.9134,
+      "step": 4311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6906920822243403e-06,
+      "loss": 1.1457,
+      "step": 4312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.690182237472838e-06,
+      "loss": 0.9793,
+      "step": 4313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6896723703897025e-06,
+      "loss": 0.767,
+      "step": 4314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.689162481034804e-06,
+      "loss": 0.9249,
+      "step": 4315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6886525694680143e-06,
+      "loss": 1.0276,
+      "step": 4316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6881426357492084e-06,
+      "loss": 1.7737,
+      "step": 4317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6876326799382645e-06,
+      "loss": 1.0852,
+      "step": 4318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6871227020950622e-06,
+      "loss": 1.066,
+      "step": 4319
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6866127022794844e-06,
+      "loss": 1.0052,
+      "step": 4320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6861026805514164e-06,
+      "loss": 1.4381,
+      "step": 4321
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.685592636970746e-06,
+      "loss": 0.9199,
+      "step": 4322
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6850825715973642e-06,
+      "loss": 1.0855,
+      "step": 4323
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6845724844911636e-06,
+      "loss": 1.2311,
+      "step": 4324
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6840623757120395e-06,
+      "loss": 0.7471,
+      "step": 4325
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6835522453198908e-06,
+      "loss": 0.9407,
+      "step": 4326
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.683042093374618e-06,
+      "loss": 0.7282,
+      "step": 4327
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.682531919936124e-06,
+      "loss": 1.3536,
+      "step": 4328
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6820217250643142e-06,
+      "loss": 0.5081,
+      "step": 4329
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6815115088190977e-06,
+      "loss": 0.87,
+      "step": 4330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.681001271260385e-06,
+      "loss": 0.6994,
+      "step": 4331
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6804910124480892e-06,
+      "loss": 0.9987,
+      "step": 4332
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6799807324421258e-06,
+      "loss": 0.7505,
+      "step": 4333
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6794704313024132e-06,
+      "loss": 1.276,
+      "step": 4334
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6789601090888726e-06,
+      "loss": 0.8051,
+      "step": 4335
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6784497658614269e-06,
+      "loss": 1.0943,
+      "step": 4336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6779394016800012e-06,
+      "loss": 0.7962,
+      "step": 4337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6774290166045247e-06,
+      "loss": 0.8746,
+      "step": 4338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.676918610694927e-06,
+      "loss": 1.061,
+      "step": 4339
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6764081840111417e-06,
+      "loss": 0.9369,
+      "step": 4340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.675897736613104e-06,
+      "loss": 0.7964,
+      "step": 4341
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6753872685607526e-06,
+      "loss": 0.9616,
+      "step": 4342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6748767799140265e-06,
+      "loss": 0.8608,
+      "step": 4343
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6743662707328693e-06,
+      "loss": 0.6517,
+      "step": 4344
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6738557410772257e-06,
+      "loss": 0.8947,
+      "step": 4345
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6733451910070438e-06,
+      "loss": 0.9856,
+      "step": 4346
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6728346205822732e-06,
+      "loss": 1.0175,
+      "step": 4347
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6723240298628666e-06,
+      "loss": 0.7868,
+      "step": 4348
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6718134189087778e-06,
+      "loss": 0.9748,
+      "step": 4349
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6713027877799651e-06,
+      "loss": 1.3974,
+      "step": 4350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6707921365363872e-06,
+      "loss": 0.8357,
+      "step": 4351
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.670281465238006e-06,
+      "loss": 0.7783,
+      "step": 4352
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6697707739447863e-06,
+      "loss": 0.9009,
+      "step": 4353
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6692600627166939e-06,
+      "loss": 0.8387,
+      "step": 4354
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6687493316136977e-06,
+      "loss": 0.9214,
+      "step": 4355
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6682385806957695e-06,
+      "loss": 0.8098,
+      "step": 4356
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6677278100228828e-06,
+      "loss": 1.1067,
+      "step": 4357
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.667217019655013e-06,
+      "loss": 0.8114,
+      "step": 4358
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.666706209652138e-06,
+      "loss": 0.9005,
+      "step": 4359
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.666195380074239e-06,
+      "loss": 1.3237,
+      "step": 4360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6656845309812989e-06,
+      "loss": 0.8652,
+      "step": 4361
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.665173662433302e-06,
+      "loss": 0.8205,
+      "step": 4362
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6646627744902358e-06,
+      "loss": 0.7573,
+      "step": 4363
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6641518672120906e-06,
+      "loss": 0.5742,
+      "step": 4364
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6636409406588577e-06,
+      "loss": 1.0902,
+      "step": 4365
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.663129994890531e-06,
+      "loss": 0.9829,
+      "step": 4366
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.662619029967108e-06,
+      "loss": 0.9755,
+      "step": 4367
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6621080459485865e-06,
+      "loss": 0.8114,
+      "step": 4368
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6615970428949678e-06,
+      "loss": 1.009,
+      "step": 4369
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6610860208662544e-06,
+      "loss": 0.8512,
+      "step": 4370
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6605749799224526e-06,
+      "loss": 0.8369,
+      "step": 4371
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6600639201235696e-06,
+      "loss": 0.7988,
+      "step": 4372
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6595528415296146e-06,
+      "loss": 1.0017,
+      "step": 4373
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6590417442006006e-06,
+      "loss": 0.8417,
+      "step": 4374
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6585306281965413e-06,
+      "loss": 0.8507,
+      "step": 4375
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6580194935774536e-06,
+      "loss": 0.9738,
+      "step": 4376
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6575083404033554e-06,
+      "loss": 0.9707,
+      "step": 4377
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.656997168734267e-06,
+      "loss": 1.354,
+      "step": 4378
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.656485978630213e-06,
+      "loss": 0.8077,
+      "step": 4379
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6559747701512177e-06,
+      "loss": 0.7028,
+      "step": 4380
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6554635433573078e-06,
+      "loss": 0.7249,
+      "step": 4381
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6549522983085138e-06,
+      "loss": 0.8283,
+      "step": 4382
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6544410350648665e-06,
+      "loss": 1.142,
+      "step": 4383
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.653929753686399e-06,
+      "loss": 0.8299,
+      "step": 4384
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6534184542331487e-06,
+      "loss": 0.7381,
+      "step": 4385
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6529071367651527e-06,
+      "loss": 1.017,
+      "step": 4386
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.652395801342451e-06,
+      "loss": 0.8516,
+      "step": 4387
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.651884448025086e-06,
+      "loss": 0.8804,
+      "step": 4388
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6513730768731014e-06,
+      "loss": 1.0244,
+      "step": 4389
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6508616879465443e-06,
+      "loss": 0.5594,
+      "step": 4390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.650350281305463e-06,
+      "loss": 1.1213,
+      "step": 4391
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6498388570099078e-06,
+      "loss": 1.1153,
+      "step": 4392
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6493274151199319e-06,
+      "loss": 0.7839,
+      "step": 4393
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6488159556955891e-06,
+      "loss": 1.1059,
+      "step": 4394
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6483044787969368e-06,
+      "loss": 1.0021,
+      "step": 4395
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6477929844840328e-06,
+      "loss": 0.7184,
+      "step": 4396
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6472814728169394e-06,
+      "loss": 1.1833,
+      "step": 4397
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6467699438557188e-06,
+      "loss": 0.7855,
+      "step": 4398
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6462583976604356e-06,
+      "loss": 1.1598,
+      "step": 4399
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6457468342911568e-06,
+      "loss": 1.0506,
+      "step": 4400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.645235253807952e-06,
+      "loss": 0.9296,
+      "step": 4401
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6447236562708912e-06,
+      "loss": 0.9627,
+      "step": 4402
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6442120417400478e-06,
+      "loss": 1.4301,
+      "step": 4403
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.643700410275497e-06,
+      "loss": 1.2241,
+      "step": 4404
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6431887619373154e-06,
+      "loss": 1.0298,
+      "step": 4405
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6426770967855823e-06,
+      "loss": 1.1232,
+      "step": 4406
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6421654148803775e-06,
+      "loss": 0.7414,
+      "step": 4407
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6416537162817853e-06,
+      "loss": 0.8999,
+      "step": 4408
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6411420010498899e-06,
+      "loss": 1.044,
+      "step": 4409
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6406302692447781e-06,
+      "loss": 1.1978,
+      "step": 4410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6401185209265385e-06,
+      "loss": 1.0884,
+      "step": 4411
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6396067561552617e-06,
+      "loss": 0.8547,
+      "step": 4412
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6390949749910405e-06,
+      "loss": 0.8718,
+      "step": 4413
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6385831774939692e-06,
+      "loss": 1.1394,
+      "step": 4414
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6380713637241446e-06,
+      "loss": 0.8713,
+      "step": 4415
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6375595337416646e-06,
+      "loss": 0.727,
+      "step": 4416
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6370476876066298e-06,
+      "loss": 0.9258,
+      "step": 4417
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6365358253791419e-06,
+      "loss": 0.6534,
+      "step": 4418
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6360239471193057e-06,
+      "loss": 0.793,
+      "step": 4419
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6355120528872263e-06,
+      "loss": 0.5565,
+      "step": 4420
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.635000142743012e-06,
+      "loss": 1.3132,
+      "step": 4421
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.634488216746772e-06,
+      "loss": 0.9947,
+      "step": 4422
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6339762749586184e-06,
+      "loss": 0.7739,
+      "step": 4423
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6334643174386643e-06,
+      "loss": 0.7924,
+      "step": 4424
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6329523442470249e-06,
+      "loss": 0.9742,
+      "step": 4425
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.632440355443817e-06,
+      "loss": 1.1258,
+      "step": 4426
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6319283510891602e-06,
+      "loss": 0.9859,
+      "step": 4427
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6314163312431747e-06,
+      "loss": 0.7574,
+      "step": 4428
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6309042959659832e-06,
+      "loss": 0.7987,
+      "step": 4429
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6303922453177098e-06,
+      "loss": 0.8106,
+      "step": 4430
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6298801793584807e-06,
+      "loss": 1.0564,
+      "step": 4431
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6293680981484244e-06,
+      "loss": 0.9403,
+      "step": 4432
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6288560017476698e-06,
+      "loss": 0.8826,
+      "step": 4433
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6283438902163493e-06,
+      "loss": 0.8454,
+      "step": 4434
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6278317636145955e-06,
+      "loss": 0.6994,
+      "step": 4435
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6273196220025442e-06,
+      "loss": 0.7835,
+      "step": 4436
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.626807465440331e-06,
+      "loss": 0.8644,
+      "step": 4437
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6262952939880959e-06,
+      "loss": 0.9343,
+      "step": 4438
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6257831077059787e-06,
+      "loss": 1.0938,
+      "step": 4439
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6252709066541212e-06,
+      "loss": 1.0798,
+      "step": 4440
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6247586908926672e-06,
+      "loss": 0.8279,
+      "step": 4441
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.624246460481763e-06,
+      "loss": 0.8762,
+      "step": 4442
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.623734215481555e-06,
+      "loss": 0.8818,
+      "step": 4443
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6232219559521928e-06,
+      "loss": 0.7192,
+      "step": 4444
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6227096819538269e-06,
+      "loss": 1.016,
+      "step": 4445
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6221973935466095e-06,
+      "loss": 0.7252,
+      "step": 4446
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6216850907906944e-06,
+      "loss": 0.8525,
+      "step": 4447
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6211727737462382e-06,
+      "loss": 0.9015,
+      "step": 4448
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6206604424733977e-06,
+      "loss": 0.8737,
+      "step": 4449
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.620148097032332e-06,
+      "loss": 0.7021,
+      "step": 4450
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6196357374832025e-06,
+      "loss": 0.7929,
+      "step": 4451
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6191233638861706e-06,
+      "loss": 1.0254,
+      "step": 4452
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6186109763014015e-06,
+      "loss": 0.9084,
+      "step": 4453
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.61809857478906e-06,
+      "loss": 0.6008,
+      "step": 4454
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6175861594093141e-06,
+      "loss": 0.8591,
+      "step": 4455
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6170737302223323e-06,
+      "loss": 0.677,
+      "step": 4456
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.616561287288286e-06,
+      "loss": 0.8638,
+      "step": 4457
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6160488306673463e-06,
+      "loss": 1.2444,
+      "step": 4458
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6155363604196875e-06,
+      "loss": 0.9871,
+      "step": 4459
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6150238766054854e-06,
+      "loss": 1.1295,
+      "step": 4460
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6145113792849167e-06,
+      "loss": 0.6605,
+      "step": 4461
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6139988685181602e-06,
+      "loss": 1.1003,
+      "step": 4462
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6134863443653957e-06,
+      "loss": 0.725,
+      "step": 4463
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6129738068868053e-06,
+      "loss": 1.0757,
+      "step": 4464
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6124612561425722e-06,
+      "loss": 0.9403,
+      "step": 4465
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6119486921928815e-06,
+      "loss": 1.0592,
+      "step": 4466
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.611436115097919e-06,
+      "loss": 0.5656,
+      "step": 4467
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6109235249178732e-06,
+      "loss": 0.9071,
+      "step": 4468
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6104109217129339e-06,
+      "loss": 0.9822,
+      "step": 4469
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6098983055432913e-06,
+      "loss": 1.034,
+      "step": 4470
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6093856764691386e-06,
+      "loss": 0.8148,
+      "step": 4471
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6088730345506696e-06,
+      "loss": 0.7677,
+      "step": 4472
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6083603798480802e-06,
+      "loss": 0.9149,
+      "step": 4473
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6078477124215668e-06,
+      "loss": 0.8878,
+      "step": 4474
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6073350323313287e-06,
+      "loss": 0.8252,
+      "step": 4475
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6068223396375658e-06,
+      "loss": 0.9093,
+      "step": 4476
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6063096344004797e-06,
+      "loss": 0.8625,
+      "step": 4477
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.605796916680273e-06,
+      "loss": 0.9128,
+      "step": 4478
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6052841865371505e-06,
+      "loss": 1.0033,
+      "step": 4479
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6047714440313184e-06,
+      "loss": 1.3893,
+      "step": 4480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6042586892229838e-06,
+      "loss": 0.8408,
+      "step": 4481
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.603745922172355e-06,
+      "loss": 0.7506,
+      "step": 4482
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6032331429396434e-06,
+      "loss": 0.9061,
+      "step": 4483
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.60272035158506e-06,
+      "loss": 0.7381,
+      "step": 4484
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6022075481688174e-06,
+      "loss": 0.8252,
+      "step": 4485
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6016947327511317e-06,
+      "loss": 0.8381,
+      "step": 4486
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6011819053922177e-06,
+      "loss": 1.2024,
+      "step": 4487
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6006690661522933e-06,
+      "loss": 1.0154,
+      "step": 4488
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6001562150915764e-06,
+      "loss": 0.9107,
+      "step": 4489
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.599643352270288e-06,
+      "loss": 0.8228,
+      "step": 4490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5991304777486496e-06,
+      "loss": 1.1395,
+      "step": 4491
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5986175915868837e-06,
+      "loss": 0.7793,
+      "step": 4492
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5981046938452148e-06,
+      "loss": 0.7047,
+      "step": 4493
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5975917845838683e-06,
+      "loss": 0.7437,
+      "step": 4494
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5970788638630717e-06,
+      "loss": 1.0996,
+      "step": 4495
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5965659317430525e-06,
+      "loss": 0.9332,
+      "step": 4496
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.596052988284041e-06,
+      "loss": 0.7743,
+      "step": 4497
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.595540033546268e-06,
+      "loss": 0.8612,
+      "step": 4498
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5950270675899659e-06,
+      "loss": 0.7922,
+      "step": 4499
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.594514090475368e-06,
+      "loss": 1.087,
+      "step": 4500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5940011022627097e-06,
+      "loss": 1.068,
+      "step": 4501
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5934881030122276e-06,
+      "loss": 0.8813,
+      "step": 4502
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.592975092784158e-06,
+      "loss": 0.8623,
+      "step": 4503
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5924620716387405e-06,
+      "loss": 0.951,
+      "step": 4504
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5919490396362154e-06,
+      "loss": 0.9129,
+      "step": 4505
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.591435996836824e-06,
+      "loss": 1.0726,
+      "step": 4506
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5909229433008085e-06,
+      "loss": 0.9929,
+      "step": 4507
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5904098790884133e-06,
+      "loss": 0.9773,
+      "step": 4508
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5898968042598835e-06,
+      "loss": 1.0138,
+      "step": 4509
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5893837188754649e-06,
+      "loss": 0.894,
+      "step": 4510
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5888706229954063e-06,
+      "loss": 1.1898,
+      "step": 4511
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5883575166799559e-06,
+      "loss": 1.032,
+      "step": 4512
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5878443999893636e-06,
+      "loss": 0.7723,
+      "step": 4513
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5873312729838813e-06,
+      "loss": 0.9052,
+      "step": 4514
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5868181357237605e-06,
+      "loss": 0.7658,
+      "step": 4515
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5863049882692566e-06,
+      "loss": 0.8084,
+      "step": 4516
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5857918306806233e-06,
+      "loss": 1.043,
+      "step": 4517
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5852786630181173e-06,
+      "loss": 1.014,
+      "step": 4518
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5847654853419954e-06,
+      "loss": 1.0013,
+      "step": 4519
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5842522977125166e-06,
+      "loss": 1.2798,
+      "step": 4520
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5837391001899407e-06,
+      "loss": 0.8519,
+      "step": 4521
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5832258928345276e-06,
+      "loss": 0.8633,
+      "step": 4522
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.58271267570654e-06,
+      "loss": 0.7575,
+      "step": 4523
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5821994488662413e-06,
+      "loss": 0.9073,
+      "step": 4524
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5816862123738952e-06,
+      "loss": 1.0406,
+      "step": 4525
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.581172966289767e-06,
+      "loss": 0.9758,
+      "step": 4526
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.580659710674124e-06,
+      "loss": 0.9768,
+      "step": 4527
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.580146445587233e-06,
+      "loss": 1.2008,
+      "step": 4528
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5796331710893635e-06,
+      "loss": 0.9392,
+      "step": 4529
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5791198872407844e-06,
+      "loss": 1.06,
+      "step": 4530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5786065941017676e-06,
+      "loss": 0.9497,
+      "step": 4531
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5780932917325851e-06,
+      "loss": 1.1699,
+      "step": 4532
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5775799801935095e-06,
+      "loss": 0.9844,
+      "step": 4533
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.577066659544815e-06,
+      "loss": 0.8659,
+      "step": 4534
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5765533298467776e-06,
+      "loss": 1.0615,
+      "step": 4535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5760399911596728e-06,
+      "loss": 1.1281,
+      "step": 4536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.575526643543779e-06,
+      "loss": 1.0093,
+      "step": 4537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5750132870593732e-06,
+      "loss": 0.8585,
+      "step": 4538
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5744999217667363e-06,
+      "loss": 1.1403,
+      "step": 4539
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5739865477261481e-06,
+      "loss": 0.7443,
+      "step": 4540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5734731649978905e-06,
+      "loss": 0.8226,
+      "step": 4541
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5729597736422459e-06,
+      "loss": 0.8204,
+      "step": 4542
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5724463737194975e-06,
+      "loss": 0.9813,
+      "step": 4543
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5719329652899307e-06,
+      "loss": 0.9136,
+      "step": 4544
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.57141954841383e-06,
+      "loss": 0.8116,
+      "step": 4545
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.570906123151483e-06,
+      "loss": 0.9944,
+      "step": 4546
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5703926895631775e-06,
+      "loss": 0.8622,
+      "step": 4547
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.569879247709201e-06,
+      "loss": 0.9114,
+      "step": 4548
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.569365797649843e-06,
+      "loss": 1.128,
+      "step": 4549
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5688523394453948e-06,
+      "loss": 0.9162,
+      "step": 4550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.568338873156148e-06,
+      "loss": 0.7485,
+      "step": 4551
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5678253988423936e-06,
+      "loss": 0.9247,
+      "step": 4552
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5673119165644262e-06,
+      "loss": 1.1505,
+      "step": 4553
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5667984263825398e-06,
+      "loss": 0.8931,
+      "step": 4554
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5662849283570294e-06,
+      "loss": 1.3159,
+      "step": 4555
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5657714225481908e-06,
+      "loss": 0.8492,
+      "step": 4556
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.565257909016322e-06,
+      "loss": 1.0976,
+      "step": 4557
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5647443878217202e-06,
+      "loss": 1.0012,
+      "step": 4558
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5642308590246843e-06,
+      "loss": 0.7148,
+      "step": 4559
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5637173226855136e-06,
+      "loss": 1.0258,
+      "step": 4560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5632037788645096e-06,
+      "loss": 1.1768,
+      "step": 4561
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5626902276219736e-06,
+      "loss": 0.8526,
+      "step": 4562
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5621766690182076e-06,
+      "loss": 0.8511,
+      "step": 4563
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5616631031135148e-06,
+      "loss": 3.703,
+      "step": 4564
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5611495299681995e-06,
+      "loss": 1.0795,
+      "step": 4565
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5606359496425668e-06,
+      "loss": 0.6869,
+      "step": 4566
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.560122362196922e-06,
+      "loss": 0.7196,
+      "step": 4567
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.559608767691572e-06,
+      "loss": 0.9818,
+      "step": 4568
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5590951661868244e-06,
+      "loss": 0.8406,
+      "step": 4569
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5585815577429873e-06,
+      "loss": 1.3976,
+      "step": 4570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5580679424203697e-06,
+      "loss": 0.8474,
+      "step": 4571
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5575543202792816e-06,
+      "loss": 1.0778,
+      "step": 4572
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5570406913800332e-06,
+      "loss": 1.1661,
+      "step": 4573
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.556527055782937e-06,
+      "loss": 0.8903,
+      "step": 4574
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5560134135483042e-06,
+      "loss": 1.1088,
+      "step": 4575
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.555499764736449e-06,
+      "loss": 0.7511,
+      "step": 4576
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.554986109407684e-06,
+      "loss": 0.9694,
+      "step": 4577
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.554472447622324e-06,
+      "loss": 0.6465,
+      "step": 4578
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5539587794406851e-06,
+      "loss": 0.9408,
+      "step": 4579
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5534451049230833e-06,
+      "loss": 0.8825,
+      "step": 4580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5529314241298347e-06,
+      "loss": 0.6219,
+      "step": 4581
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.552417737121257e-06,
+      "loss": 0.7481,
+      "step": 4582
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.551904043957669e-06,
+      "loss": 0.858,
+      "step": 4583
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5513903446993896e-06,
+      "loss": 1.1196,
+      "step": 4584
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5508766394067386e-06,
+      "loss": 0.6314,
+      "step": 4585
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.550362928140036e-06,
+      "loss": 0.8846,
+      "step": 4586
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5498492109596032e-06,
+      "loss": 0.7444,
+      "step": 4587
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5493354879257623e-06,
+      "loss": 0.7289,
+      "step": 4588
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5488217590988353e-06,
+      "loss": 1.1163,
+      "step": 4589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.548308024539146e-06,
+      "loss": 0.7406,
+      "step": 4590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.547794284307018e-06,
+      "loss": 0.8423,
+      "step": 4591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5472805384627757e-06,
+      "loss": 1.1685,
+      "step": 4592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5467667870667446e-06,
+      "loss": 0.7068,
+      "step": 4593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5462530301792504e-06,
+      "loss": 1.1664,
+      "step": 4594
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.54573926786062e-06,
+      "loss": 0.9094,
+      "step": 4595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5452255001711796e-06,
+      "loss": 0.9271,
+      "step": 4596
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5447117271712579e-06,
+      "loss": 0.9044,
+      "step": 4597
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5441979489211828e-06,
+      "loss": 0.8895,
+      "step": 4598
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5436841654812837e-06,
+      "loss": 0.9105,
+      "step": 4599
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.54317037691189e-06,
+      "loss": 1.2724,
+      "step": 4600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5426565832733318e-06,
+      "loss": 0.8496,
+      "step": 4601
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5421427846259402e-06,
+      "loss": 1.0195,
+      "step": 4602
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5416289810300466e-06,
+      "loss": 0.8763,
+      "step": 4603
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5411151725459828e-06,
+      "loss": 0.92,
+      "step": 4604
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5406013592340812e-06,
+      "loss": 0.9341,
+      "step": 4605
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5400875411546755e-06,
+      "loss": 0.8676,
+      "step": 4606
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.539573718368099e-06,
+      "loss": 1.1108,
+      "step": 4607
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5390598909346856e-06,
+      "loss": 1.1874,
+      "step": 4608
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.538546058914771e-06,
+      "loss": 0.8551,
+      "step": 4609
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5380322223686899e-06,
+      "loss": 1.2762,
+      "step": 4610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5375183813567783e-06,
+      "loss": 1.1831,
+      "step": 4611
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5370045359393723e-06,
+      "loss": 0.9757,
+      "step": 4612
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5364906861768094e-06,
+      "loss": 0.7252,
+      "step": 4613
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5359768321294267e-06,
+      "loss": 1.0112,
+      "step": 4614
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.535462973857562e-06,
+      "loss": 0.9783,
+      "step": 4615
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5349491114215538e-06,
+      "loss": 0.9189,
+      "step": 4616
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.534435244881741e-06,
+      "loss": 1.0523,
+      "step": 4617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5339213742984636e-06,
+      "loss": 0.7817,
+      "step": 4618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.53340749973206e-06,
+      "loss": 0.7794,
+      "step": 4619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5328936212428719e-06,
+      "loss": 1.0236,
+      "step": 4620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5323797388912397e-06,
+      "loss": 0.8732,
+      "step": 4621
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5318658527375043e-06,
+      "loss": 1.0074,
+      "step": 4622
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5313519628420076e-06,
+      "loss": 0.6695,
+      "step": 4623
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5308380692650916e-06,
+      "loss": 0.9937,
+      "step": 4624
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5303241720670992e-06,
+      "loss": 0.8714,
+      "step": 4625
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5298102713083736e-06,
+      "loss": 0.985,
+      "step": 4626
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5292963670492568e-06,
+      "loss": 1.0371,
+      "step": 4627
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5287824593500947e-06,
+      "loss": 1.1347,
+      "step": 4628
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5282685482712298e-06,
+      "loss": 0.9617,
+      "step": 4629
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5277546338730077e-06,
+      "loss": 0.7302,
+      "step": 4630
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5272407162157724e-06,
+      "loss": 7.9383,
+      "step": 4631
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5267267953598705e-06,
+      "loss": 1.0627,
+      "step": 4632
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5262128713656469e-06,
+      "loss": 1.0331,
+      "step": 4633
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5256989442934483e-06,
+      "loss": 0.8131,
+      "step": 4634
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5251850142036205e-06,
+      "loss": 0.7665,
+      "step": 4635
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5246710811565107e-06,
+      "loss": 0.6228,
+      "step": 4636
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5241571452124668e-06,
+      "loss": 0.9659,
+      "step": 4637
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5236432064318353e-06,
+      "loss": 0.858,
+      "step": 4638
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5231292648749644e-06,
+      "loss": 1.0135,
+      "step": 4639
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5226153206022024e-06,
+      "loss": 0.8877,
+      "step": 4640
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 1.255910873413086,
+      "eval_runtime": 31.8108,
+      "eval_samples_per_second": 39.138,
+      "eval_steps_per_second": 19.585,
+      "step": 4640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.522101373673898e-06,
+      "loss": 0.8008,
+      "step": 4641
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5215874241503993e-06,
+      "loss": 0.9732,
+      "step": 4642
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.521073472092056e-06,
+      "loss": 0.7095,
+      "step": 4643
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5205595175592178e-06,
+      "loss": 0.8681,
+      "step": 4644
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5200455606122335e-06,
+      "loss": 0.6078,
+      "step": 4645
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5195316013114542e-06,
+      "loss": 0.6003,
+      "step": 4646
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.519017639717229e-06,
+      "loss": 0.9142,
+      "step": 4647
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5185036758899098e-06,
+      "loss": 0.841,
+      "step": 4648
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5179897098898462e-06,
+      "loss": 0.7544,
+      "step": 4649
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5174757417773898e-06,
+      "loss": 0.6131,
+      "step": 4650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.516961771612892e-06,
+      "loss": 1.014,
+      "step": 4651
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.516447799456704e-06,
+      "loss": 1.1217,
+      "step": 4652
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5159338253691776e-06,
+      "loss": 0.8867,
+      "step": 4653
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5154198494106655e-06,
+      "loss": 0.8693,
+      "step": 4654
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5149058716415192e-06,
+      "loss": 1.2542,
+      "step": 4655
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5143918921220914e-06,
+      "loss": 1.1674,
+      "step": 4656
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5138779109127345e-06,
+      "loss": 0.6528,
+      "step": 4657
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5133639280738018e-06,
+      "loss": 0.9058,
+      "step": 4658
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.512849943665646e-06,
+      "loss": 0.9413,
+      "step": 4659
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5123359577486204e-06,
+      "loss": 0.8419,
+      "step": 4660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5118219703830786e-06,
+      "loss": 0.8303,
+      "step": 4661
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.511307981629374e-06,
+      "loss": 0.9389,
+      "step": 4662
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5107939915478604e-06,
+      "loss": 0.7419,
+      "step": 4663
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5102800001988914e-06,
+      "loss": 1.1577,
+      "step": 4664
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5097660076428213e-06,
+      "loss": 0.7815,
+      "step": 4665
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5092520139400045e-06,
+      "loss": 0.838,
+      "step": 4666
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.508738019150795e-06,
+      "loss": 0.7284,
+      "step": 4667
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5082240233355473e-06,
+      "loss": 0.8896,
+      "step": 4668
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5077100265546161e-06,
+      "loss": 0.7008,
+      "step": 4669
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5071960288683557e-06,
+      "loss": 0.8702,
+      "step": 4670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.506682030337121e-06,
+      "loss": 0.9281,
+      "step": 4671
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5061680310212674e-06,
+      "loss": 0.8457,
+      "step": 4672
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5056540309811493e-06,
+      "loss": 1.2093,
+      "step": 4673
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5051400302771222e-06,
+      "loss": 0.7434,
+      "step": 4674
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5046260289695403e-06,
+      "loss": 1.0835,
+      "step": 4675
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5041120271187602e-06,
+      "loss": 1.081,
+      "step": 4676
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5035980247851364e-06,
+      "loss": 0.8337,
+      "step": 4677
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5030840220290241e-06,
+      "loss": 0.9147,
+      "step": 4678
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5025700189107788e-06,
+      "loss": 0.544,
+      "step": 4679
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5020560154907563e-06,
+      "loss": 1.0187,
+      "step": 4680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5015420118293115e-06,
+      "loss": 1.1074,
+      "step": 4681
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5010280079868003e-06,
+      "loss": 0.9933,
+      "step": 4682
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5005140040235779e-06,
+      "loss": 1.1473,
+      "step": 4683
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5e-06,
+      "loss": 0.9043,
+      "step": 4684
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4994859959764222e-06,
+      "loss": 0.8236,
+      "step": 4685
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4989719920132002e-06,
+      "loss": 0.8711,
+      "step": 4686
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.498457988170689e-06,
+      "loss": 1.0482,
+      "step": 4687
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4979439845092438e-06,
+      "loss": 0.9504,
+      "step": 4688
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4974299810892212e-06,
+      "loss": 0.8095,
+      "step": 4689
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.496915977970976e-06,
+      "loss": 0.8402,
+      "step": 4690
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.496401975214864e-06,
+      "loss": 1.2106,
+      "step": 4691
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.49588797288124e-06,
+      "loss": 0.9907,
+      "step": 4692
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4953739710304598e-06,
+      "loss": 0.8719,
+      "step": 4693
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4948599697228785e-06,
+      "loss": 1.1572,
+      "step": 4694
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4943459690188514e-06,
+      "loss": 1.0115,
+      "step": 4695
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.493831968978733e-06,
+      "loss": 0.9906,
+      "step": 4696
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4933179696628793e-06,
+      "loss": 0.8554,
+      "step": 4697
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4928039711316446e-06,
+      "loss": 1.1981,
+      "step": 4698
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4922899734453844e-06,
+      "loss": 1.207,
+      "step": 4699
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4917759766644528e-06,
+      "loss": 0.9671,
+      "step": 4700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4912619808492053e-06,
+      "loss": 0.9816,
+      "step": 4701
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4907479860599957e-06,
+      "loss": 0.6174,
+      "step": 4702
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4902339923571788e-06,
+      "loss": 0.8064,
+      "step": 4703
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4897199998011086e-06,
+      "loss": 0.7253,
+      "step": 4704
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4892060084521399e-06,
+      "loss": 1.193,
+      "step": 4705
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4886920183706263e-06,
+      "loss": 0.9326,
+      "step": 4706
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4881780296169216e-06,
+      "loss": 0.8415,
+      "step": 4707
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4876640422513799e-06,
+      "loss": 1.1452,
+      "step": 4708
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4871500563343544e-06,
+      "loss": 0.972,
+      "step": 4709
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.486636071926199e-06,
+      "loss": 0.8715,
+      "step": 4710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4861220890872656e-06,
+      "loss": 1.2805,
+      "step": 4711
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4856081078779089e-06,
+      "loss": 0.9376,
+      "step": 4712
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4850941283584811e-06,
+      "loss": 1.2452,
+      "step": 4713
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.484580150589335e-06,
+      "loss": 0.9966,
+      "step": 4714
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4840661746308227e-06,
+      "loss": 0.8539,
+      "step": 4715
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4835522005432966e-06,
+      "loss": 0.9497,
+      "step": 4716
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4830382283871089e-06,
+      "loss": 0.6315,
+      "step": 4717
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4825242582226103e-06,
+      "loss": 0.7576,
+      "step": 4718
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.482010290110154e-06,
+      "loss": 1.0293,
+      "step": 4719
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4814963241100905e-06,
+      "loss": 0.8833,
+      "step": 4720
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.480982360282771e-06,
+      "loss": 0.9716,
+      "step": 4721
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4804683986885463e-06,
+      "loss": 0.8145,
+      "step": 4722
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4799544393877668e-06,
+      "loss": 0.7307,
+      "step": 4723
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4794404824407827e-06,
+      "loss": 1.1438,
+      "step": 4724
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4789265279079436e-06,
+      "loss": 0.8173,
+      "step": 4725
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4784125758496006e-06,
+      "loss": 1.0141,
+      "step": 4726
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4778986263261021e-06,
+      "loss": 0.9433,
+      "step": 4727
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4773846793977976e-06,
+      "loss": 1.294,
+      "step": 4728
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4768707351250357e-06,
+      "loss": 0.9299,
+      "step": 4729
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.476356793568165e-06,
+      "loss": 0.7354,
+      "step": 4730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4758428547875335e-06,
+      "loss": 1.3568,
+      "step": 4731
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4753289188434893e-06,
+      "loss": 0.9606,
+      "step": 4732
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4748149857963796e-06,
+      "loss": 1.0755,
+      "step": 4733
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.474301055706552e-06,
+      "loss": 1.0903,
+      "step": 4734
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4737871286343532e-06,
+      "loss": 0.9308,
+      "step": 4735
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4732732046401298e-06,
+      "loss": 0.7764,
+      "step": 4736
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.472759283784228e-06,
+      "loss": 0.9756,
+      "step": 4737
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4722453661269932e-06,
+      "loss": 1.1079,
+      "step": 4738
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.471731451728771e-06,
+      "loss": 0.9395,
+      "step": 4739
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4712175406499054e-06,
+      "loss": 0.7697,
+      "step": 4740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4707036329507428e-06,
+      "loss": 0.9847,
+      "step": 4741
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4701897286916269e-06,
+      "loss": 0.9694,
+      "step": 4742
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4696758279329009e-06,
+      "loss": 0.6737,
+      "step": 4743
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4691619307349087e-06,
+      "loss": 0.8308,
+      "step": 4744
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.468648037157993e-06,
+      "loss": 0.9527,
+      "step": 4745
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4681341472624964e-06,
+      "loss": 1.042,
+      "step": 4746
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.467620261108761e-06,
+      "loss": 0.8136,
+      "step": 4747
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.467106378757128e-06,
+      "loss": 1.0674,
+      "step": 4748
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.46659250026794e-06,
+      "loss": 0.7391,
+      "step": 4749
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4660786257015367e-06,
+      "loss": 0.9938,
+      "step": 4750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.465564755118259e-06,
+      "loss": 0.7346,
+      "step": 4751
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4650508885784465e-06,
+      "loss": 1.1062,
+      "step": 4752
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4645370261424385e-06,
+      "loss": 0.8752,
+      "step": 4753
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4640231678705738e-06,
+      "loss": 1.3382,
+      "step": 4754
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4635093138231905e-06,
+      "loss": 1.1704,
+      "step": 4755
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4629954640606277e-06,
+      "loss": 1.0616,
+      "step": 4756
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.462481618643222e-06,
+      "loss": 0.9337,
+      "step": 4757
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4619677776313104e-06,
+      "loss": 0.6388,
+      "step": 4758
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4614539410852293e-06,
+      "loss": 0.7149,
+      "step": 4759
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4609401090653145e-06,
+      "loss": 0.9816,
+      "step": 4760
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4604262816319016e-06,
+      "loss": 0.5603,
+      "step": 4761
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.459912458845325e-06,
+      "loss": 1.1593,
+      "step": 4762
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4593986407659188e-06,
+      "loss": 1.1483,
+      "step": 4763
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4588848274540175e-06,
+      "loss": 0.8038,
+      "step": 4764
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.458371018969954e-06,
+      "loss": 1.1833,
+      "step": 4765
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.45785721537406e-06,
+      "loss": 1.0259,
+      "step": 4766
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4573434167266687e-06,
+      "loss": 1.1682,
+      "step": 4767
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4568296230881104e-06,
+      "loss": 1.0218,
+      "step": 4768
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4563158345187164e-06,
+      "loss": 0.8673,
+      "step": 4769
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.455802051078817e-06,
+      "loss": 1.2174,
+      "step": 4770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4552882728287422e-06,
+      "loss": 1.2969,
+      "step": 4771
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4547744998288206e-06,
+      "loss": 0.9089,
+      "step": 4772
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4542607321393804e-06,
+      "loss": 0.9523,
+      "step": 4773
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4537469698207497e-06,
+      "loss": 0.9873,
+      "step": 4774
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.453233212933256e-06,
+      "loss": 0.9733,
+      "step": 4775
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4527194615372248e-06,
+      "loss": 1.1091,
+      "step": 4776
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4522057156929826e-06,
+      "loss": 0.6263,
+      "step": 4777
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4516919754608541e-06,
+      "loss": 1.0642,
+      "step": 4778
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4511782409011647e-06,
+      "loss": 0.5704,
+      "step": 4779
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.450664512074238e-06,
+      "loss": 1.2524,
+      "step": 4780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4501507890403971e-06,
+      "loss": 0.9645,
+      "step": 4781
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4496370718599644e-06,
+      "loss": 0.9009,
+      "step": 4782
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.449123360593262e-06,
+      "loss": 0.8337,
+      "step": 4783
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.448609655300611e-06,
+      "loss": 0.8766,
+      "step": 4784
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4480959560423308e-06,
+      "loss": 1.1721,
+      "step": 4785
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4475822628787428e-06,
+      "loss": 1.1699,
+      "step": 4786
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4470685758701655e-06,
+      "loss": 1.0389,
+      "step": 4787
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4465548950769172e-06,
+      "loss": 1.2502,
+      "step": 4788
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.446041220559315e-06,
+      "loss": 1.0301,
+      "step": 4789
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.445527552377676e-06,
+      "loss": 0.8824,
+      "step": 4790
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4450138905923167e-06,
+      "loss": 0.8305,
+      "step": 4791
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4445002352635516e-06,
+      "loss": 0.7667,
+      "step": 4792
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4439865864516959e-06,
+      "loss": 0.7625,
+      "step": 4793
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4434729442170633e-06,
+      "loss": 1.1713,
+      "step": 4794
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.442959308619967e-06,
+      "loss": 0.9361,
+      "step": 4795
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4424456797207187e-06,
+      "loss": 1.0225,
+      "step": 4796
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4419320575796306e-06,
+      "loss": 0.8923,
+      "step": 4797
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.441418442257013e-06,
+      "loss": 1.0582,
+      "step": 4798
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4409048338131759e-06,
+      "loss": 0.621,
+      "step": 4799
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.440391232308428e-06,
+      "loss": 1.2133,
+      "step": 4800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4398776378030781e-06,
+      "loss": 1.0545,
+      "step": 4801
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4393640503574335e-06,
+      "loss": 1.1722,
+      "step": 4802
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4388504700318008e-06,
+      "loss": 0.9456,
+      "step": 4803
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4383368968864857e-06,
+      "loss": 1.0316,
+      "step": 4804
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.437823330981793e-06,
+      "loss": 0.9451,
+      "step": 4805
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.437309772378027e-06,
+      "loss": 0.9699,
+      "step": 4806
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4367962211354902e-06,
+      "loss": 1.0364,
+      "step": 4807
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4362826773144865e-06,
+      "loss": 1.2889,
+      "step": 4808
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.435769140975316e-06,
+      "loss": 0.8348,
+      "step": 4809
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4352556121782803e-06,
+      "loss": 0.9759,
+      "step": 4810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4347420909836784e-06,
+      "loss": 0.677,
+      "step": 4811
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4342285774518095e-06,
+      "loss": 0.8926,
+      "step": 4812
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4337150716429711e-06,
+      "loss": 0.7531,
+      "step": 4813
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4332015736174607e-06,
+      "loss": 1.0326,
+      "step": 4814
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4326880834355737e-06,
+      "loss": 0.5433,
+      "step": 4815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4321746011576065e-06,
+      "loss": 0.9658,
+      "step": 4816
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4316611268438526e-06,
+      "loss": 1.3546,
+      "step": 4817
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4311476605546054e-06,
+      "loss": 1.1655,
+      "step": 4818
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4306342023501572e-06,
+      "loss": 0.8827,
+      "step": 4819
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4301207522908e-06,
+      "loss": 0.7959,
+      "step": 4820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4296073104368235e-06,
+      "loss": 0.9194,
+      "step": 4821
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.429093876848517e-06,
+      "loss": 1.163,
+      "step": 4822
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4285804515861702e-06,
+      "loss": 0.7149,
+      "step": 4823
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4280670347100696e-06,
+      "loss": 0.9857,
+      "step": 4824
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4275536262805028e-06,
+      "loss": 0.9923,
+      "step": 4825
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4270402263577546e-06,
+      "loss": 1.0568,
+      "step": 4826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4265268350021098e-06,
+      "loss": 0.7516,
+      "step": 4827
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4260134522738522e-06,
+      "loss": 1.038,
+      "step": 4828
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4255000782332642e-06,
+      "loss": 1.0564,
+      "step": 4829
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4249867129406267e-06,
+      "loss": 1.0834,
+      "step": 4830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4244733564562215e-06,
+      "loss": 0.8406,
+      "step": 4831
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4239600088403272e-06,
+      "loss": 0.9333,
+      "step": 4832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4234466701532229e-06,
+      "loss": 0.7902,
+      "step": 4833
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4229333404551851e-06,
+      "loss": 1.4,
+      "step": 4834
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.422420019806491e-06,
+      "loss": 0.9976,
+      "step": 4835
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4219067082674156e-06,
+      "loss": 0.8845,
+      "step": 4836
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.421393405898232e-06,
+      "loss": 1.3702,
+      "step": 4837
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4208801127592154e-06,
+      "loss": 1.2271,
+      "step": 4838
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4203668289106368e-06,
+      "loss": 1.0921,
+      "step": 4839
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4198535544127672e-06,
+      "loss": 1.0574,
+      "step": 4840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4193402893258765e-06,
+      "loss": 0.9153,
+      "step": 4841
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4188270337102333e-06,
+      "loss": 0.8642,
+      "step": 4842
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4183137876261053e-06,
+      "loss": 1.0519,
+      "step": 4843
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4178005511337588e-06,
+      "loss": 1.0209,
+      "step": 4844
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.41728732429346e-06,
+      "loss": 0.8919,
+      "step": 4845
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4167741071654725e-06,
+      "loss": 0.9406,
+      "step": 4846
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4162608998100598e-06,
+      "loss": 1.1556,
+      "step": 4847
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4157477022874837e-06,
+      "loss": 0.6858,
+      "step": 4848
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4152345146580051e-06,
+      "loss": 0.8049,
+      "step": 4849
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4147213369818832e-06,
+      "loss": 0.7999,
+      "step": 4850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4142081693193772e-06,
+      "loss": 0.8283,
+      "step": 4851
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4136950117307435e-06,
+      "loss": 0.9008,
+      "step": 4852
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4131818642762392e-06,
+      "loss": 0.6911,
+      "step": 4853
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4126687270161192e-06,
+      "loss": 1.1784,
+      "step": 4854
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.4121556000106365e-06,
+      "loss": 0.8888,
+      "step": 4855
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4116424833200444e-06,
+      "loss": 1.0246,
+      "step": 4856
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.411129377004594e-06,
+      "loss": 1.0708,
+      "step": 4857
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4106162811245352e-06,
+      "loss": 0.716,
+      "step": 4858
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4101031957401167e-06,
+      "loss": 0.7072,
+      "step": 4859
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4095901209115868e-06,
+      "loss": 0.7883,
+      "step": 4860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4090770566991916e-06,
+      "loss": 0.7575,
+      "step": 4861
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4085640031631764e-06,
+      "loss": 0.6909,
+      "step": 4862
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4080509603637849e-06,
+      "loss": 0.9289,
+      "step": 4863
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4075379283612596e-06,
+      "loss": 1.2823,
+      "step": 4864
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4070249072158425e-06,
+      "loss": 0.8557,
+      "step": 4865
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4065118969877731e-06,
+      "loss": 1.1499,
+      "step": 4866
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.40599889773729e-06,
+      "loss": 0.9489,
+      "step": 4867
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4054859095246318e-06,
+      "loss": 1.0556,
+      "step": 4868
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4049729324100342e-06,
+      "loss": 0.984,
+      "step": 4869
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4044599664537322e-06,
+      "loss": 1.1518,
+      "step": 4870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4039470117159594e-06,
+      "loss": 0.7441,
+      "step": 4871
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.403434068256948e-06,
+      "loss": 0.8745,
+      "step": 4872
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.402921136136929e-06,
+      "loss": 0.9623,
+      "step": 4873
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4024082154161318e-06,
+      "loss": 0.8139,
+      "step": 4874
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4018953061547853e-06,
+      "loss": 1.0812,
+      "step": 4875
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4013824084131164e-06,
+      "loss": 0.8983,
+      "step": 4876
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4008695222513507e-06,
+      "loss": 0.7613,
+      "step": 4877
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4003566477297123e-06,
+      "loss": 0.7784,
+      "step": 4878
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.399843784908424e-06,
+      "loss": 0.9661,
+      "step": 4879
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3993309338477076e-06,
+      "loss": 0.8766,
+      "step": 4880
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3988180946077828e-06,
+      "loss": 0.8037,
+      "step": 4881
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3983052672488684e-06,
+      "loss": 0.5523,
+      "step": 4882
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3977924518311822e-06,
+      "loss": 1.1756,
+      "step": 4883
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3972796484149403e-06,
+      "loss": 0.9986,
+      "step": 4884
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.396766857060357e-06,
+      "loss": 0.9296,
+      "step": 4885
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3962540778276453e-06,
+      "loss": 0.834,
+      "step": 4886
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.395741310777017e-06,
+      "loss": 0.8235,
+      "step": 4887
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.395228555968682e-06,
+      "loss": 1.1326,
+      "step": 4888
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3947158134628496e-06,
+      "loss": 0.9195,
+      "step": 4889
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3942030833197272e-06,
+      "loss": 0.6525,
+      "step": 4890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3936903655995206e-06,
+      "loss": 4.525,
+      "step": 4891
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3931776603624343e-06,
+      "loss": 0.9381,
+      "step": 4892
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3926649676686714e-06,
+      "loss": 0.976,
+      "step": 4893
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3921522875784335e-06,
+      "loss": 0.7991,
+      "step": 4894
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3916396201519203e-06,
+      "loss": 0.4908,
+      "step": 4895
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3911269654493306e-06,
+      "loss": 0.7898,
+      "step": 4896
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3906143235308614e-06,
+      "loss": 0.9188,
+      "step": 4897
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3901016944567088e-06,
+      "loss": 0.9859,
+      "step": 4898
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3895890782870664e-06,
+      "loss": 1.0025,
+      "step": 4899
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3890764750821271e-06,
+      "loss": 0.8727,
+      "step": 4900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3885638849020814e-06,
+      "loss": 0.868,
+      "step": 4901
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3880513078071192e-06,
+      "loss": 1.2056,
+      "step": 4902
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3875387438574283e-06,
+      "loss": 0.9903,
+      "step": 4903
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3870261931131946e-06,
+      "loss": 0.9121,
+      "step": 4904
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3865136556346044e-06,
+      "loss": 0.6984,
+      "step": 4905
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.38600113148184e-06,
+      "loss": 0.7799,
+      "step": 4906
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3854886207150833e-06,
+      "loss": 0.9722,
+      "step": 4907
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3849761233945149e-06,
+      "loss": 1.1209,
+      "step": 4908
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3844636395803126e-06,
+      "loss": 1.1307,
+      "step": 4909
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3839511693326542e-06,
+      "loss": 0.7169,
+      "step": 4910
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3834387127117141e-06,
+      "loss": 0.7417,
+      "step": 4911
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3829262697776675e-06,
+      "loss": 1.0939,
+      "step": 4912
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.382413840590686e-06,
+      "loss": 0.7724,
+      "step": 4913
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3819014252109403e-06,
+      "loss": 0.9854,
+      "step": 4914
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3813890236985988e-06,
+      "loss": 0.9919,
+      "step": 4915
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3808766361138295e-06,
+      "loss": 0.9861,
+      "step": 4916
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3803642625167982e-06,
+      "loss": 0.8914,
+      "step": 4917
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3798519029676684e-06,
+      "loss": 0.9158,
+      "step": 4918
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3793395575266026e-06,
+      "loss": 0.9443,
+      "step": 4919
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3788272262537623e-06,
+      "loss": 1.0811,
+      "step": 4920
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3783149092093058e-06,
+      "loss": 1.3089,
+      "step": 4921
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3778026064533908e-06,
+      "loss": 0.6777,
+      "step": 4922
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3772903180461734e-06,
+      "loss": 0.5768,
+      "step": 4923
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3767780440478075e-06,
+      "loss": 1.0745,
+      "step": 4924
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3762657845184452e-06,
+      "loss": 1.0912,
+      "step": 4925
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3757535395182371e-06,
+      "loss": 1.0922,
+      "step": 4926
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3752413091073326e-06,
+      "loss": 0.6692,
+      "step": 4927
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.374729093345879e-06,
+      "loss": 1.1593,
+      "step": 4928
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3742168922940216e-06,
+      "loss": 1.0019,
+      "step": 4929
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3737047060119042e-06,
+      "loss": 0.8119,
+      "step": 4930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3731925345596692e-06,
+      "loss": 1.1369,
+      "step": 4931
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3726803779974565e-06,
+      "loss": 1.0798,
+      "step": 4932
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.372168236385405e-06,
+      "loss": 0.8162,
+      "step": 4933
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3716561097836508e-06,
+      "loss": 0.6469,
+      "step": 4934
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3711439982523303e-06,
+      "loss": 1.0969,
+      "step": 4935
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3706319018515758e-06,
+      "loss": 1.0706,
+      "step": 4936
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3701198206415194e-06,
+      "loss": 0.9633,
+      "step": 4937
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3696077546822907e-06,
+      "loss": 0.9333,
+      "step": 4938
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3690957040340173e-06,
+      "loss": 0.6104,
+      "step": 4939
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3685836687568258e-06,
+      "loss": 1.0161,
+      "step": 4940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3680716489108397e-06,
+      "loss": 0.8222,
+      "step": 4941
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3675596445561828e-06,
+      "loss": 0.8591,
+      "step": 4942
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3670476557529752e-06,
+      "loss": 1.0169,
+      "step": 4943
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3665356825613357e-06,
+      "loss": 1.1299,
+      "step": 4944
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3660237250413817e-06,
+      "loss": 1.1539,
+      "step": 4945
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.365511783253228e-06,
+      "loss": 1.1139,
+      "step": 4946
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3649998572569886e-06,
+      "loss": 0.905,
+      "step": 4947
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3644879471127742e-06,
+      "loss": 1.0048,
+      "step": 4948
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3639760528806944e-06,
+      "loss": 1.0199,
+      "step": 4949
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.363464174620858e-06,
+      "loss": 1.0005,
+      "step": 4950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3629523123933705e-06,
+      "loss": 1.0854,
+      "step": 4951
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3624404662583356e-06,
+      "loss": 0.867,
+      "step": 4952
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3619286362758557e-06,
+      "loss": 0.8117,
+      "step": 4953
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.361416822506031e-06,
+      "loss": 0.8809,
+      "step": 4954
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.36090502500896e-06,
+      "loss": 0.8877,
+      "step": 4955
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3603932438447384e-06,
+      "loss": 0.7033,
+      "step": 4956
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3598814790734616e-06,
+      "loss": 0.734,
+      "step": 4957
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3593697307552221e-06,
+      "loss": 1.104,
+      "step": 4958
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3588579989501102e-06,
+      "loss": 0.9694,
+      "step": 4959
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3583462837182147e-06,
+      "loss": 1.1275,
+      "step": 4960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3578345851196226e-06,
+      "loss": 0.6905,
+      "step": 4961
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3573229032144184e-06,
+      "loss": 0.9673,
+      "step": 4962
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.356811238062685e-06,
+      "loss": 0.8317,
+      "step": 4963
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.356299589724503e-06,
+      "loss": 0.8765,
+      "step": 4964
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3557879582599522e-06,
+      "loss": 1.106,
+      "step": 4965
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.355276343729109e-06,
+      "loss": 0.9165,
+      "step": 4966
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3547647461920486e-06,
+      "loss": 0.7364,
+      "step": 4967
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3542531657088435e-06,
+      "loss": 1.2841,
+      "step": 4968
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.353741602339565e-06,
+      "loss": 0.864,
+      "step": 4969
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3532300561442819e-06,
+      "loss": 1.1829,
+      "step": 4970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3527185271830605e-06,
+      "loss": 0.6971,
+      "step": 4971
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.352207015515967e-06,
+      "loss": 0.8143,
+      "step": 4972
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3516955212030635e-06,
+      "loss": 0.9369,
+      "step": 4973
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3511840443044112e-06,
+      "loss": 0.9989,
+      "step": 4974
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3506725848800686e-06,
+      "loss": 1.2327,
+      "step": 4975
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3501611429900924e-06,
+      "loss": 1.0217,
+      "step": 4976
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3496497186945373e-06,
+      "loss": 1.1602,
+      "step": 4977
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3491383120534554e-06,
+      "loss": 0.8961,
+      "step": 4978
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3486269231268985e-06,
+      "loss": 0.6816,
+      "step": 4979
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3481155519749144e-06,
+      "loss": 0.9985,
+      "step": 4980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3476041986575494e-06,
+      "loss": 1.1164,
+      "step": 4981
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3470928632348478e-06,
+      "loss": 0.9392,
+      "step": 4982
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3465815457668516e-06,
+      "loss": 1.3598,
+      "step": 4983
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3460702463136012e-06,
+      "loss": 0.7446,
+      "step": 4984
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3455589649351342e-06,
+      "loss": 1.3451,
+      "step": 4985
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3450477016914865e-06,
+      "loss": 1.0962,
+      "step": 4986
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3445364566426923e-06,
+      "loss": 0.8267,
+      "step": 4987
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3440252298487828e-06,
+      "loss": 0.9075,
+      "step": 4988
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3435140213697871e-06,
+      "loss": 1.0633,
+      "step": 4989
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.343002831265733e-06,
+      "loss": 1.0466,
+      "step": 4990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3424916595966455e-06,
+      "loss": 0.689,
+      "step": 4991
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3419805064225473e-06,
+      "loss": 0.6899,
+      "step": 4992
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3414693718034586e-06,
+      "loss": 1.0334,
+      "step": 4993
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.3409582557993994e-06,
+      "loss": 0.996,
+      "step": 4994
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3404471584703854e-06,
+      "loss": 0.8663,
+      "step": 4995
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.339936079876431e-06,
+      "loss": 0.7192,
+      "step": 4996
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3394250200775477e-06,
+      "loss": 0.8029,
+      "step": 4997
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3389139791337459e-06,
+      "loss": 1.1833,
+      "step": 4998
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.338402957105033e-06,
+      "loss": 1.1582,
+      "step": 4999
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.337891954051414e-06,
+      "loss": 0.9414,
+      "step": 5000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3373809700328919e-06,
+      "loss": 1.0018,
+      "step": 5001
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3368700051094688e-06,
+      "loss": 0.7182,
+      "step": 5002
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3363590593411426e-06,
+      "loss": 1.2684,
+      "step": 5003
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.33584813278791e-06,
+      "loss": 0.8968,
+      "step": 5004
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3353372255097645e-06,
+      "loss": 0.7109,
+      "step": 5005
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3348263375666985e-06,
+      "loss": 1.1712,
+      "step": 5006
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.334315469018702e-06,
+      "loss": 0.8192,
+      "step": 5007
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3338046199257608e-06,
+      "loss": 1.044,
+      "step": 5008
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.333293790347862e-06,
+      "loss": 0.9414,
+      "step": 5009
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3327829803449875e-06,
+      "loss": 0.9695,
+      "step": 5010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3322721899771175e-06,
+      "loss": 0.6701,
+      "step": 5011
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3317614193042305e-06,
+      "loss": 1.3061,
+      "step": 5012
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3312506683863023e-06,
+      "loss": 0.9289,
+      "step": 5013
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3307399372833066e-06,
+      "loss": 0.6164,
+      "step": 5014
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.3302292260552144e-06,
+      "loss": 0.8431,
+      "step": 5015
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.329718534761994e-06,
+      "loss": 0.915,
+      "step": 5016
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.329207863463613e-06,
+      "loss": 0.8227,
+      "step": 5017
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3286972122200352e-06,
+      "loss": 0.9067,
+      "step": 5018
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3281865810912225e-06,
+      "loss": 0.7254,
+      "step": 5019
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3276759701371339e-06,
+      "loss": 1.2454,
+      "step": 5020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3271653794177269e-06,
+      "loss": 0.7986,
+      "step": 5021
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3266548089929563e-06,
+      "loss": 0.7706,
+      "step": 5022
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3261442589227741e-06,
+      "loss": 1.0138,
+      "step": 5023
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3256337292671308e-06,
+      "loss": 0.4142,
+      "step": 5024
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3251232200859736e-06,
+      "loss": 1.0367,
+      "step": 5025
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.324612731439248e-06,
+      "loss": 0.9717,
+      "step": 5026
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.324102263386896e-06,
+      "loss": 0.675,
+      "step": 5027
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3235918159888586e-06,
+      "loss": 1.0115,
+      "step": 5028
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3230813893050733e-06,
+      "loss": 1.0298,
+      "step": 5029
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.322570983395476e-06,
+      "loss": 1.0714,
+      "step": 5030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3220605983199987e-06,
+      "loss": 0.7385,
+      "step": 5031
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3215502341385734e-06,
+      "loss": 0.8049,
+      "step": 5032
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3210398909111275e-06,
+      "loss": 0.9971,
+      "step": 5033
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3205295686975871e-06,
+      "loss": 1.1559,
+      "step": 5034
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3200192675578747e-06,
+      "loss": 1.1394,
+      "step": 5035
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3195089875519115e-06,
+      "loss": 0.5379,
+      "step": 5036
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3189987287396157e-06,
+      "loss": 1.2706,
+      "step": 5037
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3184884911809021e-06,
+      "loss": 1.0968,
+      "step": 5038
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3179782749356858e-06,
+      "loss": 1.1031,
+      "step": 5039
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3174680800638764e-06,
+      "loss": 0.9128,
+      "step": 5040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3169579066253822e-06,
+      "loss": 0.7973,
+      "step": 5041
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3164477546801095e-06,
+      "loss": 0.9056,
+      "step": 5042
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3159376242879606e-06,
+      "loss": 0.751,
+      "step": 5043
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3154275155088372e-06,
+      "loss": 1.1284,
+      "step": 5044
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3149174284026363e-06,
+      "loss": 1.1953,
+      "step": 5045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3144073630292544e-06,
+      "loss": 0.8494,
+      "step": 5046
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3138973194485837e-06,
+      "loss": 0.8173,
+      "step": 5047
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3133872977205158e-06,
+      "loss": 0.7158,
+      "step": 5048
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.312877297904938e-06,
+      "loss": 1.03,
+      "step": 5049
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3123673200617357e-06,
+      "loss": 0.9833,
+      "step": 5050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3118573642507919e-06,
+      "loss": 0.8674,
+      "step": 5051
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3113474305319862e-06,
+      "loss": 1.0072,
+      "step": 5052
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3108375189651961e-06,
+      "loss": 0.9005,
+      "step": 5053
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3103276296102975e-06,
+      "loss": 1.116,
+      "step": 5054
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3098177625271624e-06,
+      "loss": 0.9757,
+      "step": 5055
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3093079177756602e-06,
+      "loss": 0.9509,
+      "step": 5056
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3087980954156582e-06,
+      "loss": 0.8378,
+      "step": 5057
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3082882955070212e-06,
+      "loss": 0.9119,
+      "step": 5058
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3077785181096107e-06,
+      "loss": 1.0209,
+      "step": 5059
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3072687632832853e-06,
+      "loss": 0.8976,
+      "step": 5060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3067590310879032e-06,
+      "loss": 1.1455,
+      "step": 5061
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3062493215833175e-06,
+      "loss": 0.6794,
+      "step": 5062
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.3057396348293793e-06,
+      "loss": 0.728,
+      "step": 5063
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3052299708859373e-06,
+      "loss": 0.8789,
+      "step": 5064
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3047203298128375e-06,
+      "loss": 0.8877,
+      "step": 5065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.304210711669923e-06,
+      "loss": 0.9553,
+      "step": 5066
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3037011165170343e-06,
+      "loss": 0.8033,
+      "step": 5067
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3031915444140086e-06,
+      "loss": 0.6761,
+      "step": 5068
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3026819954206824e-06,
+      "loss": 1.0453,
+      "step": 5069
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3021724695968874e-06,
+      "loss": 1.0234,
+      "step": 5070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3016629670024534e-06,
+      "loss": 1.053,
+      "step": 5071
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.301153487697207e-06,
+      "loss": 0.7365,
+      "step": 5072
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.300644031740973e-06,
+      "loss": 0.9073,
+      "step": 5073
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.3001345991935723e-06,
+      "loss": 1.2211,
+      "step": 5074
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2996251901148235e-06,
+      "loss": 0.923,
+      "step": 5075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2991158045645434e-06,
+      "loss": 1.1176,
+      "step": 5076
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.298606442602545e-06,
+      "loss": 1.1306,
+      "step": 5077
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2980971042886383e-06,
+      "loss": 1.0646,
+      "step": 5078
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2975877896826314e-06,
+      "loss": 1.2412,
+      "step": 5079
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2970784988443288e-06,
+      "loss": 1.0118,
+      "step": 5080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2965692318335332e-06,
+      "loss": 1.0385,
+      "step": 5081
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.296059988710043e-06,
+      "loss": 0.8514,
+      "step": 5082
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.295550769533655e-06,
+      "loss": 0.8339,
+      "step": 5083
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2950415743641635e-06,
+      "loss": 0.7652,
+      "step": 5084
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.294532403261359e-06,
+      "loss": 0.9372,
+      "step": 5085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2940232562850292e-06,
+      "loss": 0.8763,
+      "step": 5086
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2935141334949598e-06,
+      "loss": 0.8192,
+      "step": 5087
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2930050349509328e-06,
+      "loss": 1.0884,
+      "step": 5088
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.292495960712728e-06,
+      "loss": 1.021,
+      "step": 5089
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2919869108401208e-06,
+      "loss": 0.9407,
+      "step": 5090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.291477885392887e-06,
+      "loss": 0.8128,
+      "step": 5091
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2909688844307967e-06,
+      "loss": 1.022,
+      "step": 5092
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2904599080136178e-06,
+      "loss": 0.8021,
+      "step": 5093
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2899509562011156e-06,
+      "loss": 0.5246,
+      "step": 5094
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2894420290530524e-06,
+      "loss": 0.6505,
+      "step": 5095
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2889331266291875e-06,
+      "loss": 1.0124,
+      "step": 5096
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2884242489892772e-06,
+      "loss": 0.474,
+      "step": 5097
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.287915396193075e-06,
+      "loss": 0.5359,
+      "step": 5098
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2874065683003323e-06,
+      "loss": 0.912,
+      "step": 5099
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2868977653707963e-06,
+      "loss": 0.9761,
+      "step": 5100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2863889874642122e-06,
+      "loss": 1.1104,
+      "step": 5101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2858802346403213e-06,
+      "loss": 0.5643,
+      "step": 5102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2853715069588629e-06,
+      "loss": 0.8707,
+      "step": 5103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2848628044795726e-06,
+      "loss": 0.9085,
+      "step": 5104
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2843541272621832e-06,
+      "loss": 1.2015,
+      "step": 5105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2838454753664258e-06,
+      "loss": 0.7874,
+      "step": 5106
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.283336848852027e-06,
+      "loss": 0.8102,
+      "step": 5107
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2828282477787105e-06,
+      "loss": 0.6175,
+      "step": 5108
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2823196722061979e-06,
+      "loss": 1.1287,
+      "step": 5109
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.281811122194207e-06,
+      "loss": 1.1952,
+      "step": 5110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2813025978024529e-06,
+      "loss": 0.7769,
+      "step": 5111
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2807940990906473e-06,
+      "loss": 1.0825,
+      "step": 5112
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2802856261185001e-06,
+      "loss": 1.1316,
+      "step": 5113
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2797771789457172e-06,
+      "loss": 0.9168,
+      "step": 5114
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2792687576320014e-06,
+      "loss": 0.6459,
+      "step": 5115
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2787603622370525e-06,
+      "loss": 0.8956,
+      "step": 5116
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2782519928205679e-06,
+      "loss": 0.7492,
+      "step": 5117
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2777436494422409e-06,
+      "loss": 1.0082,
+      "step": 5118
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2772353321617628e-06,
+      "loss": 1.0413,
+      "step": 5119
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2767270410388214e-06,
+      "loss": 0.8379,
+      "step": 5120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2762187761331012e-06,
+      "loss": 0.6841,
+      "step": 5121
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2757105375042842e-06,
+      "loss": 0.9415,
+      "step": 5122
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2752023252120483e-06,
+      "loss": 0.8955,
+      "step": 5123
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2746941393160698e-06,
+      "loss": 0.8463,
+      "step": 5124
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2741859798760204e-06,
+      "loss": 1.2037,
+      "step": 5125
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2736778469515694e-06,
+      "loss": 1.2971,
+      "step": 5126
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2731697406023824e-06,
+      "loss": 0.9764,
+      "step": 5127
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2726616608881235e-06,
+      "loss": 0.6562,
+      "step": 5128
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2721536078684525e-06,
+      "loss": 0.7698,
+      "step": 5129
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2716455816030255e-06,
+      "loss": 0.8674,
+      "step": 5130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2711375821514967e-06,
+      "loss": 1.0503,
+      "step": 5131
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.270629609573516e-06,
+      "loss": 0.8876,
+      "step": 5132
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2701216639287308e-06,
+      "loss": 0.8905,
+      "step": 5133
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2696137452767855e-06,
+      "loss": 0.8391,
+      "step": 5134
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2691058536773204e-06,
+      "loss": 1.1749,
+      "step": 5135
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2685979891899745e-06,
+      "loss": 0.6852,
+      "step": 5136
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2680901518743817e-06,
+      "loss": 1.2155,
+      "step": 5137
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2675823417901733e-06,
+      "loss": 0.669,
+      "step": 5138
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2670745589969778e-06,
+      "loss": 0.9381,
+      "step": 5139
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2665668035544202e-06,
+      "loss": 0.9538,
+      "step": 5140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.266059075522122e-06,
+      "loss": 1.0348,
+      "step": 5141
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2655513749597019e-06,
+      "loss": 1.3019,
+      "step": 5142
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2650437019267756e-06,
+      "loss": 1.4363,
+      "step": 5143
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.264536056482955e-06,
+      "loss": 1.1112,
+      "step": 5144
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.264028438687849e-06,
+      "loss": 0.9141,
+      "step": 5145
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2635208486010627e-06,
+      "loss": 1.0754,
+      "step": 5146
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2630132862821993e-06,
+      "loss": 0.9186,
+      "step": 5147
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2625057517908577e-06,
+      "loss": 1.1649,
+      "step": 5148
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.261998245186634e-06,
+      "loss": 1.0225,
+      "step": 5149
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2614907665291198e-06,
+      "loss": 0.9245,
+      "step": 5150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2609833158779054e-06,
+      "loss": 0.7669,
+      "step": 5151
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2604758932925768e-06,
+      "loss": 1.1806,
+      "step": 5152
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.259968498832716e-06,
+      "loss": 0.6936,
+      "step": 5153
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2594611325579033e-06,
+      "loss": 0.8462,
+      "step": 5154
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.258953794527714e-06,
+      "loss": 1.0719,
+      "step": 5155
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2584464848017215e-06,
+      "loss": 0.7496,
+      "step": 5156
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2579392034394941e-06,
+      "loss": 0.9445,
+      "step": 5157
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2574319505005997e-06,
+      "loss": 1.0566,
+      "step": 5158
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2569247260446e-06,
+      "loss": 0.87,
+      "step": 5159
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2564175301310549e-06,
+      "loss": 1.1433,
+      "step": 5160
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.25591036281952e-06,
+      "loss": 0.8836,
+      "step": 5161
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2554032241695486e-06,
+      "loss": 0.7654,
+      "step": 5162
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2548961142406898e-06,
+      "loss": 1.0668,
+      "step": 5163
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.254389033092489e-06,
+      "loss": 1.1037,
+      "step": 5164
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2538819807844902e-06,
+      "loss": 1.0945,
+      "step": 5165
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2533749573762317e-06,
+      "loss": 1.1307,
+      "step": 5166
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2528679629272497e-06,
+      "loss": 0.8638,
+      "step": 5167
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2523609974970764e-06,
+      "loss": 0.8069,
+      "step": 5168
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2518540611452409e-06,
+      "loss": 1.0769,
+      "step": 5169
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2513471539312687e-06,
+      "loss": 1.1874,
+      "step": 5170
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2508402759146823e-06,
+      "loss": 1.0229,
+      "step": 5171
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2503334271550002e-06,
+      "loss": 0.8218,
+      "step": 5172
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2498266077117376e-06,
+      "loss": 0.817,
+      "step": 5173
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2493198176444073e-06,
+      "loss": 1.3177,
+      "step": 5174
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2488130570125169e-06,
+      "loss": 0.8488,
+      "step": 5175
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2483063258755715e-06,
+      "loss": 1.0535,
+      "step": 5176
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.247799624293073e-06,
+      "loss": 0.9755,
+      "step": 5177
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2472929523245192e-06,
+      "loss": 1.0674,
+      "step": 5178
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2467863100294043e-06,
+      "loss": 0.8505,
+      "step": 5179
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.24627969746722e-06,
+      "loss": 0.6414,
+      "step": 5180
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2457731146974538e-06,
+      "loss": 1.0963,
+      "step": 5181
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.24526656177959e-06,
+      "loss": 0.87,
+      "step": 5182
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2447600387731084e-06,
+      "loss": 0.6958,
+      "step": 5183
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2442535457374866e-06,
+      "loss": 0.9776,
+      "step": 5184
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2437470827321983e-06,
+      "loss": 0.9333,
+      "step": 5185
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2432406498167134e-06,
+      "loss": 0.7893,
+      "step": 5186
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2427342470504978e-06,
+      "loss": 0.9548,
+      "step": 5187
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2422278744930158e-06,
+      "loss": 0.7414,
+      "step": 5188
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.241721532203726e-06,
+      "loss": 1.1615,
+      "step": 5189
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2412152202420842e-06,
+      "loss": 0.9259,
+      "step": 5190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2407089386675431e-06,
+      "loss": 0.8901,
+      "step": 5191
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2402026875395512e-06,
+      "loss": 0.8066,
+      "step": 5192
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2396964669175536e-06,
+      "loss": 1.0259,
+      "step": 5193
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2391902768609914e-06,
+      "loss": 1.0984,
+      "step": 5194
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2386841174293038e-06,
+      "loss": 1.1064,
+      "step": 5195
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2381779886819245e-06,
+      "loss": 0.8569,
+      "step": 5196
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2376718906782844e-06,
+      "loss": 0.9969,
+      "step": 5197
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2371658234778106e-06,
+      "loss": 1.2771,
+      "step": 5198
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2366597871399268e-06,
+      "loss": 0.7568,
+      "step": 5199
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2361537817240527e-06,
+      "loss": 0.9683,
+      "step": 5200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.235647807289605e-06,
+      "loss": 0.9355,
+      "step": 5201
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.235141863895995e-06,
+      "loss": 0.9439,
+      "step": 5202
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.234635951602634e-06,
+      "loss": 0.7679,
+      "step": 5203
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2341300704689261e-06,
+      "loss": 1.2125,
+      "step": 5204
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.233624220554273e-06,
+      "loss": 1.0356,
+      "step": 5205
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2331184019180732e-06,
+      "loss": 1.2578,
+      "step": 5206
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2326126146197207e-06,
+      "loss": 0.8292,
+      "step": 5207
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2321068587186064e-06,
+      "loss": 1.059,
+      "step": 5208
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2316011342741167e-06,
+      "loss": 0.9682,
+      "step": 5209
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2310954413456356e-06,
+      "loss": 1.0102,
+      "step": 5210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2305897799925428e-06,
+      "loss": 0.8268,
+      "step": 5211
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.230084150274214e-06,
+      "loss": 0.7343,
+      "step": 5212
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2295785522500212e-06,
+      "loss": 1.3721,
+      "step": 5213
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2290729859793333e-06,
+      "loss": 1.0903,
+      "step": 5214
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2285674515215145e-06,
+      "loss": 1.1619,
+      "step": 5215
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2280619489359265e-06,
+      "loss": 0.9392,
+      "step": 5216
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.227556478281925e-06,
+      "loss": 0.9834,
+      "step": 5217
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2270510396188656e-06,
+      "loss": 1.0274,
+      "step": 5218
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2265456330060972e-06,
+      "loss": 0.6784,
+      "step": 5219
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2260402585029658e-06,
+      "loss": 0.8498,
+      "step": 5220
+    },
+    {
+      "epoch": 2.21,
+      "eval_loss": 1.2553883790969849,
+      "eval_runtime": 32.0755,
+      "eval_samples_per_second": 38.815,
+      "eval_steps_per_second": 19.423,
+      "step": 5220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2255349161688134e-06,
+      "loss": 0.7775,
+      "step": 5221
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2250296060629785e-06,
+      "loss": 1.0985,
+      "step": 5222
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2245243282447962e-06,
+      "loss": 0.8488,
+      "step": 5223
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2240190827735965e-06,
+      "loss": 0.8491,
+      "step": 5224
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2235138697087074e-06,
+      "loss": 0.7469,
+      "step": 5225
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2230086891094517e-06,
+      "loss": 1.036,
+      "step": 5226
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2225035410351493e-06,
+      "loss": 1.0114,
+      "step": 5227
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2219984255451157e-06,
+      "loss": 0.9593,
+      "step": 5228
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2214933426986623e-06,
+      "loss": 0.7037,
+      "step": 5229
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2209882925550972e-06,
+      "loss": 0.9488,
+      "step": 5230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2204832751737243e-06,
+      "loss": 0.8032,
+      "step": 5231
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2199782906138448e-06,
+      "loss": 0.965,
+      "step": 5232
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2194733389347546e-06,
+      "loss": 1.1399,
+      "step": 5233
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2189684201957465e-06,
+      "loss": 0.6262,
+      "step": 5234
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2184635344561085e-06,
+      "loss": 1.3945,
+      "step": 5235
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2179586817751264e-06,
+      "loss": 0.7874,
+      "step": 5236
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2174538622120804e-06,
+      "loss": 1.0088,
+      "step": 5237
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.216949075826248e-06,
+      "loss": 1.2195,
+      "step": 5238
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2164443226769019e-06,
+      "loss": 0.8252,
+      "step": 5239
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2159396028233122e-06,
+      "loss": 1.0914,
+      "step": 5240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2154349163247436e-06,
+      "loss": 0.9154,
+      "step": 5241
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2149302632404578e-06,
+      "loss": 0.9713,
+      "step": 5242
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.214425643629712e-06,
+      "loss": 0.7903,
+      "step": 5243
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2139210575517598e-06,
+      "loss": 0.8949,
+      "step": 5244
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2134165050658512e-06,
+      "loss": 0.9112,
+      "step": 5245
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2129119862312319e-06,
+      "loss": 0.9781,
+      "step": 5246
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2124075011071436e-06,
+      "loss": 0.8175,
+      "step": 5247
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.211903049752824e-06,
+      "loss": 1.0336,
+      "step": 5248
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.211398632227507e-06,
+      "loss": 0.8713,
+      "step": 5249
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2108942485904222e-06,
+      "loss": 0.7384,
+      "step": 5250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2103898989007961e-06,
+      "loss": 0.9225,
+      "step": 5251
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2098855832178498e-06,
+      "loss": 0.9063,
+      "step": 5252
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.209381301600802e-06,
+      "loss": 0.7522,
+      "step": 5253
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.208877054108865e-06,
+      "loss": 0.8119,
+      "step": 5254
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.208372840801251e-06,
+      "loss": 0.8081,
+      "step": 5255
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2078686617371646e-06,
+      "loss": 0.9575,
+      "step": 5256
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2073645169758077e-06,
+      "loss": 1.2199,
+      "step": 5257
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2068604065763784e-06,
+      "loss": 1.1233,
+      "step": 5258
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2063563305980703e-06,
+      "loss": 1.1075,
+      "step": 5259
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2058522891000731e-06,
+      "loss": 0.6833,
+      "step": 5260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.205348282141572e-06,
+      "loss": 0.9535,
+      "step": 5261
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.20484430978175e-06,
+      "loss": 0.8404,
+      "step": 5262
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.204340372079784e-06,
+      "loss": 0.9941,
+      "step": 5263
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2038364690948472e-06,
+      "loss": 1.079,
+      "step": 5264
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2033326008861094e-06,
+      "loss": 0.631,
+      "step": 5265
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.202828767512736e-06,
+      "loss": 1.0357,
+      "step": 5266
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.202324969033888e-06,
+      "loss": 0.7528,
+      "step": 5267
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2018212055087225e-06,
+      "loss": 1.0235,
+      "step": 5268
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2013174769963927e-06,
+      "loss": 0.7799,
+      "step": 5269
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2008137835560478e-06,
+      "loss": 6.4903,
+      "step": 5270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2003101252468321e-06,
+      "loss": 1.1408,
+      "step": 5271
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1998065021278872e-06,
+      "loss": 0.9125,
+      "step": 5272
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.199302914258349e-06,
+      "loss": 1.0343,
+      "step": 5273
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.19879936169735e-06,
+      "loss": 4.1928,
+      "step": 5274
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1982958445040187e-06,
+      "loss": 0.612,
+      "step": 5275
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1977923627374786e-06,
+      "loss": 0.7106,
+      "step": 5276
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1972889164568504e-06,
+      "loss": 1.0144,
+      "step": 5277
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.19678550572125e-06,
+      "loss": 0.9744,
+      "step": 5278
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1962821305897887e-06,
+      "loss": 0.653,
+      "step": 5279
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1957787911215739e-06,
+      "loss": 0.6347,
+      "step": 5280
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1952754873757088e-06,
+      "loss": 0.9323,
+      "step": 5281
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1947722194112928e-06,
+      "loss": 0.9485,
+      "step": 5282
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1942689872874207e-06,
+      "loss": 1.1124,
+      "step": 5283
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1937657910631825e-06,
+      "loss": 0.6921,
+      "step": 5284
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1932626307976658e-06,
+      "loss": 1.1904,
+      "step": 5285
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1927595065499521e-06,
+      "loss": 0.973,
+      "step": 5286
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1922564183791198e-06,
+      "loss": 0.9564,
+      "step": 5287
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1917533663442422e-06,
+      "loss": 1.0116,
+      "step": 5288
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1912503505043893e-06,
+      "loss": 0.8092,
+      "step": 5289
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1907473709186258e-06,
+      "loss": 1.1995,
+      "step": 5290
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1902444276460125e-06,
+      "loss": 1.1699,
+      "step": 5291
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1897415207456074e-06,
+      "loss": 1.0366,
+      "step": 5292
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1892386502764622e-06,
+      "loss": 0.9922,
+      "step": 5293
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1887358162976252e-06,
+      "loss": 1.1505,
+      "step": 5294
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1882330188681404e-06,
+      "loss": 0.9102,
+      "step": 5295
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1877302580470471e-06,
+      "loss": 0.8839,
+      "step": 5296
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1872275338933808e-06,
+      "loss": 0.9341,
+      "step": 5297
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1867248464661721e-06,
+      "loss": 0.8133,
+      "step": 5298
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1862221958244484e-06,
+      "loss": 1.4248,
+      "step": 5299
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1857195820272323e-06,
+      "loss": 0.9625,
+      "step": 5300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.185217005133541e-06,
+      "loss": 1.067,
+      "step": 5301
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1847144652023891e-06,
+      "loss": 0.8805,
+      "step": 5302
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.184211962292785e-06,
+      "loss": 0.7554,
+      "step": 5303
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1837094964637344e-06,
+      "loss": 1.0664,
+      "step": 5304
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.183207067774238e-06,
+      "loss": 1.0556,
+      "step": 5305
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1827046762832916e-06,
+      "loss": 0.8551,
+      "step": 5306
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1822023220498877e-06,
+      "loss": 0.719,
+      "step": 5307
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1817000051330138e-06,
+      "loss": 0.9996,
+      "step": 5308
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.181197725591653e-06,
+      "loss": 0.6987,
+      "step": 5309
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.180695483484784e-06,
+      "loss": 0.8044,
+      "step": 5310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1801932788713817e-06,
+      "loss": 1.0992,
+      "step": 5311
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1796911118104152e-06,
+      "loss": 0.825,
+      "step": 5312
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1791889823608504e-06,
+      "loss": 0.9986,
+      "step": 5313
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1786868905816492e-06,
+      "loss": 0.6536,
+      "step": 5314
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1781848365317682e-06,
+      "loss": 0.8982,
+      "step": 5315
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.177682820270159e-06,
+      "loss": 0.883,
+      "step": 5316
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1771808418557703e-06,
+      "loss": 1.1213,
+      "step": 5317
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.176678901347545e-06,
+      "loss": 0.8456,
+      "step": 5318
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1761769988044222e-06,
+      "loss": 1.2501,
+      "step": 5319
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1756751342853366e-06,
+      "loss": 1.2543,
+      "step": 5320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1751733078492177e-06,
+      "loss": 1.0468,
+      "step": 5321
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1746715195549922e-06,
+      "loss": 0.9206,
+      "step": 5322
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1741697694615805e-06,
+      "loss": 1.0234,
+      "step": 5323
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1736680576278995e-06,
+      "loss": 1.0093,
+      "step": 5324
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1731663841128615e-06,
+      "loss": 1.1131,
+      "step": 5325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1726647489753735e-06,
+      "loss": 0.9221,
+      "step": 5326
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1721631522743393e-06,
+      "loss": 0.9865,
+      "step": 5327
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1716615940686564e-06,
+      "loss": 0.8229,
+      "step": 5328
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1711600744172206e-06,
+      "loss": 0.7211,
+      "step": 5329
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1706585933789205e-06,
+      "loss": 0.9733,
+      "step": 5330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1701571510126417e-06,
+      "loss": 0.907,
+      "step": 5331
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.169655747377264e-06,
+      "loss": 1.131,
+      "step": 5332
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1691543825316636e-06,
+      "loss": 1.1474,
+      "step": 5333
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1686530565347123e-06,
+      "loss": 1.0332,
+      "step": 5334
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1681517694452763e-06,
+      "loss": 1.0908,
+      "step": 5335
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1676505213222181e-06,
+      "loss": 1.0586,
+      "step": 5336
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1671493122243957e-06,
+      "loss": 1.0358,
+      "step": 5337
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.166648142210662e-06,
+      "loss": 0.7107,
+      "step": 5338
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1661470113398657e-06,
+      "loss": 0.8981,
+      "step": 5339
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1656459196708507e-06,
+      "loss": 0.8059,
+      "step": 5340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1651448672624562e-06,
+      "loss": 0.9758,
+      "step": 5341
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1646438541735165e-06,
+      "loss": 0.7526,
+      "step": 5342
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1641428804628621e-06,
+      "loss": 0.8609,
+      "step": 5343
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.163641946189319e-06,
+      "loss": 0.8529,
+      "step": 5344
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1631410514117072e-06,
+      "loss": 0.9427,
+      "step": 5345
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1626401961888437e-06,
+      "loss": 1.0748,
+      "step": 5346
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1621393805795395e-06,
+      "loss": 1.046,
+      "step": 5347
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1616386046426016e-06,
+      "loss": 1.2588,
+      "step": 5348
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1611378684368324e-06,
+      "loss": 0.8081,
+      "step": 5349
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1606371720210294e-06,
+      "loss": 0.8868,
+      "step": 5350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1601365154539846e-06,
+      "loss": 0.991,
+      "step": 5351
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1596358987944882e-06,
+      "loss": 0.9343,
+      "step": 5352
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1591353221013224e-06,
+      "loss": 0.6657,
+      "step": 5353
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1586347854332668e-06,
+      "loss": 1.0431,
+      "step": 5354
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1581342888490947e-06,
+      "loss": 0.9434,
+      "step": 5355
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1576338324075762e-06,
+      "loss": 0.8932,
+      "step": 5356
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1571334161674758e-06,
+      "loss": 0.6926,
+      "step": 5357
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1566330401875529e-06,
+      "loss": 0.9223,
+      "step": 5358
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.156132704526564e-06,
+      "loss": 1.1165,
+      "step": 5359
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1556324092432591e-06,
+      "loss": 1.1443,
+      "step": 5360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.155132154396384e-06,
+      "loss": 1.3044,
+      "step": 5361
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1546319400446799e-06,
+      "loss": 0.5993,
+      "step": 5362
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.154131766246883e-06,
+      "loss": 1.042,
+      "step": 5363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1536316330617247e-06,
+      "loss": 0.9126,
+      "step": 5364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1531315405479317e-06,
+      "loss": 0.9286,
+      "step": 5365
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1526314887642264e-06,
+      "loss": 0.9093,
+      "step": 5366
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1521314777693262e-06,
+      "loss": 0.7723,
+      "step": 5367
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.151631507621943e-06,
+      "loss": 0.9462,
+      "step": 5368
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1511315783807842e-06,
+      "loss": 1.0319,
+      "step": 5369
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1506316901045533e-06,
+      "loss": 0.9479,
+      "step": 5370
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1501318428519482e-06,
+      "loss": 0.8461,
+      "step": 5371
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.149632036681662e-06,
+      "loss": 1.0551,
+      "step": 5372
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1491322716523829e-06,
+      "loss": 1.1683,
+      "step": 5373
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1486325478227948e-06,
+      "loss": 0.9147,
+      "step": 5374
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1481328652515762e-06,
+      "loss": 0.8889,
+      "step": 5375
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1476332239974012e-06,
+      "loss": 0.8479,
+      "step": 5376
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1471336241189385e-06,
+      "loss": 0.9882,
+      "step": 5377
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.146634065674852e-06,
+      "loss": 0.9199,
+      "step": 5378
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1461345487238018e-06,
+      "loss": 0.6761,
+      "step": 5379
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1456350733244413e-06,
+      "loss": 1.0592,
+      "step": 5380
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1451356395354212e-06,
+      "loss": 1.0004,
+      "step": 5381
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1446362474153857e-06,
+      "loss": 0.8881,
+      "step": 5382
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1441368970229744e-06,
+      "loss": 0.8281,
+      "step": 5383
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1436375884168223e-06,
+      "loss": 0.8842,
+      "step": 5384
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1431383216555591e-06,
+      "loss": 0.6791,
+      "step": 5385
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1426390967978102e-06,
+      "loss": 0.6976,
+      "step": 5386
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1421399139021954e-06,
+      "loss": 0.9302,
+      "step": 5387
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1416407730273295e-06,
+      "loss": 0.9145,
+      "step": 5388
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1411416742318242e-06,
+      "loss": 0.8918,
+      "step": 5389
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1406426175742837e-06,
+      "loss": 1.2348,
+      "step": 5390
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1401436031133086e-06,
+      "loss": 1.0305,
+      "step": 5391
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1396446309074945e-06,
+      "loss": 1.3039,
+      "step": 5392
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1391457010154316e-06,
+      "loss": 0.9937,
+      "step": 5393
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1386468134957054e-06,
+      "loss": 1.0313,
+      "step": 5394
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.138147968406896e-06,
+      "loss": 0.8152,
+      "step": 5395
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1376491658075801e-06,
+      "loss": 1.2447,
+      "step": 5396
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1371504057563272e-06,
+      "loss": 1.2757,
+      "step": 5397
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1366516883117036e-06,
+      "loss": 0.9726,
+      "step": 5398
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.136153013532269e-06,
+      "loss": 1.2059,
+      "step": 5399
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.13565438147658e-06,
+      "loss": 0.8954,
+      "step": 5400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1351557922031861e-06,
+      "loss": 0.9263,
+      "step": 5401
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1346572457706335e-06,
+      "loss": 0.8168,
+      "step": 5402
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1341587422374618e-06,
+      "loss": 0.3203,
+      "step": 5403
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1336602816622075e-06,
+      "loss": 0.6415,
+      "step": 5404
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1331618641034002e-06,
+      "loss": 1.0666,
+      "step": 5405
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1326634896195656e-06,
+      "loss": 0.7638,
+      "step": 5406
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1321651582692237e-06,
+      "loss": 0.6214,
+      "step": 5407
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1316668701108902e-06,
+      "loss": 0.8457,
+      "step": 5408
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1311686252030747e-06,
+      "loss": 0.9098,
+      "step": 5409
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1306704236042818e-06,
+      "loss": 0.7846,
+      "step": 5410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1301722653730128e-06,
+      "loss": 0.8527,
+      "step": 5411
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.129674150567762e-06,
+      "loss": 1.1,
+      "step": 5412
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1291760792470192e-06,
+      "loss": 0.7884,
+      "step": 5413
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1286780514692687e-06,
+      "loss": 1.1804,
+      "step": 5414
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1281800672929905e-06,
+      "loss": 0.7112,
+      "step": 5415
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1276821267766587e-06,
+      "loss": 0.9248,
+      "step": 5416
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1271842299787422e-06,
+      "loss": 1.1373,
+      "step": 5417
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1266863769577063e-06,
+      "loss": 0.6384,
+      "step": 5418
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1261885677720096e-06,
+      "loss": 0.7609,
+      "step": 5419
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1256908024801059e-06,
+      "loss": 1.3385,
+      "step": 5420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.125193081140444e-06,
+      "loss": 0.9717,
+      "step": 5421
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1246954038114673e-06,
+      "loss": 0.8653,
+      "step": 5422
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1241977705516145e-06,
+      "loss": 1.0102,
+      "step": 5423
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1237001814193184e-06,
+      "loss": 1.1443,
+      "step": 5424
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1232026364730068e-06,
+      "loss": 1.1813,
+      "step": 5425
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1227051357711036e-06,
+      "loss": 1.1975,
+      "step": 5426
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.122207679372026e-06,
+      "loss": 1.1125,
+      "step": 5427
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1217102673341865e-06,
+      "loss": 1.0264,
+      "step": 5428
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1212128997159921e-06,
+      "loss": 1.2372,
+      "step": 5429
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1207155765758453e-06,
+      "loss": 0.8794,
+      "step": 5430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1202182979721423e-06,
+      "loss": 0.9407,
+      "step": 5431
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.119721063963275e-06,
+      "loss": 0.8086,
+      "step": 5432
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.11922387460763e-06,
+      "loss": 1.0168,
+      "step": 5433
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.118726729963588e-06,
+      "loss": 1.0615,
+      "step": 5434
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.118229630089525e-06,
+      "loss": 1.1048,
+      "step": 5435
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.117732575043812e-06,
+      "loss": 0.7094,
+      "step": 5436
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1172355648848137e-06,
+      "loss": 0.8766,
+      "step": 5437
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1167385996708905e-06,
+      "loss": 0.9605,
+      "step": 5438
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1162416794603972e-06,
+      "loss": 0.9749,
+      "step": 5439
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1157448043116826e-06,
+      "loss": 0.8983,
+      "step": 5440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1152479742830923e-06,
+      "loss": 0.8734,
+      "step": 5441
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1147511894329643e-06,
+      "loss": 0.9245,
+      "step": 5442
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1142544498196326e-06,
+      "loss": 0.9288,
+      "step": 5443
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.113757755501425e-06,
+      "loss": 0.8636,
+      "step": 5444
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1132611065366652e-06,
+      "loss": 0.6519,
+      "step": 5445
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1127645029836703e-06,
+      "loss": 0.9744,
+      "step": 5446
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.112267944900752e-06,
+      "loss": 0.8793,
+      "step": 5447
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1117714323462188e-06,
+      "loss": 0.9501,
+      "step": 5448
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1112749653783715e-06,
+      "loss": 1.0083,
+      "step": 5449
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1107785440555064e-06,
+      "loss": 0.9109,
+      "step": 5450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1102821684359148e-06,
+      "loss": 0.9918,
+      "step": 5451
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1097858385778817e-06,
+      "loss": 0.5663,
+      "step": 5452
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1092895545396878e-06,
+      "loss": 0.7422,
+      "step": 5453
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1087933163796075e-06,
+      "loss": 0.7777,
+      "step": 5454
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1082971241559097e-06,
+      "loss": 0.923,
+      "step": 5455
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.10780097792686e-06,
+      "loss": 0.6923,
+      "step": 5456
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1073048777507158e-06,
+      "loss": 0.5937,
+      "step": 5457
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1068088236857308e-06,
+      "loss": 0.6745,
+      "step": 5458
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1063128157901525e-06,
+      "loss": 1.0872,
+      "step": 5459
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1058168541222237e-06,
+      "loss": 0.8736,
+      "step": 5460
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.105320938740181e-06,
+      "loss": 1.1257,
+      "step": 5461
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1048250697022553e-06,
+      "loss": 0.871,
+      "step": 5462
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.104329247066674e-06,
+      "loss": 0.6894,
+      "step": 5463
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1038334708916574e-06,
+      "loss": 1.1136,
+      "step": 5464
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.10333774123542e-06,
+      "loss": 1.2706,
+      "step": 5465
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1028420581561719e-06,
+      "loss": 0.8787,
+      "step": 5466
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1023464217121175e-06,
+      "loss": 0.8076,
+      "step": 5467
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.101850831961455e-06,
+      "loss": 1.048,
+      "step": 5468
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1013552889623783e-06,
+      "loss": 1.0998,
+      "step": 5469
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1008597927730746e-06,
+      "loss": 0.178,
+      "step": 5470
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1003643434517271e-06,
+      "loss": 0.651,
+      "step": 5471
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0998689410565119e-06,
+      "loss": 0.9799,
+      "step": 5472
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0993735856456005e-06,
+      "loss": 1.0023,
+      "step": 5473
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0988782772771583e-06,
+      "loss": 0.7209,
+      "step": 5474
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0983830160093463e-06,
+      "loss": 1.0614,
+      "step": 5475
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0978878019003183e-06,
+      "loss": 0.5385,
+      "step": 5476
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0973926350082236e-06,
+      "loss": 0.8723,
+      "step": 5477
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0968975153912068e-06,
+      "loss": 0.9633,
+      "step": 5478
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0964024431074053e-06,
+      "loss": 0.9008,
+      "step": 5479
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0959074182149515e-06,
+      "loss": 0.6825,
+      "step": 5480
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0954124407719726e-06,
+      "loss": 0.8924,
+      "step": 5481
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0949175108365898e-06,
+      "loss": 0.8912,
+      "step": 5482
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.094422628466919e-06,
+      "loss": 0.8748,
+      "step": 5483
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0939277937210698e-06,
+      "loss": 0.847,
+      "step": 5484
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0934330066571481e-06,
+      "loss": 0.945,
+      "step": 5485
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0929382673332524e-06,
+      "loss": 0.9303,
+      "step": 5486
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0924435758074758e-06,
+      "loss": 0.8904,
+      "step": 5487
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0919489321379062e-06,
+      "loss": 0.8871,
+      "step": 5488
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.091454336382626e-06,
+      "loss": 0.7114,
+      "step": 5489
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0909597885997117e-06,
+      "loss": 1.0453,
+      "step": 5490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0904652888472342e-06,
+      "loss": 0.7848,
+      "step": 5491
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0899708371832585e-06,
+      "loss": 1.2517,
+      "step": 5492
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0894764336658446e-06,
+      "loss": 1.0818,
+      "step": 5493
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0889820783530466e-06,
+      "loss": 0.8483,
+      "step": 5494
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0884877713029124e-06,
+      "loss": 1.0954,
+      "step": 5495
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.087993512573485e-06,
+      "loss": 0.4843,
+      "step": 5496
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0874993022228014e-06,
+      "loss": 0.7639,
+      "step": 5497
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0870051403088927e-06,
+      "loss": 0.6345,
+      "step": 5498
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0865110268897845e-06,
+      "loss": 0.7991,
+      "step": 5499
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0860169620234969e-06,
+      "loss": 1.0544,
+      "step": 5500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0855229457680443e-06,
+      "loss": 0.8142,
+      "step": 5501
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.085028978181435e-06,
+      "loss": 0.8939,
+      "step": 5502
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0845350593216715e-06,
+      "loss": 1.0398,
+      "step": 5503
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0840411892467516e-06,
+      "loss": 1.0968,
+      "step": 5504
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0835473680146661e-06,
+      "loss": 0.9991,
+      "step": 5505
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0830535956834008e-06,
+      "loss": 1.054,
+      "step": 5506
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0825598723109347e-06,
+      "loss": 0.9214,
+      "step": 5507
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0820661979552436e-06,
+      "loss": 0.7025,
+      "step": 5508
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.081572572674295e-06,
+      "loss": 0.7603,
+      "step": 5509
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.081078996526051e-06,
+      "loss": 0.8089,
+      "step": 5510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0805854695684696e-06,
+      "loss": 0.7492,
+      "step": 5511
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0800919918595012e-06,
+      "loss": 0.9388,
+      "step": 5512
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.079598563457091e-06,
+      "loss": 0.802,
+      "step": 5513
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.079105184419178e-06,
+      "loss": 1.2445,
+      "step": 5514
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.078611854803697e-06,
+      "loss": 1.2033,
+      "step": 5515
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0781185746685755e-06,
+      "loss": 1.171,
+      "step": 5516
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0776253440717355e-06,
+      "loss": 0.7621,
+      "step": 5517
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0771321630710933e-06,
+      "loss": 1.0256,
+      "step": 5518
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0766390317245592e-06,
+      "loss": 0.6364,
+      "step": 5519
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0761459500900379e-06,
+      "loss": 0.9444,
+      "step": 5520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.075652918225428e-06,
+      "loss": 0.9039,
+      "step": 5521
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0751599361886222e-06,
+      "loss": 0.7789,
+      "step": 5522
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0746670040375085e-06,
+      "loss": 0.8942,
+      "step": 5523
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0741741218299678e-06,
+      "loss": 0.9762,
+      "step": 5524
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.073681289623875e-06,
+      "loss": 1.0485,
+      "step": 5525
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0731885074770999e-06,
+      "loss": 0.9724,
+      "step": 5526
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0726957754475059e-06,
+      "loss": 0.942,
+      "step": 5527
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0722030935929514e-06,
+      "loss": 0.8427,
+      "step": 5528
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.071710461971287e-06,
+      "loss": 0.8424,
+      "step": 5529
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0712178806403595e-06,
+      "loss": 0.9594,
+      "step": 5530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0707253496580088e-06,
+      "loss": 0.8142,
+      "step": 5531
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0702328690820693e-06,
+      "loss": 0.7808,
+      "step": 5532
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0697404389703684e-06,
+      "loss": 0.7498,
+      "step": 5533
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0692480593807288e-06,
+      "loss": 1.2075,
+      "step": 5534
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0687557303709674e-06,
+      "loss": 1.2115,
+      "step": 5535
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0682634519988938e-06,
+      "loss": 1.2559,
+      "step": 5536
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.067771224322312e-06,
+      "loss": 1.021,
+      "step": 5537
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0672790473990218e-06,
+      "loss": 0.3379,
+      "step": 5538
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0667869212868153e-06,
+      "loss": 0.8244,
+      "step": 5539
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0662948460434788e-06,
+      "loss": 1.1714,
+      "step": 5540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.065802821726793e-06,
+      "loss": 0.8891,
+      "step": 5541
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0653108483945327e-06,
+      "loss": 0.6891,
+      "step": 5542
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0648189261044663e-06,
+      "loss": 0.9611,
+      "step": 5543
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.064327054914356e-06,
+      "loss": 1.3206,
+      "step": 5544
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0638352348819593e-06,
+      "loss": 1.0822,
+      "step": 5545
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0633434660650267e-06,
+      "loss": 0.5955,
+      "step": 5546
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0628517485213027e-06,
+      "loss": 0.9566,
+      "step": 5547
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0623600823085256e-06,
+      "loss": 0.8415,
+      "step": 5548
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0618684674844283e-06,
+      "loss": 0.7681,
+      "step": 5549
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0613769041067371e-06,
+      "loss": 1.1137,
+      "step": 5550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0608853922331723e-06,
+      "loss": 0.9755,
+      "step": 5551
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0603939319214488e-06,
+      "loss": 1.0505,
+      "step": 5552
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0599025232292755e-06,
+      "loss": 1.0099,
+      "step": 5553
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0594111662143534e-06,
+      "loss": 1.3881,
+      "step": 5554
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0589198609343798e-06,
+      "loss": 0.8741,
+      "step": 5555
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0584286074470445e-06,
+      "loss": 1.0091,
+      "step": 5556
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0579374058100317e-06,
+      "loss": 0.784,
+      "step": 5557
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0574462560810192e-06,
+      "loss": 1.2409,
+      "step": 5558
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0569551583176788e-06,
+      "loss": 1.0499,
+      "step": 5559
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0564641125776771e-06,
+      "loss": 1.1239,
+      "step": 5560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0559731189186733e-06,
+      "loss": 0.9539,
+      "step": 5561
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0554821773983208e-06,
+      "loss": 0.9617,
+      "step": 5562
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0549912880742676e-06,
+      "loss": 0.9174,
+      "step": 5563
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0545004510041546e-06,
+      "loss": 1.1683,
+      "step": 5564
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0540096662456172e-06,
+      "loss": 0.7619,
+      "step": 5565
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0535189338562837e-06,
+      "loss": 0.8897,
+      "step": 5566
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0530282538937784e-06,
+      "loss": 1.1419,
+      "step": 5567
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0525376264157174e-06,
+      "loss": 1.1083,
+      "step": 5568
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0520470514797115e-06,
+      "loss": 1.0301,
+      "step": 5569
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0515565291433645e-06,
+      "loss": 0.8222,
+      "step": 5570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0510660594642755e-06,
+      "loss": 0.9888,
+      "step": 5571
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0505756425000358e-06,
+      "loss": 0.8828,
+      "step": 5572
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.050085278308232e-06,
+      "loss": 1.0834,
+      "step": 5573
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0495949669464427e-06,
+      "loss": 1.1304,
+      "step": 5574
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0491047084722429e-06,
+      "loss": 0.9505,
+      "step": 5575
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0486145029431988e-06,
+      "loss": 0.9708,
+      "step": 5576
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0481243504168717e-06,
+      "loss": 0.9581,
+      "step": 5577
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0476342509508166e-06,
+      "loss": 0.8152,
+      "step": 5578
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.047144204602582e-06,
+      "loss": 0.9301,
+      "step": 5579
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.04665421142971e-06,
+      "loss": 0.9321,
+      "step": 5580
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0461642714897368e-06,
+      "loss": 0.9386,
+      "step": 5581
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0456743848401926e-06,
+      "loss": 0.8713,
+      "step": 5582
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0451845515386009e-06,
+      "loss": 0.9747,
+      "step": 5583
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0446947716424788e-06,
+      "loss": 0.7594,
+      "step": 5584
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.044205045209338e-06,
+      "loss": 1.0137,
+      "step": 5585
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0437153722966825e-06,
+      "loss": 1.0349,
+      "step": 5586
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0432257529620114e-06,
+      "loss": 0.8547,
+      "step": 5587
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0427361872628166e-06,
+      "loss": 1.1956,
+      "step": 5588
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0422466752565836e-06,
+      "loss": 1.2097,
+      "step": 5589
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.041757217000793e-06,
+      "loss": 1.0321,
+      "step": 5590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0412678125529175e-06,
+      "loss": 1.2391,
+      "step": 5591
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0407784619704244e-06,
+      "loss": 1.0366,
+      "step": 5592
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0402891653107736e-06,
+      "loss": 1.03,
+      "step": 5593
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0397999226314206e-06,
+      "loss": 1.0931,
+      "step": 5594
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0393107339898129e-06,
+      "loss": 0.9874,
+      "step": 5595
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0388215994433915e-06,
+      "loss": 1.0469,
+      "step": 5596
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0383325190495925e-06,
+      "loss": 1.4566,
+      "step": 5597
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0378434928658448e-06,
+      "loss": 0.85,
+      "step": 5598
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0373545209495705e-06,
+      "loss": 1.0095,
+      "step": 5599
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0368656033581863e-06,
+      "loss": 1.0836,
+      "step": 5600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0363767401491018e-06,
+      "loss": 0.6743,
+      "step": 5601
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0358879313797203e-06,
+      "loss": 0.9972,
+      "step": 5602
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0353991771074389e-06,
+      "loss": 0.9784,
+      "step": 5603
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0349104773896479e-06,
+      "loss": 0.9544,
+      "step": 5604
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0344218322837323e-06,
+      "loss": 0.941,
+      "step": 5605
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0339332418470697e-06,
+      "loss": 0.922,
+      "step": 5606
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0334447061370314e-06,
+      "loss": 1.3452,
+      "step": 5607
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0329562252109824e-06,
+      "loss": 1.0233,
+      "step": 5608
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.032467799126281e-06,
+      "loss": 0.8953,
+      "step": 5609
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0319794279402799e-06,
+      "loss": 0.9431,
+      "step": 5610
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0314911117103234e-06,
+      "loss": 1.0185,
+      "step": 5611
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0310028504937524e-06,
+      "loss": 1.2902,
+      "step": 5612
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.030514644347899e-06,
+      "loss": 1.0695,
+      "step": 5613
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0300264933300893e-06,
+      "loss": 0.5198,
+      "step": 5614
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0295383974976434e-06,
+      "loss": 1.0448,
+      "step": 5615
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0290503569078742e-06,
+      "loss": 0.6908,
+      "step": 5616
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.028562371618089e-06,
+      "loss": 0.8946,
+      "step": 5617
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0280744416855875e-06,
+      "loss": 0.794,
+      "step": 5618
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0275865671676642e-06,
+      "loss": 1.2755,
+      "step": 5619
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0270987481216062e-06,
+      "loss": 1.2957,
+      "step": 5620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0266109846046944e-06,
+      "loss": 0.8069,
+      "step": 5621
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0261232766742032e-06,
+      "loss": 1.0499,
+      "step": 5622
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0256356243874e-06,
+      "loss": 0.7847,
+      "step": 5623
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0251480278015468e-06,
+      "loss": 1.0619,
+      "step": 5624
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0246604869738974e-06,
+      "loss": 1.1191,
+      "step": 5625
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0241730019617001e-06,
+      "loss": 0.7128,
+      "step": 5626
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.023685572822197e-06,
+      "loss": 1.0521,
+      "step": 5627
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.023198199612623e-06,
+      "loss": 1.1505,
+      "step": 5628
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0227108823902065e-06,
+      "loss": 1.0007,
+      "step": 5629
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0222236212121696e-06,
+      "loss": 1.1228,
+      "step": 5630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0217364161357271e-06,
+      "loss": 1.0124,
+      "step": 5631
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.021249267218088e-06,
+      "loss": 1.0742,
+      "step": 5632
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0207621745164542e-06,
+      "loss": 0.9745,
+      "step": 5633
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0202751380880218e-06,
+      "loss": 0.9823,
+      "step": 5634
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0197881579899796e-06,
+      "loss": 0.8782,
+      "step": 5635
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0193012342795096e-06,
+      "loss": 0.9444,
+      "step": 5636
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0188143670137878e-06,
+      "loss": 1.0402,
+      "step": 5637
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0183275562499832e-06,
+      "loss": 0.8741,
+      "step": 5638
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.017840802045258e-06,
+      "loss": 0.6655,
+      "step": 5639
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0173541044567682e-06,
+      "loss": 0.7727,
+      "step": 5640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0168674635416623e-06,
+      "loss": 0.9226,
+      "step": 5641
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.016380879357084e-06,
+      "loss": 0.7566,
+      "step": 5642
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0158943519601684e-06,
+      "loss": 1.085,
+      "step": 5643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0154078814080447e-06,
+      "loss": 0.9829,
+      "step": 5644
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0149214677578358e-06,
+      "loss": 0.7347,
+      "step": 5645
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.014435111066657e-06,
+      "loss": 1.1564,
+      "step": 5646
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0139488113916175e-06,
+      "loss": 0.9506,
+      "step": 5647
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0134625687898193e-06,
+      "loss": 0.9628,
+      "step": 5648
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.012976383318359e-06,
+      "loss": 0.6731,
+      "step": 5649
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0124902550343253e-06,
+      "loss": 1.0014,
+      "step": 5650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0120041839948004e-06,
+      "loss": 1.2275,
+      "step": 5651
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.01151817025686e-06,
+      "loss": 1.0345,
+      "step": 5652
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0110322138775727e-06,
+      "loss": 0.8594,
+      "step": 5653
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.010546314914001e-06,
+      "loss": 1.031,
+      "step": 5654
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0100604734231996e-06,
+      "loss": 1.0155,
+      "step": 5655
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0095746894622174e-06,
+      "loss": 1.1846,
+      "step": 5656
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0090889630880968e-06,
+      "loss": 0.8112,
+      "step": 5657
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0086032943578723e-06,
+      "loss": 0.9647,
+      "step": 5658
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0081176833285728e-06,
+      "loss": 1.1277,
+      "step": 5659
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0076321300572192e-06,
+      "loss": 0.9114,
+      "step": 5660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0071466346008263e-06,
+      "loss": 1.1983,
+      "step": 5661
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0066611970164028e-06,
+      "loss": 0.8069,
+      "step": 5662
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0061758173609485e-06,
+      "loss": 0.8537,
+      "step": 5663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0056904956914596e-06,
+      "loss": 0.6401,
+      "step": 5664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0052052320649226e-06,
+      "loss": 1.2306,
+      "step": 5665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0047200265383186e-06,
+      "loss": 0.6256,
+      "step": 5666
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0042348791686212e-06,
+      "loss": 0.596,
+      "step": 5667
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0037497900127978e-06,
+      "loss": 0.555,
+      "step": 5668
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0032647591278088e-06,
+      "loss": 0.6682,
+      "step": 5669
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0027797865706072e-06,
+      "loss": 0.8264,
+      "step": 5670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.002294872398139e-06,
+      "loss": 0.7466,
+      "step": 5671
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.001810016667346e-06,
+      "loss": 1.0534,
+      "step": 5672
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.001325219435159e-06,
+      "loss": 0.9703,
+      "step": 5673
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0008404807585054e-06,
+      "loss": 0.9969,
+      "step": 5674
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0003558006943038e-06,
+      "loss": 0.9495,
+      "step": 5675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.99871179299466e-07,
+      "loss": 1.0127,
+      "step": 5676
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.993866166308982e-07,
+      "loss": 0.7911,
+      "step": 5677
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.989021127454973e-07,
+      "loss": 0.943,
+      "step": 5678
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.984176677001569e-07,
+      "loss": 0.9224,
+      "step": 5679
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.979332815517606e-07,
+      "loss": 0.9209,
+      "step": 5680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.97448954357186e-07,
+      "loss": 1.0427,
+      "step": 5681
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.969646861733044e-07,
+      "loss": 1.0678,
+      "step": 5682
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.964804770569794e-07,
+      "loss": 1.0003,
+      "step": 5683
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.959963270650678e-07,
+      "loss": 1.0226,
+      "step": 5684
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.955122362544194e-07,
+      "loss": 0.9695,
+      "step": 5685
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.95028204681878e-07,
+      "loss": 0.9178,
+      "step": 5686
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.945442324042792e-07,
+      "loss": 0.9137,
+      "step": 5687
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.940603194784522e-07,
+      "loss": 0.7362,
+      "step": 5688
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.935764659612195e-07,
+      "loss": 0.9433,
+      "step": 5689
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.930926719093956e-07,
+      "loss": 0.8519,
+      "step": 5690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.926089373797888e-07,
+      "loss": 0.8369,
+      "step": 5691
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.92125262429201e-07,
+      "loss": 1.0073,
+      "step": 5692
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.916416471144258e-07,
+      "loss": 1.0991,
+      "step": 5693
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.911580914922507e-07,
+      "loss": 0.7468,
+      "step": 5694
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.90674595619456e-07,
+      "loss": 0.8652,
+      "step": 5695
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.901911595528146e-07,
+      "loss": 0.9203,
+      "step": 5696
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.897077833490928e-07,
+      "loss": 1.0501,
+      "step": 5697
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.8922446706505e-07,
+      "loss": 1.0124,
+      "step": 5698
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.887412107574376e-07,
+      "loss": 1.122,
+      "step": 5699
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.882580144830007e-07,
+      "loss": 1.2016,
+      "step": 5700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.877748782984781e-07,
+      "loss": 0.9083,
+      "step": 5701
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.872918022606007e-07,
+      "loss": 0.9164,
+      "step": 5702
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.868087864260916e-07,
+      "loss": 0.8549,
+      "step": 5703
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.863258308516684e-07,
+      "loss": 1.106,
+      "step": 5704
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.858429355940405e-07,
+      "loss": 0.9267,
+      "step": 5705
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.853601007099106e-07,
+      "loss": 0.9877,
+      "step": 5706
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.848773262559742e-07,
+      "loss": 0.8867,
+      "step": 5707
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.843946122889191e-07,
+      "loss": 1.0883,
+      "step": 5708
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.839119588654282e-07,
+      "loss": 0.7517,
+      "step": 5709
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.83429366042175e-07,
+      "loss": 0.8648,
+      "step": 5710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.829468338758265e-07,
+      "loss": 4.9295,
+      "step": 5711
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.82464362423043e-07,
+      "loss": 0.7768,
+      "step": 5712
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.81981951740477e-07,
+      "loss": 1.229,
+      "step": 5713
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.814996018847748e-07,
+      "loss": 0.8507,
+      "step": 5714
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.810173129125744e-07,
+      "loss": 0.8083,
+      "step": 5715
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.805350848805079e-07,
+      "loss": 1.1288,
+      "step": 5716
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.800529178451993e-07,
+      "loss": 0.9093,
+      "step": 5717
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.795708118632657e-07,
+      "loss": 0.7488,
+      "step": 5718
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.790887669913172e-07,
+      "loss": 0.9771,
+      "step": 5719
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.786067832859568e-07,
+      "loss": 0.9837,
+      "step": 5720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.781248608037798e-07,
+      "loss": 0.6505,
+      "step": 5721
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.776429996013747e-07,
+      "loss": 0.9533,
+      "step": 5722
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.771611997353228e-07,
+      "loss": 1.1618,
+      "step": 5723
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.766794612621985e-07,
+      "loss": 0.8804,
+      "step": 5724
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.761977842385682e-07,
+      "loss": 0.8591,
+      "step": 5725
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.757161687209913e-07,
+      "loss": 1.016,
+      "step": 5726
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.752346147660208e-07,
+      "loss": 0.9102,
+      "step": 5727
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.747531224302014e-07,
+      "loss": 1.0704,
+      "step": 5728
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.742716917700714e-07,
+      "loss": 0.8974,
+      "step": 5729
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.737903228421605e-07,
+      "loss": 0.9141,
+      "step": 5730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.733090157029936e-07,
+      "loss": 1.4132,
+      "step": 5731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.728277704090862e-07,
+      "loss": 0.8187,
+      "step": 5732
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.723465870169473e-07,
+      "loss": 0.5881,
+      "step": 5733
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.718654655830784e-07,
+      "loss": 0.8466,
+      "step": 5734
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.713844061639738e-07,
+      "loss": 1.1821,
+      "step": 5735
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.70903408816121e-07,
+      "loss": 0.9295,
+      "step": 5736
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.704224735959988e-07,
+      "loss": 0.7644,
+      "step": 5737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.699416005600814e-07,
+      "loss": 0.9565,
+      "step": 5738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.69460789764833e-07,
+      "loss": 0.7779,
+      "step": 5739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.689800412667115e-07,
+      "loss": 0.9018,
+      "step": 5740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.684993551221676e-07,
+      "loss": 0.6847,
+      "step": 5741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.680187313876444e-07,
+      "loss": 0.9703,
+      "step": 5742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.675381701195782e-07,
+      "loss": 0.9022,
+      "step": 5743
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.670576713743974e-07,
+      "loss": 0.8756,
+      "step": 5744
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.665772352085229e-07,
+      "loss": 1.0295,
+      "step": 5745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.660968616783689e-07,
+      "loss": 0.9031,
+      "step": 5746
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.656165508403423e-07,
+      "loss": 0.8802,
+      "step": 5747
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.65136302750842e-07,
+      "loss": 0.7678,
+      "step": 5748
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.6465611746626e-07,
+      "loss": 1.0276,
+      "step": 5749
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.641759950429806e-07,
+      "loss": 1.1372,
+      "step": 5750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.636959355373807e-07,
+      "loss": 0.6135,
+      "step": 5751
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.6321593900583e-07,
+      "loss": 0.6944,
+      "step": 5752
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.627360055046914e-07,
+      "loss": 0.8763,
+      "step": 5753
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.62256135090319e-07,
+      "loss": 1.0627,
+      "step": 5754
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.617763278190608e-07,
+      "loss": 0.8739,
+      "step": 5755
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.61296583747257e-07,
+      "loss": 0.7879,
+      "step": 5756
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.608169029312395e-07,
+      "loss": 0.8761,
+      "step": 5757
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.60337285427334e-07,
+      "loss": 0.9822,
+      "step": 5758
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.598577312918585e-07,
+      "loss": 0.9296,
+      "step": 5759
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.593782405811221e-07,
+      "loss": 0.8456,
+      "step": 5760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.588988133514296e-07,
+      "loss": 0.5821,
+      "step": 5761
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.584194496590754e-07,
+      "loss": 1.0433,
+      "step": 5762
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.579401495603475e-07,
+      "loss": 0.8332,
+      "step": 5763
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.574609131115266e-07,
+      "loss": 1.1448,
+      "step": 5764
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.569817403688858e-07,
+      "loss": 0.9922,
+      "step": 5765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.565026313886903e-07,
+      "loss": 0.726,
+      "step": 5766
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.560235862271976e-07,
+      "loss": 0.848,
+      "step": 5767
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.555446049406602e-07,
+      "loss": 1.2689,
+      "step": 5768
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.550656875853196e-07,
+      "loss": 1.0744,
+      "step": 5769
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.54586834217412e-07,
+      "loss": 0.9555,
+      "step": 5770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.541080448931652e-07,
+      "loss": 0.9321,
+      "step": 5771
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.536293196687998e-07,
+      "loss": 1.0804,
+      "step": 5772
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.531506586005288e-07,
+      "loss": 0.7908,
+      "step": 5773
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.526720617445576e-07,
+      "loss": 0.8079,
+      "step": 5774
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.521935291570837e-07,
+      "loss": 1.3369,
+      "step": 5775
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.517150608942988e-07,
+      "loss": 1.1937,
+      "step": 5776
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.512366570123847e-07,
+      "loss": 1.1402,
+      "step": 5777
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.50758317567517e-07,
+      "loss": 0.9033,
+      "step": 5778
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.502800426158633e-07,
+      "loss": 0.8485,
+      "step": 5779
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.498018322135839e-07,
+      "loss": 1.2659,
+      "step": 5780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.493236864168313e-07,
+      "loss": 0.7082,
+      "step": 5781
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.488456052817499e-07,
+      "loss": 0.8108,
+      "step": 5782
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.483675888644781e-07,
+      "loss": 1.0428,
+      "step": 5783
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.47889637221145e-07,
+      "loss": 1.1762,
+      "step": 5784
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.47411750407873e-07,
+      "loss": 1.1719,
+      "step": 5785
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.469339284807767e-07,
+      "loss": 0.8256,
+      "step": 5786
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.464561714959628e-07,
+      "loss": 1.1009,
+      "step": 5787
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.459784795095308e-07,
+      "loss": 1.0169,
+      "step": 5788
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.455008525775725e-07,
+      "loss": 0.9554,
+      "step": 5789
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.45023290756171e-07,
+      "loss": 0.8044,
+      "step": 5790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.445457941014042e-07,
+      "loss": 1.1016,
+      "step": 5791
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.440683626693403e-07,
+      "loss": 0.7359,
+      "step": 5792
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.4359099651604e-07,
+      "loss": 1.2955,
+      "step": 5793
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.431136956975571e-07,
+      "loss": 0.7264,
+      "step": 5794
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.426364602699371e-07,
+      "loss": 0.823,
+      "step": 5795
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.421592902892184e-07,
+      "loss": 0.918,
+      "step": 5796
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.416821858114303e-07,
+      "loss": 1.0281,
+      "step": 5797
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.412051468925972e-07,
+      "loss": 0.773,
+      "step": 5798
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.407281735887333e-07,
+      "loss": 1.062,
+      "step": 5799
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.402512659558459e-07,
+      "loss": 0.9163,
+      "step": 5800
+    },
+    {
+      "epoch": 2.46,
+      "eval_loss": 1.254994511604309,
+      "eval_runtime": 32.2921,
+      "eval_samples_per_second": 38.554,
+      "eval_steps_per_second": 19.293,
+      "step": 5800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.397744240499344e-07,
+      "loss": 0.7872,
+      "step": 5801
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.392976479269908e-07,
+      "loss": 0.9198,
+      "step": 5802
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.388209376429994e-07,
+      "loss": 0.7928,
+      "step": 5803
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.383442932539359e-07,
+      "loss": 0.7124,
+      "step": 5804
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.3786771481577e-07,
+      "loss": 0.9956,
+      "step": 5805
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.373912023844623e-07,
+      "loss": 0.661,
+      "step": 5806
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.369147560159659e-07,
+      "loss": 1.1512,
+      "step": 5807
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.364383757662262e-07,
+      "loss": 0.9678,
+      "step": 5808
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.359620616911807e-07,
+      "loss": 0.8289,
+      "step": 5809
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.354858138467596e-07,
+      "loss": 0.7766,
+      "step": 5810
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.350096322888848e-07,
+      "loss": 1.0407,
+      "step": 5811
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.345335170734703e-07,
+      "loss": 0.7063,
+      "step": 5812
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.340574682564233e-07,
+      "loss": 0.7756,
+      "step": 5813
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.335814858936424e-07,
+      "loss": 0.8492,
+      "step": 5814
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.331055700410183e-07,
+      "loss": 1.204,
+      "step": 5815
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.326297207544345e-07,
+      "loss": 1.0995,
+      "step": 5816
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.321539380897652e-07,
+      "loss": 0.6479,
+      "step": 5817
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.316782221028793e-07,
+      "loss": 0.9467,
+      "step": 5818
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.312025728496355e-07,
+      "loss": 0.9558,
+      "step": 5819
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.307269903858864e-07,
+      "loss": 0.9443,
+      "step": 5820
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.302514747674754e-07,
+      "loss": 0.7586,
+      "step": 5821
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.297760260502389e-07,
+      "loss": 0.7004,
+      "step": 5822
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.293006442900053e-07,
+      "loss": 0.9509,
+      "step": 5823
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.288253295425944e-07,
+      "loss": 0.9799,
+      "step": 5824
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.283500818638194e-07,
+      "loss": 0.5699,
+      "step": 5825
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.278749013094846e-07,
+      "loss": 0.7408,
+      "step": 5826
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.273997879353863e-07,
+      "loss": 1.0885,
+      "step": 5827
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.269247417973148e-07,
+      "loss": 0.7065,
+      "step": 5828
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.264497629510501e-07,
+      "loss": 0.681,
+      "step": 5829
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.259748514523654e-07,
+      "loss": 1.0489,
+      "step": 5830
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.255000073570262e-07,
+      "loss": 0.8951,
+      "step": 5831
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.250252307207893e-07,
+      "loss": 0.893,
+      "step": 5832
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.245505215994048e-07,
+      "loss": 1.0024,
+      "step": 5833
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.240758800486126e-07,
+      "loss": 1.0111,
+      "step": 5834
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.23601306124148e-07,
+      "loss": 1.1438,
+      "step": 5835
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.231267998817361e-07,
+      "loss": 0.7422,
+      "step": 5836
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.226523613770938e-07,
+      "loss": 1.1649,
+      "step": 5837
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.221779906659314e-07,
+      "loss": 0.8297,
+      "step": 5838
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.217036878039503e-07,
+      "loss": 0.8245,
+      "step": 5839
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.212294528468445e-07,
+      "loss": 0.7356,
+      "step": 5840
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.207552858502994e-07,
+      "loss": 0.9133,
+      "step": 5841
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.202811868699925e-07,
+      "loss": 1.0049,
+      "step": 5842
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.198071559615942e-07,
+      "loss": 0.9889,
+      "step": 5843
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.19333193180766e-07,
+      "loss": 0.9936,
+      "step": 5844
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.188592985831622e-07,
+      "loss": 0.9316,
+      "step": 5845
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.18385472224428e-07,
+      "loss": 0.757,
+      "step": 5846
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.179117141602013e-07,
+      "loss": 1.2357,
+      "step": 5847
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.174380244461122e-07,
+      "loss": 0.846,
+      "step": 5848
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.169644031377818e-07,
+      "loss": 1.1591,
+      "step": 5849
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.164908502908242e-07,
+      "loss": 0.8099,
+      "step": 5850
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.16017365960845e-07,
+      "loss": 0.7981,
+      "step": 5851
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.155439502034421e-07,
+      "loss": 0.733,
+      "step": 5852
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.150706030742046e-07,
+      "loss": 0.9223,
+      "step": 5853
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.14597324628714e-07,
+      "loss": 1.0449,
+      "step": 5854
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.141241149225441e-07,
+      "loss": 0.6145,
+      "step": 5855
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.1365097401126e-07,
+      "loss": 0.9283,
+      "step": 5856
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.131779019504184e-07,
+      "loss": 0.8545,
+      "step": 5857
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.127048987955698e-07,
+      "loss": 1.0627,
+      "step": 5858
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.122319646022548e-07,
+      "loss": 1.14,
+      "step": 5859
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.117590994260062e-07,
+      "loss": 0.7992,
+      "step": 5860
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.112863033223489e-07,
+      "loss": 1.1175,
+      "step": 5861
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.108135763467999e-07,
+      "loss": 0.8232,
+      "step": 5862
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.103409185548678e-07,
+      "loss": 0.9881,
+      "step": 5863
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.098683300020524e-07,
+      "loss": 1.055,
+      "step": 5864
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.093958107438475e-07,
+      "loss": 0.9958,
+      "step": 5865
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.08923360835737e-07,
+      "loss": 0.9394,
+      "step": 5866
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.084509803331965e-07,
+      "loss": 1.0047,
+      "step": 5867
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.079786692916946e-07,
+      "loss": 1.0145,
+      "step": 5868
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.075064277666906e-07,
+      "loss": 0.9273,
+      "step": 5869
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.070342558136364e-07,
+      "loss": 1.0051,
+      "step": 5870
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.065621534879751e-07,
+      "loss": 0.6146,
+      "step": 5871
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.060901208451429e-07,
+      "loss": 0.9508,
+      "step": 5872
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.056181579405666e-07,
+      "loss": 0.5159,
+      "step": 5873
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.051462648296652e-07,
+      "loss": 0.6363,
+      "step": 5874
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.046744415678493e-07,
+      "loss": 0.6207,
+      "step": 5875
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.042026882105215e-07,
+      "loss": 0.9515,
+      "step": 5876
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.037310048130764e-07,
+      "loss": 0.8991,
+      "step": 5877
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.032593914308994e-07,
+      "loss": 0.7615,
+      "step": 5878
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.027878481193691e-07,
+      "loss": 1.0612,
+      "step": 5879
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.023163749338551e-07,
+      "loss": 0.8829,
+      "step": 5880
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.018449719297188e-07,
+      "loss": 1.2153,
+      "step": 5881
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.013736391623136e-07,
+      "loss": 0.4863,
+      "step": 5882
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.009023766869842e-07,
+      "loss": 0.8512,
+      "step": 5883
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.004311845590673e-07,
+      "loss": 1.1605,
+      "step": 5884
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.999600628338917e-07,
+      "loss": 0.9494,
+      "step": 5885
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.994890115667765e-07,
+      "loss": 0.8229,
+      "step": 5886
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.990180308130351e-07,
+      "loss": 0.7966,
+      "step": 5887
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.985471206279705e-07,
+      "loss": 1.5166,
+      "step": 5888
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.980762810668781e-07,
+      "loss": 1.305,
+      "step": 5889
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.97605512185045e-07,
+      "loss": 1.0149,
+      "step": 5890
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.971348140377498e-07,
+      "loss": 0.965,
+      "step": 5891
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.966641866802631e-07,
+      "loss": 1.2144,
+      "step": 5892
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.961936301678469e-07,
+      "loss": 1.1267,
+      "step": 5893
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.957231445557543e-07,
+      "loss": 0.9579,
+      "step": 5894
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.952527298992324e-07,
+      "loss": 0.8805,
+      "step": 5895
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.947823862535174e-07,
+      "loss": 1.2994,
+      "step": 5896
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.943121136738383e-07,
+      "loss": 0.8814,
+      "step": 5897
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.938419122154156e-07,
+      "loss": 0.7733,
+      "step": 5898
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.933717819334615e-07,
+      "loss": 1.1495,
+      "step": 5899
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.929017228831794e-07,
+      "loss": 0.898,
+      "step": 5900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.924317351197646e-07,
+      "loss": 0.8665,
+      "step": 5901
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.91961818698405e-07,
+      "loss": 0.8968,
+      "step": 5902
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.914919736742787e-07,
+      "loss": 0.8135,
+      "step": 5903
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.910222001025561e-07,
+      "loss": 1.1318,
+      "step": 5904
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.90552498038399e-07,
+      "loss": 0.7661,
+      "step": 5905
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.90082867536961e-07,
+      "loss": 0.7421,
+      "step": 5906
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.896133086533869e-07,
+      "loss": 0.8446,
+      "step": 5907
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.891438214428138e-07,
+      "loss": 0.4413,
+      "step": 5908
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.886744059603694e-07,
+      "loss": 0.8975,
+      "step": 5909
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.88205062261174e-07,
+      "loss": 0.9425,
+      "step": 5910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.877357904003392e-07,
+      "loss": 1.2894,
+      "step": 5911
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.872665904329675e-07,
+      "loss": 0.8432,
+      "step": 5912
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.867974624141535e-07,
+      "loss": 0.8542,
+      "step": 5913
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.863284063989838e-07,
+      "loss": 1.1062,
+      "step": 5914
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.858594224425347e-07,
+      "loss": 1.1742,
+      "step": 5915
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.853905105998768e-07,
+      "loss": 0.7796,
+      "step": 5916
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.849216709260702e-07,
+      "loss": 1.0989,
+      "step": 5917
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.844529034761674e-07,
+      "loss": 0.8862,
+      "step": 5918
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.839842083052119e-07,
+      "loss": 1.0909,
+      "step": 5919
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.83515585468239e-07,
+      "loss": 1.0974,
+      "step": 5920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.830470350202756e-07,
+      "loss": 0.8694,
+      "step": 5921
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.825785570163398e-07,
+      "loss": 1.0705,
+      "step": 5922
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.821101515114412e-07,
+      "loss": 1.3108,
+      "step": 5923
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.816418185605809e-07,
+      "loss": 0.7485,
+      "step": 5924
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.811735582187525e-07,
+      "loss": 0.6066,
+      "step": 5925
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.807053705409396e-07,
+      "loss": 0.7792,
+      "step": 5926
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.80237255582118e-07,
+      "loss": 1.0073,
+      "step": 5927
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.797692133972547e-07,
+      "loss": 0.6968,
+      "step": 5928
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.793012440413084e-07,
+      "loss": 0.9631,
+      "step": 5929
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.788333475692291e-07,
+      "loss": 1.3349,
+      "step": 5930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.783655240359576e-07,
+      "loss": 0.8017,
+      "step": 5931
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.778977734964281e-07,
+      "loss": 0.69,
+      "step": 5932
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.774300960055643e-07,
+      "loss": 0.9857,
+      "step": 5933
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.769624916182821e-07,
+      "loss": 1.0378,
+      "step": 5934
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.764949603894885e-07,
+      "loss": 0.8924,
+      "step": 5935
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.760275023740819e-07,
+      "loss": 1.0818,
+      "step": 5936
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.755601176269528e-07,
+      "loss": 1.0862,
+      "step": 5937
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.750928062029819e-07,
+      "loss": 0.9234,
+      "step": 5938
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.746255681570422e-07,
+      "loss": 0.4702,
+      "step": 5939
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.741584035439987e-07,
+      "loss": 0.9426,
+      "step": 5940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.736913124187057e-07,
+      "loss": 1.1552,
+      "step": 5941
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.732242948360115e-07,
+      "loss": 0.9139,
+      "step": 5942
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.727573508507526e-07,
+      "loss": 0.6156,
+      "step": 5943
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.722904805177604e-07,
+      "loss": 0.7885,
+      "step": 5944
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.718236838918543e-07,
+      "loss": 1.0192,
+      "step": 5945
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.713569610278477e-07,
+      "loss": 0.8147,
+      "step": 5946
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.708903119805437e-07,
+      "loss": 1.1004,
+      "step": 5947
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.704237368047381e-07,
+      "loss": 0.9874,
+      "step": 5948
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.699572355552162e-07,
+      "loss": 0.881,
+      "step": 5949
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.694908082867567e-07,
+      "loss": 0.974,
+      "step": 5950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.690244550541275e-07,
+      "loss": 1.0425,
+      "step": 5951
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.685581759120898e-07,
+      "loss": 0.9515,
+      "step": 5952
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.680919709153942e-07,
+      "loss": 0.9601,
+      "step": 5953
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.67625840118784e-07,
+      "loss": 0.8799,
+      "step": 5954
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.67159783576994e-07,
+      "loss": 0.9174,
+      "step": 5955
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.666938013447484e-07,
+      "loss": 1.0619,
+      "step": 5956
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.662278934767646e-07,
+      "loss": 0.9349,
+      "step": 5957
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.657620600277511e-07,
+      "loss": 1.2058,
+      "step": 5958
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.652963010524057e-07,
+      "loss": 0.8883,
+      "step": 5959
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.648306166054205e-07,
+      "loss": 1.0495,
+      "step": 5960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.643650067414759e-07,
+      "loss": 1.4784,
+      "step": 5961
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.638994715152452e-07,
+      "loss": 1.0604,
+      "step": 5962
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.634340109813935e-07,
+      "loss": 0.7977,
+      "step": 5963
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.629686251945752e-07,
+      "loss": 0.8536,
+      "step": 5964
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.625033142094376e-07,
+      "loss": 1.1721,
+      "step": 5965
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.62038078080618e-07,
+      "loss": 0.8995,
+      "step": 5966
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.615729168627462e-07,
+      "loss": 0.6846,
+      "step": 5967
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.611078306104409e-07,
+      "loss": 0.945,
+      "step": 5968
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.606428193783161e-07,
+      "loss": 0.9723,
+      "step": 5969
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.601778832209723e-07,
+      "loss": 1.0181,
+      "step": 5970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.597130221930052e-07,
+      "loss": 1.0815,
+      "step": 5971
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.59248236348998e-07,
+      "loss": 0.9428,
+      "step": 5972
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.587835257435285e-07,
+      "loss": 1.0334,
+      "step": 5973
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.583188904311628e-07,
+      "loss": 0.9926,
+      "step": 5974
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.578543304664605e-07,
+      "loss": 0.9538,
+      "step": 5975
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.573898459039698e-07,
+      "loss": 0.7467,
+      "step": 5976
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.569254367982336e-07,
+      "loss": 1.1127,
+      "step": 5977
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.564611032037824e-07,
+      "loss": 0.8656,
+      "step": 5978
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.5599684517514e-07,
+      "loss": 0.7794,
+      "step": 5979
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.5553266276682e-07,
+      "loss": 0.7362,
+      "step": 5980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.550685560333287e-07,
+      "loss": 0.7002,
+      "step": 5981
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.546045250291614e-07,
+      "loss": 0.9441,
+      "step": 5982
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.541405698088062e-07,
+      "loss": 0.7756,
+      "step": 5983
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.53676690426742e-07,
+      "loss": 1.1561,
+      "step": 5984
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.532128869374392e-07,
+      "loss": 0.7359,
+      "step": 5985
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.527491593953571e-07,
+      "loss": 0.9637,
+      "step": 5986
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.52285507854949e-07,
+      "loss": 1.2135,
+      "step": 5987
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.51821932370657e-07,
+      "loss": 0.6531,
+      "step": 5988
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.513584329969162e-07,
+      "loss": 0.9435,
+      "step": 5989
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.508950097881504e-07,
+      "loss": 1.0946,
+      "step": 5990
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.504316627987764e-07,
+      "loss": 0.7415,
+      "step": 5991
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.499683920832024e-07,
+      "loss": 0.8021,
+      "step": 5992
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.495051976958255e-07,
+      "loss": 0.9517,
+      "step": 5993
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.490420796910357e-07,
+      "loss": 1.2423,
+      "step": 5994
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.485790381232128e-07,
+      "loss": 1.043,
+      "step": 5995
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.481160730467292e-07,
+      "loss": 1.1528,
+      "step": 5996
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.476531845159458e-07,
+      "loss": 0.9018,
+      "step": 5997
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.471903725852169e-07,
+      "loss": 0.8304,
+      "step": 5998
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.467276373088872e-07,
+      "loss": 0.9868,
+      "step": 5999
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.462649787412922e-07,
+      "loss": 1.1099,
+      "step": 6000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.458023969367576e-07,
+      "loss": 0.8922,
+      "step": 6001
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.453398919496015e-07,
+      "loss": 0.7979,
+      "step": 6002
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.448774638341314e-07,
+      "loss": 1.1917,
+      "step": 6003
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.444151126446482e-07,
+      "loss": 1.0601,
+      "step": 6004
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.439528384354398e-07,
+      "loss": 0.8784,
+      "step": 6005
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.434906412607904e-07,
+      "loss": 0.8427,
+      "step": 6006
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.4302852117497e-07,
+      "loss": 1.156,
+      "step": 6007
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.425664782322433e-07,
+      "loss": 1.1157,
+      "step": 6008
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.421045124868633e-07,
+      "loss": 0.82,
+      "step": 6009
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.41642623993076e-07,
+      "loss": 0.9313,
+      "step": 6010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.411808128051164e-07,
+      "loss": 1.5779,
+      "step": 6011
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.407190789772126e-07,
+      "loss": 1.1395,
+      "step": 6012
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.402574225635805e-07,
+      "loss": 0.6281,
+      "step": 6013
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.397958436184315e-07,
+      "loss": 0.6887,
+      "step": 6014
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.393343421959632e-07,
+      "loss": 1.009,
+      "step": 6015
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.388729183503676e-07,
+      "loss": 0.9951,
+      "step": 6016
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.384115721358247e-07,
+      "loss": 0.7307,
+      "step": 6017
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.379503036065083e-07,
+      "loss": 0.6663,
+      "step": 6018
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.374891128165799e-07,
+      "loss": 0.8795,
+      "step": 6019
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.370279998201948e-07,
+      "loss": 1.1275,
+      "step": 6020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.365669646714984e-07,
+      "loss": 0.9172,
+      "step": 6021
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.361060074246249e-07,
+      "loss": 0.84,
+      "step": 6022
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.356451281337026e-07,
+      "loss": 1.0383,
+      "step": 6023
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.351843268528475e-07,
+      "loss": 0.7674,
+      "step": 6024
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.347236036361693e-07,
+      "loss": 0.7571,
+      "step": 6025
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.342629585377659e-07,
+      "loss": 0.689,
+      "step": 6026
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.338023916117278e-07,
+      "loss": 0.971,
+      "step": 6027
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.333419029121363e-07,
+      "loss": 0.9754,
+      "step": 6028
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.328814924930628e-07,
+      "loss": 1.174,
+      "step": 6029
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.324211604085697e-07,
+      "loss": 0.9549,
+      "step": 6030
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.319609067127102e-07,
+      "loss": 0.9645,
+      "step": 6031
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.31500731459528e-07,
+      "loss": 0.7902,
+      "step": 6032
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.310406347030588e-07,
+      "loss": 1.2499,
+      "step": 6033
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.305806164973271e-07,
+      "loss": 1.0968,
+      "step": 6034
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.301206768963496e-07,
+      "loss": 0.9176,
+      "step": 6035
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.29660815954134e-07,
+      "loss": 0.903,
+      "step": 6036
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.292010337246785e-07,
+      "loss": 1.259,
+      "step": 6037
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.287413302619706e-07,
+      "loss": 0.8005,
+      "step": 6038
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.282817056199908e-07,
+      "loss": 0.9607,
+      "step": 6039
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.278221598527086e-07,
+      "loss": 1.1205,
+      "step": 6040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.273626930140853e-07,
+      "loss": 0.8166,
+      "step": 6041
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.269033051580721e-07,
+      "loss": 0.78,
+      "step": 6042
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.264439963386116e-07,
+      "loss": 0.6601,
+      "step": 6043
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.259847666096368e-07,
+      "loss": 1.139,
+      "step": 6044
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.255256160250725e-07,
+      "loss": 0.9817,
+      "step": 6045
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.250665446388315e-07,
+      "loss": 0.97,
+      "step": 6046
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.246075525048207e-07,
+      "loss": 1.127,
+      "step": 6047
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.241486396769343e-07,
+      "loss": 0.9558,
+      "step": 6048
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.236898062090607e-07,
+      "loss": 1.1192,
+      "step": 6049
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.232310521550755e-07,
+      "loss": 1.1713,
+      "step": 6050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.227723775688475e-07,
+      "loss": 0.9249,
+      "step": 6051
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.223137825042356e-07,
+      "loss": 1.2035,
+      "step": 6052
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.218552670150884e-07,
+      "loss": 0.8998,
+      "step": 6053
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.213968311552459e-07,
+      "loss": 1.0464,
+      "step": 6054
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.209384749785396e-07,
+      "loss": 0.8578,
+      "step": 6055
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.204801985387895e-07,
+      "loss": 0.7735,
+      "step": 6056
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.200220018898081e-07,
+      "loss": 0.9546,
+      "step": 6057
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.195638850853981e-07,
+      "loss": 0.7995,
+      "step": 6058
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.191058481793522e-07,
+      "loss": 0.6203,
+      "step": 6059
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.186478912254545e-07,
+      "loss": 0.6662,
+      "step": 6060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.181900142774786e-07,
+      "loss": 0.9896,
+      "step": 6061
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.177322173891907e-07,
+      "loss": 1.1268,
+      "step": 6062
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.172745006143451e-07,
+      "loss": 0.8108,
+      "step": 6063
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.168168640066891e-07,
+      "loss": 0.7803,
+      "step": 6064
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.163593076199575e-07,
+      "loss": 0.8598,
+      "step": 6065
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.159018315078803e-07,
+      "loss": 1.0163,
+      "step": 6066
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.154444357241734e-07,
+      "loss": 0.9762,
+      "step": 6067
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.149871203225464e-07,
+      "loss": 1.0154,
+      "step": 6068
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.145298853566973e-07,
+      "loss": 1.2579,
+      "step": 6069
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.140727308803169e-07,
+      "loss": 1.2139,
+      "step": 6070
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.136156569470838e-07,
+      "loss": 0.9935,
+      "step": 6071
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.131586636106698e-07,
+      "loss": 0.9222,
+      "step": 6072
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.127017509247357e-07,
+      "loss": 0.8346,
+      "step": 6073
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.122449189429337e-07,
+      "loss": 0.85,
+      "step": 6074
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.117881677189053e-07,
+      "loss": 0.5385,
+      "step": 6075
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.113314973062841e-07,
+      "loss": 0.9457,
+      "step": 6076
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.108749077586926e-07,
+      "loss": 1.0242,
+      "step": 6077
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.104183991297452e-07,
+      "loss": 1.0322,
+      "step": 6078
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.099619714730454e-07,
+      "loss": 1.1153,
+      "step": 6079
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.095056248421887e-07,
+      "loss": 0.654,
+      "step": 6080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.090493592907605e-07,
+      "loss": 1.4159,
+      "step": 6081
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.085931748723358e-07,
+      "loss": 0.9648,
+      "step": 6082
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.08137071640481e-07,
+      "loss": 0.7654,
+      "step": 6083
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.076810496487537e-07,
+      "loss": 0.7458,
+      "step": 6084
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.072251089507e-07,
+      "loss": 1.0148,
+      "step": 6085
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.067692495998582e-07,
+      "loss": 1.0618,
+      "step": 6086
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.063134716497553e-07,
+      "loss": 0.8812,
+      "step": 6087
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.058577751539108e-07,
+      "loss": 1.3118,
+      "step": 6088
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.054021601658336e-07,
+      "loss": 0.7818,
+      "step": 6089
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.049466267390224e-07,
+      "loss": 0.8063,
+      "step": 6090
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.044911749269677e-07,
+      "loss": 0.5875,
+      "step": 6091
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.040358047831489e-07,
+      "loss": 0.805,
+      "step": 6092
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.035805163610376e-07,
+      "loss": 0.6556,
+      "step": 6093
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.031253097140937e-07,
+      "loss": 1.0155,
+      "step": 6094
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.02670184895769e-07,
+      "loss": 1.0623,
+      "step": 6095
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.022151419595055e-07,
+      "loss": 1.074,
+      "step": 6096
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.01760180958736e-07,
+      "loss": 1.0997,
+      "step": 6097
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.013053019468817e-07,
+      "loss": 0.9029,
+      "step": 6098
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.008505049773569e-07,
+      "loss": 1.0979,
+      "step": 6099
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.003957901035635e-07,
+      "loss": 1.2684,
+      "step": 6100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.999411573788965e-07,
+      "loss": 0.8139,
+      "step": 6101
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.994866068567381e-07,
+      "loss": 0.8254,
+      "step": 6102
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.990321385904652e-07,
+      "loss": 0.9024,
+      "step": 6103
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.985777526334405e-07,
+      "loss": 0.8529,
+      "step": 6104
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.981234490390202e-07,
+      "loss": 0.7159,
+      "step": 6105
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.976692278605487e-07,
+      "loss": 0.6895,
+      "step": 6106
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.972150891513625e-07,
+      "loss": 0.8559,
+      "step": 6107
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.96761032964787e-07,
+      "loss": 0.9732,
+      "step": 6108
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.963070593541392e-07,
+      "loss": 1.0929,
+      "step": 6109
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.958531683727242e-07,
+      "loss": 1.1632,
+      "step": 6110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.95399360073841e-07,
+      "loss": 0.7755,
+      "step": 6111
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.949456345107753e-07,
+      "loss": 0.7484,
+      "step": 6112
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.944919917368057e-07,
+      "loss": 0.5925,
+      "step": 6113
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.940384318051987e-07,
+      "loss": 0.8435,
+      "step": 6114
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.935849547692139e-07,
+      "loss": 0.819,
+      "step": 6115
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.931315606820981e-07,
+      "loss": 0.7938,
+      "step": 6116
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.926782495970903e-07,
+      "loss": 0.8905,
+      "step": 6117
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.9222502156742e-07,
+      "loss": 0.9742,
+      "step": 6118
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.917718766463055e-07,
+      "loss": 0.7419,
+      "step": 6119
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.913188148869569e-07,
+      "loss": 0.9065,
+      "step": 6120
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.908658363425728e-07,
+      "loss": 0.7579,
+      "step": 6121
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.904129410663438e-07,
+      "loss": 0.8065,
+      "step": 6122
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.899601291114493e-07,
+      "loss": 0.8377,
+      "step": 6123
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.895074005310593e-07,
+      "loss": 0.8431,
+      "step": 6124
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.890547553783349e-07,
+      "loss": 1.0297,
+      "step": 6125
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.886021937064271e-07,
+      "loss": 0.5958,
+      "step": 6126
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.881497155684755e-07,
+      "loss": 1.0966,
+      "step": 6127
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.876973210176123e-07,
+      "loss": 0.8661,
+      "step": 6128
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.872450101069578e-07,
+      "loss": 0.8658,
+      "step": 6129
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.86792782889624e-07,
+      "loss": 0.8498,
+      "step": 6130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.863406394187119e-07,
+      "loss": 1.0911,
+      "step": 6131
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.858885797473132e-07,
+      "loss": 1.2841,
+      "step": 6132
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.854366039285104e-07,
+      "loss": 0.6827,
+      "step": 6133
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.849847120153756e-07,
+      "loss": 0.7304,
+      "step": 6134
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.845329040609701e-07,
+      "loss": 0.9298,
+      "step": 6135
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.840811801183472e-07,
+      "loss": 0.6544,
+      "step": 6136
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.836295402405484e-07,
+      "loss": 1.0409,
+      "step": 6137
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.831779844806074e-07,
+      "loss": 0.7958,
+      "step": 6138
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.827265128915449e-07,
+      "loss": 1.1603,
+      "step": 6139
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.822751255263765e-07,
+      "loss": 0.9354,
+      "step": 6140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.81823822438103e-07,
+      "loss": 0.8882,
+      "step": 6141
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.813726036797188e-07,
+      "loss": 1.0379,
+      "step": 6142
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.80921469304206e-07,
+      "loss": 0.5655,
+      "step": 6143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.804704193645386e-07,
+      "loss": 1.1281,
+      "step": 6144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.800194539136792e-07,
+      "loss": 0.9646,
+      "step": 6145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.795685730045821e-07,
+      "loss": 0.757,
+      "step": 6146
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.791177766901895e-07,
+      "loss": 1.0194,
+      "step": 6147
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.786670650234357e-07,
+      "loss": 0.9646,
+      "step": 6148
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.78216438057245e-07,
+      "loss": 1.0538,
+      "step": 6149
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.777658958445296e-07,
+      "loss": 1.1038,
+      "step": 6150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.773154384381943e-07,
+      "loss": 1.1907,
+      "step": 6151
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.768650658911321e-07,
+      "loss": 0.9753,
+      "step": 6152
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.764147782562272e-07,
+      "loss": 0.8311,
+      "step": 6153
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.75964575586353e-07,
+      "loss": 1.1428,
+      "step": 6154
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.755144579343744e-07,
+      "loss": 0.9182,
+      "step": 6155
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.750644253531439e-07,
+      "loss": 0.7168,
+      "step": 6156
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.746144778955066e-07,
+      "loss": 0.7318,
+      "step": 6157
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.74164615614295e-07,
+      "loss": 0.8546,
+      "step": 6158
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.737148385623343e-07,
+      "loss": 0.8163,
+      "step": 6159
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.732651467924371e-07,
+      "loss": 0.9282,
+      "step": 6160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.728155403574084e-07,
+      "loss": 1.061,
+      "step": 6161
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.723660193100404e-07,
+      "loss": 0.8776,
+      "step": 6162
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.71916583703119e-07,
+      "loss": 0.8504,
+      "step": 6163
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.714672335894165e-07,
+      "loss": 0.9459,
+      "step": 6164
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.710179690216972e-07,
+      "loss": 1.0773,
+      "step": 6165
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.705687900527145e-07,
+      "loss": 0.9314,
+      "step": 6166
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.701196967352124e-07,
+      "loss": 0.7457,
+      "step": 6167
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.696706891219238e-07,
+      "loss": 0.8716,
+      "step": 6168
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.692217672655726e-07,
+      "loss": 0.7645,
+      "step": 6169
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.687729312188721e-07,
+      "loss": 0.9888,
+      "step": 6170
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.683241810345261e-07,
+      "loss": 0.8454,
+      "step": 6171
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.678755167652271e-07,
+      "loss": 0.6774,
+      "step": 6172
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.674269384636595e-07,
+      "loss": 0.6327,
+      "step": 6173
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.66978446182495e-07,
+      "loss": 0.977,
+      "step": 6174
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.665300399743977e-07,
+      "loss": 1.1175,
+      "step": 6175
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.660817198920194e-07,
+      "loss": 1.0265,
+      "step": 6176
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.656334859880034e-07,
+      "loss": 1.1755,
+      "step": 6177
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.651853383149832e-07,
+      "loss": 0.967,
+      "step": 6178
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.647372769255797e-07,
+      "loss": 1.2477,
+      "step": 6179
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.642893018724063e-07,
+      "loss": 0.5992,
+      "step": 6180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.638414132080657e-07,
+      "loss": 0.9308,
+      "step": 6181
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.633936109851485e-07,
+      "loss": 0.7956,
+      "step": 6182
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.629458952562383e-07,
+      "loss": 0.8906,
+      "step": 6183
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.624982660739056e-07,
+      "loss": 0.8844,
+      "step": 6184
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.620507234907127e-07,
+      "loss": 0.7549,
+      "step": 6185
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.616032675592113e-07,
+      "loss": 0.8041,
+      "step": 6186
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.61155898331942e-07,
+      "loss": 1.0564,
+      "step": 6187
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.607086158614368e-07,
+      "loss": 0.8075,
+      "step": 6188
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.602614202002154e-07,
+      "loss": 0.8859,
+      "step": 6189
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.598143114007901e-07,
+      "loss": 1.1408,
+      "step": 6190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.593672895156592e-07,
+      "loss": 0.6577,
+      "step": 6191
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.589203545973157e-07,
+      "loss": 0.8587,
+      "step": 6192
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.584735066982377e-07,
+      "loss": 0.9754,
+      "step": 6193
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.580267458708965e-07,
+      "loss": 0.8323,
+      "step": 6194
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.575800721677505e-07,
+      "loss": 1.0972,
+      "step": 6195
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.571334856412504e-07,
+      "loss": 0.9797,
+      "step": 6196
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.566869863438343e-07,
+      "loss": 0.7978,
+      "step": 6197
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.562405743279322e-07,
+      "loss": 0.9029,
+      "step": 6198
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.557942496459609e-07,
+      "loss": 1.0203,
+      "step": 6199
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.553480123503315e-07,
+      "loss": 0.8432,
+      "step": 6200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.549018624934405e-07,
+      "loss": 0.9743,
+      "step": 6201
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.544558001276767e-07,
+      "loss": 1.0888,
+      "step": 6202
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.540098253054169e-07,
+      "loss": 0.9389,
+      "step": 6203
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.535639380790293e-07,
+      "loss": 1.1883,
+      "step": 6204
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.531181385008699e-07,
+      "loss": 0.8739,
+      "step": 6205
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.526724266232865e-07,
+      "loss": 0.8931,
+      "step": 6206
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.522268024986153e-07,
+      "loss": 1.2542,
+      "step": 6207
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.51781266179183e-07,
+      "loss": 1.158,
+      "step": 6208
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.513358177173047e-07,
+      "loss": 4.2794,
+      "step": 6209
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.508904571652865e-07,
+      "loss": 0.9358,
+      "step": 6210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.504451845754232e-07,
+      "loss": 0.8334,
+      "step": 6211
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.500000000000003e-07,
+      "loss": 0.6875,
+      "step": 6212
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.495549034912916e-07,
+      "loss": 1.2984,
+      "step": 6213
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.49109895101562e-07,
+      "loss": 1.1817,
+      "step": 6214
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.486649748830654e-07,
+      "loss": 0.8289,
+      "step": 6215
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.482201428880449e-07,
+      "loss": 0.9044,
+      "step": 6216
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.477753991687344e-07,
+      "loss": 1.0027,
+      "step": 6217
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.473307437773559e-07,
+      "loss": 0.9,
+      "step": 6218
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.468861767661226e-07,
+      "loss": 1.1856,
+      "step": 6219
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.464416981872357e-07,
+      "loss": 0.8357,
+      "step": 6220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.459973080928874e-07,
+      "loss": 0.9464,
+      "step": 6221
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.455530065352591e-07,
+      "loss": 0.7796,
+      "step": 6222
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.451087935665222e-07,
+      "loss": 0.8165,
+      "step": 6223
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.446646692388359e-07,
+      "loss": 0.9623,
+      "step": 6224
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.442206336043517e-07,
+      "loss": 0.9702,
+      "step": 6225
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.437766867152078e-07,
+      "loss": 0.7365,
+      "step": 6226
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.433328286235351e-07,
+      "loss": 0.9386,
+      "step": 6227
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.428890593814507e-07,
+      "loss": 0.7303,
+      "step": 6228
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.42445379041064e-07,
+      "loss": 1.393,
+      "step": 6229
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.42001787654473e-07,
+      "loss": 1.371,
+      "step": 6230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.415582852737655e-07,
+      "loss": 0.7301,
+      "step": 6231
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.411148719510175e-07,
+      "loss": 0.8459,
+      "step": 6232
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.406715477382969e-07,
+      "loss": 1.2881,
+      "step": 6233
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.402283126876584e-07,
+      "loss": 1.5249,
+      "step": 6234
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.397851668511491e-07,
+      "loss": 0.8254,
+      "step": 6235
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.393421102808024e-07,
+      "loss": 1.2661,
+      "step": 6236
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.388991430286454e-07,
+      "loss": 1.0469,
+      "step": 6237
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.384562651466904e-07,
+      "loss": 0.9926,
+      "step": 6238
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.380134766869423e-07,
+      "loss": 0.7632,
+      "step": 6239
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.375707777013933e-07,
+      "loss": 1.2331,
+      "step": 6240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.371281682420273e-07,
+      "loss": 0.816,
+      "step": 6241
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.366856483608154e-07,
+      "loss": 0.6584,
+      "step": 6242
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.362432181097202e-07,
+      "loss": 0.8715,
+      "step": 6243
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.358008775406921e-07,
+      "loss": 0.5685,
+      "step": 6244
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.353586267056718e-07,
+      "loss": 1.0399,
+      "step": 6245
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.349164656565904e-07,
+      "loss": 0.9429,
+      "step": 6246
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.344743944453661e-07,
+      "loss": 1.0142,
+      "step": 6247
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.340324131239092e-07,
+      "loss": 0.8532,
+      "step": 6248
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.33590521744117e-07,
+      "loss": 0.7979,
+      "step": 6249
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.331487203578784e-07,
+      "loss": 0.9061,
+      "step": 6250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.327070090170693e-07,
+      "loss": 1.0588,
+      "step": 6251
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.322653877735583e-07,
+      "loss": 1.1098,
+      "step": 6252
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.318238566792003e-07,
+      "loss": 0.8845,
+      "step": 6253
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.313824157858419e-07,
+      "loss": 0.9511,
+      "step": 6254
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.309410651453167e-07,
+      "loss": 0.9199,
+      "step": 6255
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.304998048094505e-07,
+      "loss": 1.1229,
+      "step": 6256
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.300586348300558e-07,
+      "loss": 0.9342,
+      "step": 6257
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.296175552589366e-07,
+      "loss": 1.2469,
+      "step": 6258
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.291765661478853e-07,
+      "loss": 1.0232,
+      "step": 6259
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.287356675486845e-07,
+      "loss": 1.1008,
+      "step": 6260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.282948595131045e-07,
+      "loss": 0.7105,
+      "step": 6261
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.278541420929066e-07,
+      "loss": 1.0498,
+      "step": 6262
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.274135153398403e-07,
+      "loss": 1.075,
+      "step": 6263
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.269729793056459e-07,
+      "loss": 0.9454,
+      "step": 6264
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.265325340420509e-07,
+      "loss": 0.9821,
+      "step": 6265
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.260921796007741e-07,
+      "loss": 0.9944,
+      "step": 6266
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.256519160335229e-07,
+      "loss": 0.886,
+      "step": 6267
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.252117433919946e-07,
+      "loss": 0.8485,
+      "step": 6268
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.247716617278742e-07,
+      "loss": 1.2469,
+      "step": 6269
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.243316710928383e-07,
+      "loss": 0.6865,
+      "step": 6270
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.238917715385502e-07,
+      "loss": 0.7222,
+      "step": 6271
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.234519631166653e-07,
+      "loss": 1.0945,
+      "step": 6272
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.230122458788256e-07,
+      "loss": 1.0019,
+      "step": 6273
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.225726198766645e-07,
+      "loss": 1.0759,
+      "step": 6274
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.221330851618041e-07,
+      "loss": 0.9878,
+      "step": 6275
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.21693641785855e-07,
+      "loss": 0.8061,
+      "step": 6276
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.21254289800418e-07,
+      "loss": 1.2012,
+      "step": 6277
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.208150292570824e-07,
+      "loss": 1.019,
+      "step": 6278
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.203758602074276e-07,
+      "loss": 0.9902,
+      "step": 6279
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.199367827030219e-07,
+      "loss": 1.0335,
+      "step": 6280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.194977967954223e-07,
+      "loss": 0.7133,
+      "step": 6281
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.190589025361757e-07,
+      "loss": 0.7815,
+      "step": 6282
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.186200999768186e-07,
+      "loss": 0.278,
+      "step": 6283
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.181813891688753e-07,
+      "loss": 0.8153,
+      "step": 6284
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.177427701638613e-07,
+      "loss": 1.1031,
+      "step": 6285
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.173042430132792e-07,
+      "loss": 1.1351,
+      "step": 6286
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.168658077686226e-07,
+      "loss": 0.8849,
+      "step": 6287
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.164274644813723e-07,
+      "loss": 0.6345,
+      "step": 6288
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.159892132030017e-07,
+      "loss": 0.925,
+      "step": 6289
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.155510539849695e-07,
+      "loss": 0.8765,
+      "step": 6290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.151129868787264e-07,
+      "loss": 0.9662,
+      "step": 6291
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.146750119357105e-07,
+      "loss": 0.8229,
+      "step": 6292
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.142371292073506e-07,
+      "loss": 0.9559,
+      "step": 6293
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.137993387450629e-07,
+      "loss": 0.8378,
+      "step": 6294
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.133616406002546e-07,
+      "loss": 1.0055,
+      "step": 6295
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.129240348243201e-07,
+      "loss": 1.1665,
+      "step": 6296
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.124865214686458e-07,
+      "loss": 0.805,
+      "step": 6297
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.120491005846041e-07,
+      "loss": 0.8181,
+      "step": 6298
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.116117722235589e-07,
+      "loss": 1.0035,
+      "step": 6299
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.111745364368613e-07,
+      "loss": 1.31,
+      "step": 6300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.107373932758536e-07,
+      "loss": 0.9718,
+      "step": 6301
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.103003427918651e-07,
+      "loss": 0.8709,
+      "step": 6302
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.098633850362159e-07,
+      "loss": 1.2467,
+      "step": 6303
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.094265200602148e-07,
+      "loss": 1.082,
+      "step": 6304
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.089897479151588e-07,
+      "loss": 0.8854,
+      "step": 6305
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.085530686523348e-07,
+      "loss": 0.8965,
+      "step": 6306
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.081164823230197e-07,
+      "loss": 0.8558,
+      "step": 6307
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.076799889784771e-07,
+      "loss": 0.9265,
+      "step": 6308
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.07243588669962e-07,
+      "loss": 0.9562,
+      "step": 6309
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.068072814487165e-07,
+      "loss": 0.9635,
+      "step": 6310
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.063710673659737e-07,
+      "loss": 1.1731,
+      "step": 6311
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.059349464729552e-07,
+      "loss": 0.8695,
+      "step": 6312
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.054989188208699e-07,
+      "loss": 0.9094,
+      "step": 6313
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.050629844609186e-07,
+      "loss": 0.8719,
+      "step": 6314
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.046271434442885e-07,
+      "loss": 0.8028,
+      "step": 6315
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.041913958221582e-07,
+      "loss": 0.7525,
+      "step": 6316
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.03755741645693e-07,
+      "loss": 1.458,
+      "step": 6317
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.033201809660491e-07,
+      "loss": 0.9369,
+      "step": 6318
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.028847138343708e-07,
+      "loss": 0.7432,
+      "step": 6319
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.024493403017925e-07,
+      "loss": 0.9115,
+      "step": 6320
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.020140604194353e-07,
+      "loss": 1.2846,
+      "step": 6321
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.015788742384123e-07,
+      "loss": 0.8192,
+      "step": 6322
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.011437818098225e-07,
+      "loss": 0.9919,
+      "step": 6323
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.00708783184757e-07,
+      "loss": 0.7151,
+      "step": 6324
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.002738784142922e-07,
+      "loss": 0.7183,
+      "step": 6325
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.99839067549498e-07,
+      "loss": 1.0283,
+      "step": 6326
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.994043506414295e-07,
+      "loss": 1.0461,
+      "step": 6327
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.98969727741133e-07,
+      "loss": 0.9935,
+      "step": 6328
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.985351988996417e-07,
+      "loss": 1.043,
+      "step": 6329
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.981007641679805e-07,
+      "loss": 1.0554,
+      "step": 6330
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.9766642359716e-07,
+      "loss": 0.7318,
+      "step": 6331
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.972321772381831e-07,
+      "loss": 0.9566,
+      "step": 6332
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.967980251420382e-07,
+      "loss": 0.753,
+      "step": 6333
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.963639673597066e-07,
+      "loss": 1.0421,
+      "step": 6334
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.959300039421546e-07,
+      "loss": 0.9042,
+      "step": 6335
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.954961349403404e-07,
+      "loss": 0.8732,
+      "step": 6336
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.950623604052088e-07,
+      "loss": 0.6844,
+      "step": 6337
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.946286803876956e-07,
+      "loss": 0.7425,
+      "step": 6338
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.941950949387235e-07,
+      "loss": 0.9137,
+      "step": 6339
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.937616041092056e-07,
+      "loss": 0.7861,
+      "step": 6340
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.933282079500438e-07,
+      "loss": 0.8105,
+      "step": 6341
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.928949065121277e-07,
+      "loss": 0.7902,
+      "step": 6342
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.924616998463374e-07,
+      "loss": 0.9513,
+      "step": 6343
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.9202858800354e-07,
+      "loss": 0.9858,
+      "step": 6344
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.915955710345936e-07,
+      "loss": 1.1503,
+      "step": 6345
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.911626489903426e-07,
+      "loss": 0.8411,
+      "step": 6346
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.907298219216234e-07,
+      "loss": 1.2601,
+      "step": 6347
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.902970898792576e-07,
+      "loss": 1.0104,
+      "step": 6348
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.898644529140596e-07,
+      "loss": 1.3447,
+      "step": 6349
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.894319110768292e-07,
+      "loss": 1.0452,
+      "step": 6350
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.889994644183576e-07,
+      "loss": 1.3411,
+      "step": 6351
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.885671129894221e-07,
+      "loss": 0.9543,
+      "step": 6352
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.881348568407923e-07,
+      "loss": 0.9213,
+      "step": 6353
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.877026960232229e-07,
+      "loss": 1.2606,
+      "step": 6354
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.872706305874602e-07,
+      "loss": 0.8842,
+      "step": 6355
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.86838660584238e-07,
+      "loss": 1.0826,
+      "step": 6356
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.8640678606428e-07,
+      "loss": 0.6663,
+      "step": 6357
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.859750070782967e-07,
+      "loss": 0.8421,
+      "step": 6358
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.855433236769896e-07,
+      "loss": 0.9363,
+      "step": 6359
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.851117359110468e-07,
+      "loss": 0.8642,
+      "step": 6360
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.846802438311477e-07,
+      "loss": 0.7521,
+      "step": 6361
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.842488474879578e-07,
+      "loss": 1.0911,
+      "step": 6362
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.838175469321333e-07,
+      "loss": 0.663,
+      "step": 6363
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.833863422143183e-07,
+      "loss": 1.1939,
+      "step": 6364
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.829552333851465e-07,
+      "loss": 0.9875,
+      "step": 6365
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.825242204952386e-07,
+      "loss": 0.5274,
+      "step": 6366
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.820933035952063e-07,
+      "loss": 0.9224,
+      "step": 6367
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.816624827356475e-07,
+      "loss": 1.15,
+      "step": 6368
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.812317579671517e-07,
+      "loss": 1.0046,
+      "step": 6369
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.80801129340294e-07,
+      "loss": 0.6001,
+      "step": 6370
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.803705969056406e-07,
+      "loss": 0.9741,
+      "step": 6371
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.799401607137461e-07,
+      "loss": 0.9183,
+      "step": 6372
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.795098208151524e-07,
+      "loss": 0.8398,
+      "step": 6373
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.790795772603917e-07,
+      "loss": 0.9724,
+      "step": 6374
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.786494300999833e-07,
+      "loss": 0.6914,
+      "step": 6375
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.782193793844368e-07,
+      "loss": 0.7964,
+      "step": 6376
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.777894251642494e-07,
+      "loss": 1.3138,
+      "step": 6377
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.773595674899082e-07,
+      "loss": 0.8003,
+      "step": 6378
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.769298064118866e-07,
+      "loss": 0.8619,
+      "step": 6379
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.765001419806493e-07,
+      "loss": 1.0539,
+      "step": 6380
+    },
+    {
+      "epoch": 2.71,
+      "eval_loss": 1.2545257806777954,
+      "eval_runtime": 32.1284,
+      "eval_samples_per_second": 38.751,
+      "eval_steps_per_second": 19.391,
+      "step": 6380
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.760705742466477e-07,
+      "loss": 0.6857,
+      "step": 6381
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.756411032603235e-07,
+      "loss": 0.7803,
+      "step": 6382
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.752117290721049e-07,
+      "loss": 0.982,
+      "step": 6383
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.747824517324115e-07,
+      "loss": 0.9549,
+      "step": 6384
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.743532712916479e-07,
+      "loss": 1.0251,
+      "step": 6385
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.739241878002121e-07,
+      "loss": 1.1536,
+      "step": 6386
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.734952013084862e-07,
+      "loss": 0.8734,
+      "step": 6387
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.730663118668437e-07,
+      "loss": 0.9066,
+      "step": 6388
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.726375195256451e-07,
+      "loss": 0.9321,
+      "step": 6389
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.72208824335241e-07,
+      "loss": 1.0934,
+      "step": 6390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.717802263459686e-07,
+      "loss": 0.9762,
+      "step": 6391
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.713517256081555e-07,
+      "loss": 1.0507,
+      "step": 6392
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.709233221721174e-07,
+      "loss": 0.798,
+      "step": 6393
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.704950160881589e-07,
+      "loss": 0.8571,
+      "step": 6394
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.700668074065713e-07,
+      "loss": 1.1381,
+      "step": 6395
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.696386961776372e-07,
+      "loss": 0.3895,
+      "step": 6396
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.692106824516252e-07,
+      "loss": 0.8106,
+      "step": 6397
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.687827662787949e-07,
+      "loss": 0.722,
+      "step": 6398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.683549477093919e-07,
+      "loss": 0.8787,
+      "step": 6399
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.679272267936522e-07,
+      "loss": 0.718,
+      "step": 6400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.674996035818005e-07,
+      "loss": 0.9281,
+      "step": 6401
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.670720781240482e-07,
+      "loss": 1.2608,
+      "step": 6402
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.666446504705971e-07,
+      "loss": 0.8764,
+      "step": 6403
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.662173206716358e-07,
+      "loss": 0.9546,
+      "step": 6404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.657900887773431e-07,
+      "loss": 1.2821,
+      "step": 6405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.653629548378858e-07,
+      "loss": 1.0745,
+      "step": 6406
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.64935918903418e-07,
+      "loss": 1.0235,
+      "step": 6407
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.645089810240837e-07,
+      "loss": 0.8898,
+      "step": 6408
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.640821412500155e-07,
+      "loss": 1.0019,
+      "step": 6409
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.636553996313329e-07,
+      "loss": 0.9079,
+      "step": 6410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.632287562181459e-07,
+      "loss": 0.7231,
+      "step": 6411
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.628022110605508e-07,
+      "loss": 1.2053,
+      "step": 6412
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.623757642086347e-07,
+      "loss": 1.0692,
+      "step": 6413
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.619494157124708e-07,
+      "loss": 0.9315,
+      "step": 6414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.615231656221225e-07,
+      "loss": 1.0016,
+      "step": 6415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.610970139876409e-07,
+      "loss": 1.09,
+      "step": 6416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.606709608590665e-07,
+      "loss": 1.0847,
+      "step": 6417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.602450062864262e-07,
+      "loss": 0.7461,
+      "step": 6418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.598191503197378e-07,
+      "loss": 0.8307,
+      "step": 6419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.593933930090049e-07,
+      "loss": 1.0188,
+      "step": 6420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.589677344042222e-07,
+      "loss": 0.922,
+      "step": 6421
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.5854217455537e-07,
+      "loss": 1.0673,
+      "step": 6422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.581167135124202e-07,
+      "loss": 0.9063,
+      "step": 6423
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.576913513253303e-07,
+      "loss": 0.8589,
+      "step": 6424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.572660880440482e-07,
+      "loss": 1.0638,
+      "step": 6425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.568409237185081e-07,
+      "loss": 0.7973,
+      "step": 6426
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.564158583986349e-07,
+      "loss": 0.9945,
+      "step": 6427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.559908921343397e-07,
+      "loss": 0.8359,
+      "step": 6428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.555660249755242e-07,
+      "loss": 0.9653,
+      "step": 6429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.55141256972076e-07,
+      "loss": 0.8383,
+      "step": 6430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.547165881738729e-07,
+      "loss": 0.7916,
+      "step": 6431
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.542920186307805e-07,
+      "loss": 0.9856,
+      "step": 6432
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.538675483926535e-07,
+      "loss": 0.7592,
+      "step": 6433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.534431775093328e-07,
+      "loss": 0.7565,
+      "step": 6434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.530189060306501e-07,
+      "loss": 0.6916,
+      "step": 6435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.525947340064232e-07,
+      "loss": 1.329,
+      "step": 6436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.521706614864601e-07,
+      "loss": 0.9924,
+      "step": 6437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.517466885205567e-07,
+      "loss": 1.0379,
+      "step": 6438
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.51322815158496e-07,
+      "loss": 0.7018,
+      "step": 6439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.508990414500512e-07,
+      "loss": 0.7214,
+      "step": 6440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.504753674449816e-07,
+      "loss": 1.0415,
+      "step": 6441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.500517931930371e-07,
+      "loss": 1.1664,
+      "step": 6442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.496283187439536e-07,
+      "loss": 1.0754,
+      "step": 6443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.492049441474571e-07,
+      "loss": 0.9501,
+      "step": 6444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.487816694532611e-07,
+      "loss": 0.7351,
+      "step": 6445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.483584947110683e-07,
+      "loss": 0.5986,
+      "step": 6446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.479354199705675e-07,
+      "loss": 0.9293,
+      "step": 6447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.47512445281438e-07,
+      "loss": 1.1046,
+      "step": 6448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.470895706933459e-07,
+      "loss": 0.8393,
+      "step": 6449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.466667962559468e-07,
+      "loss": 1.0501,
+      "step": 6450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.462441220188828e-07,
+      "loss": 0.8059,
+      "step": 6451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.458215480317859e-07,
+      "loss": 0.6951,
+      "step": 6452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.453990743442758e-07,
+      "loss": 0.6543,
+      "step": 6453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.449767010059605e-07,
+      "loss": 1.1133,
+      "step": 6454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.445544280664353e-07,
+      "loss": 1.0028,
+      "step": 6455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.441322555752856e-07,
+      "loss": 0.7046,
+      "step": 6456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.437101835820825e-07,
+      "loss": 1.0707,
+      "step": 6457
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.432882121363879e-07,
+      "loss": 1.1937,
+      "step": 6458
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.428663412877488e-07,
+      "loss": 0.7647,
+      "step": 6459
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.424445710857048e-07,
+      "loss": 0.9687,
+      "step": 6460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.420229015797791e-07,
+      "loss": 0.8943,
+      "step": 6461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.416013328194866e-07,
+      "loss": 0.8763,
+      "step": 6462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.411798648543274e-07,
+      "loss": 1.1021,
+      "step": 6463
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.407584977337926e-07,
+      "loss": 0.8248,
+      "step": 6464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.403372315073587e-07,
+      "loss": 0.6978,
+      "step": 6465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.399160662244931e-07,
+      "loss": 1.2437,
+      "step": 6466
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.394950019346488e-07,
+      "loss": 0.9643,
+      "step": 6467
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.390740386872687e-07,
+      "loss": 0.8533,
+      "step": 6468
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.386531765317837e-07,
+      "loss": 0.9054,
+      "step": 6469
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.382324155176115e-07,
+      "loss": 0.8742,
+      "step": 6470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.378117556941596e-07,
+      "loss": 1.055,
+      "step": 6471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.373911971108224e-07,
+      "loss": 0.9719,
+      "step": 6472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.369707398169831e-07,
+      "loss": 0.816,
+      "step": 6473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.365503838620116e-07,
+      "loss": 0.7475,
+      "step": 6474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.361301292952693e-07,
+      "loss": 1.1277,
+      "step": 6475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.357099761661017e-07,
+      "loss": 0.8285,
+      "step": 6476
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.352899245238449e-07,
+      "loss": 0.8416,
+      "step": 6477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.348699744178218e-07,
+      "loss": 0.7066,
+      "step": 6478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.344501258973446e-07,
+      "loss": 0.6872,
+      "step": 6479
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.34030379011712e-07,
+      "loss": 0.9709,
+      "step": 6480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.336107338102127e-07,
+      "loss": 0.9931,
+      "step": 6481
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.331911903421207e-07,
+      "loss": 0.8026,
+      "step": 6482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.327717486567018e-07,
+      "loss": 1.116,
+      "step": 6483
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.323524088032064e-07,
+      "loss": 0.919,
+      "step": 6484
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.319331708308752e-07,
+      "loss": 1.0745,
+      "step": 6485
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.315140347889355e-07,
+      "loss": 0.9946,
+      "step": 6486
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.310950007266036e-07,
+      "loss": 0.8159,
+      "step": 6487
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.306760686930827e-07,
+      "loss": 1.4081,
+      "step": 6488
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.302572387375654e-07,
+      "loss": 0.8316,
+      "step": 6489
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.298385109092315e-07,
+      "loss": 0.7992,
+      "step": 6490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.294198852572496e-07,
+      "loss": 0.643,
+      "step": 6491
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.290013618307747e-07,
+      "loss": 0.6285,
+      "step": 6492
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.285829406789516e-07,
+      "loss": 0.9211,
+      "step": 6493
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.281646218509114e-07,
+      "loss": 0.9752,
+      "step": 6494
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.277464053957749e-07,
+      "loss": 1.0382,
+      "step": 6495
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.273282913626492e-07,
+      "loss": 0.7915,
+      "step": 6496
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.269102798006307e-07,
+      "loss": 0.8505,
+      "step": 6497
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.264923707588037e-07,
+      "loss": 0.9071,
+      "step": 6498
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.26074564286239e-07,
+      "loss": 0.8172,
+      "step": 6499
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.256568604319975e-07,
+      "loss": 0.7564,
+      "step": 6500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.252392592451259e-07,
+      "loss": 0.7912,
+      "step": 6501
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.248217607746603e-07,
+      "loss": 0.8395,
+      "step": 6502
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.244043650696251e-07,
+      "loss": 0.8115,
+      "step": 6503
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.239870721790305e-07,
+      "loss": 0.8101,
+      "step": 6504
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.235698821518767e-07,
+      "loss": 0.7965,
+      "step": 6505
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.231527950371516e-07,
+      "loss": 0.7299,
+      "step": 6506
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.227358108838294e-07,
+      "loss": 1.1986,
+      "step": 6507
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.223189297408745e-07,
+      "loss": 1.2124,
+      "step": 6508
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.219021516572369e-07,
+      "loss": 0.916,
+      "step": 6509
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.214854766818567e-07,
+      "loss": 0.8837,
+      "step": 6510
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.210689048636593e-07,
+      "loss": 0.9822,
+      "step": 6511
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.206524362515615e-07,
+      "loss": 0.8818,
+      "step": 6512
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.202360708944644e-07,
+      "loss": 0.9367,
+      "step": 6513
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.1981980884126e-07,
+      "loss": 0.6733,
+      "step": 6514
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.194036501408252e-07,
+      "loss": 1.046,
+      "step": 6515
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.189875948420276e-07,
+      "loss": 0.8613,
+      "step": 6516
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.185716429937204e-07,
+      "loss": 0.7764,
+      "step": 6517
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.181557946447465e-07,
+      "loss": 0.7077,
+      "step": 6518
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.17740049843934e-07,
+      "loss": 1.0088,
+      "step": 6519
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.173244086401032e-07,
+      "loss": 0.8561,
+      "step": 6520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.169088710820578e-07,
+      "loss": 1.0805,
+      "step": 6521
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.164934372185921e-07,
+      "loss": 0.9125,
+      "step": 6522
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.160781070984864e-07,
+      "loss": 0.7905,
+      "step": 6523
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.156628807705108e-07,
+      "loss": 1.135,
+      "step": 6524
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.152477582834212e-07,
+      "loss": 1.4011,
+      "step": 6525
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.148327396859622e-07,
+      "loss": 0.7254,
+      "step": 6526
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.144178250268673e-07,
+      "loss": 0.744,
+      "step": 6527
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.140030143548556e-07,
+      "loss": 1.0048,
+      "step": 6528
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.135883077186354e-07,
+      "loss": 0.8705,
+      "step": 6529
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.131737051669032e-07,
+      "loss": 1.0471,
+      "step": 6530
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.127592067483416e-07,
+      "loss": 0.8154,
+      "step": 6531
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.123448125116227e-07,
+      "loss": 1.3728,
+      "step": 6532
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.119305225054049e-07,
+      "loss": 1.1007,
+      "step": 6533
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.115163367783354e-07,
+      "loss": 0.885,
+      "step": 6534
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.111022553790496e-07,
+      "loss": 0.9406,
+      "step": 6535
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.106882783561684e-07,
+      "loss": 1.0341,
+      "step": 6536
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.102744057583036e-07,
+      "loss": 1.439,
+      "step": 6537
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.098606376340514e-07,
+      "loss": 0.6819,
+      "step": 6538
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.094469740319989e-07,
+      "loss": 1.0155,
+      "step": 6539
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.09033415000718e-07,
+      "loss": 0.8907,
+      "step": 6540
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.086199605887705e-07,
+      "loss": 0.8208,
+      "step": 6541
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.082066108447051e-07,
+      "loss": 0.8655,
+      "step": 6542
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.07793365817059e-07,
+      "loss": 1.2304,
+      "step": 6543
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.07380225554355e-07,
+      "loss": 0.7905,
+      "step": 6544
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.069671901051062e-07,
+      "loss": 0.7548,
+      "step": 6545
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.065542595178112e-07,
+      "loss": 0.8964,
+      "step": 6546
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.06141433840958e-07,
+      "loss": 0.7382,
+      "step": 6547
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.057287131230208e-07,
+      "loss": 1.0803,
+      "step": 6548
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.053160974124625e-07,
+      "loss": 0.9824,
+      "step": 6549
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.049035867577337e-07,
+      "loss": 1.1075,
+      "step": 6550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.044911812072724e-07,
+      "loss": 0.9736,
+      "step": 6551
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.040788808095036e-07,
+      "loss": 1.0735,
+      "step": 6552
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.036666856128414e-07,
+      "loss": 1.0127,
+      "step": 6553
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.032545956656856e-07,
+      "loss": 0.7356,
+      "step": 6554
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.028426110164259e-07,
+      "loss": 1.0483,
+      "step": 6555
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.024307317134368e-07,
+      "loss": 0.8807,
+      "step": 6556
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.020189578050844e-07,
+      "loss": 1.1198,
+      "step": 6557
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.016072893397184e-07,
+      "loss": 1.0672,
+      "step": 6558
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.011957263656789e-07,
+      "loss": 0.7609,
+      "step": 6559
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.007842689312916e-07,
+      "loss": 1.0014,
+      "step": 6560
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.003729170848718e-07,
+      "loss": 1.1831,
+      "step": 6561
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.999616708747202e-07,
+      "loss": 0.827,
+      "step": 6562
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.995505303491269e-07,
+      "loss": 1.1052,
+      "step": 6563
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.991394955563696e-07,
+      "loss": 0.9942,
+      "step": 6564
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.987285665447118e-07,
+      "loss": 0.9956,
+      "step": 6565
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.983177433624066e-07,
+      "loss": 1.0859,
+      "step": 6566
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.979070260576929e-07,
+      "loss": 0.7246,
+      "step": 6567
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.974964146787992e-07,
+      "loss": 0.8389,
+      "step": 6568
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.970859092739392e-07,
+      "loss": 0.7411,
+      "step": 6569
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.966755098913167e-07,
+      "loss": 1.3455,
+      "step": 6570
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.962652165791199e-07,
+      "loss": 1.3862,
+      "step": 6571
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.958550293855285e-07,
+      "loss": 0.5103,
+      "step": 6572
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.954449483587061e-07,
+      "loss": 0.8344,
+      "step": 6573
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.950349735468065e-07,
+      "loss": 1.1233,
+      "step": 6574
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.946251049979688e-07,
+      "loss": 0.8389,
+      "step": 6575
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.942153427603216e-07,
+      "loss": 0.9522,
+      "step": 6576
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.938056868819791e-07,
+      "loss": 1.2423,
+      "step": 6577
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.933961374110446e-07,
+      "loss": 0.9128,
+      "step": 6578
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.929866943956084e-07,
+      "loss": 1.1827,
+      "step": 6579
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.925773578837486e-07,
+      "loss": 0.8981,
+      "step": 6580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.921681279235297e-07,
+      "loss": 1.5603,
+      "step": 6581
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.917590045630052e-07,
+      "loss": 1.1346,
+      "step": 6582
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.913499878502142e-07,
+      "loss": 0.672,
+      "step": 6583
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.909410778331855e-07,
+      "loss": 0.8192,
+      "step": 6584
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.905322745599333e-07,
+      "loss": 0.77,
+      "step": 6585
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.901235780784608e-07,
+      "loss": 0.8515,
+      "step": 6586
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.897149884367577e-07,
+      "loss": 0.8705,
+      "step": 6587
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.893065056828026e-07,
+      "loss": 0.8793,
+      "step": 6588
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.888981298645589e-07,
+      "loss": 1.022,
+      "step": 6589
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.884898610299802e-07,
+      "loss": 0.7797,
+      "step": 6590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.880816992270057e-07,
+      "loss": 1.1072,
+      "step": 6591
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.876736445035633e-07,
+      "loss": 0.7996,
+      "step": 6592
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.872656969075668e-07,
+      "loss": 1.046,
+      "step": 6593
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.868578564869191e-07,
+      "loss": 1.0244,
+      "step": 6594
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.864501232895099e-07,
+      "loss": 0.9981,
+      "step": 6595
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.860424973632151e-07,
+      "loss": 0.9411,
+      "step": 6596
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.856349787559006e-07,
+      "loss": 1.7989,
+      "step": 6597
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.852275675154167e-07,
+      "loss": 0.781,
+      "step": 6598
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.848202636896036e-07,
+      "loss": 1.1002,
+      "step": 6599
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.844130673262871e-07,
+      "loss": 0.8278,
+      "step": 6600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.840059784732814e-07,
+      "loss": 1.1788,
+      "step": 6601
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.83598997178388e-07,
+      "loss": 0.7295,
+      "step": 6602
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.831921234893958e-07,
+      "loss": 0.9226,
+      "step": 6603
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.827853574540801e-07,
+      "loss": 1.086,
+      "step": 6604
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.823786991202049e-07,
+      "loss": 1.2158,
+      "step": 6605
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.819721485355207e-07,
+      "loss": 0.874,
+      "step": 6606
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.815657057477657e-07,
+      "loss": 0.9021,
+      "step": 6607
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.811593708046645e-07,
+      "loss": 0.9287,
+      "step": 6608
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.807531437539317e-07,
+      "loss": 0.9305,
+      "step": 6609
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.803470246432659e-07,
+      "loss": 0.6112,
+      "step": 6610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.799410135203554e-07,
+      "loss": 1.0361,
+      "step": 6611
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.795351104328741e-07,
+      "loss": 0.6344,
+      "step": 6612
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.791293154284849e-07,
+      "loss": 0.9735,
+      "step": 6613
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.787236285548365e-07,
+      "loss": 0.8022,
+      "step": 6614
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.783180498595663e-07,
+      "loss": 1.2343,
+      "step": 6615
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.779125793902967e-07,
+      "loss": 1.0711,
+      "step": 6616
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.775072171946413e-07,
+      "loss": 0.9527,
+      "step": 6617
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.77101963320197e-07,
+      "loss": 1.0889,
+      "step": 6618
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.766968178145505e-07,
+      "loss": 0.6484,
+      "step": 6619
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.762917807252742e-07,
+      "loss": 0.8916,
+      "step": 6620
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.758868520999292e-07,
+      "loss": 0.757,
+      "step": 6621
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.754820319860622e-07,
+      "loss": 0.8059,
+      "step": 6622
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.750773204312088e-07,
+      "loss": 0.8707,
+      "step": 6623
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.746727174828914e-07,
+      "loss": 0.8642,
+      "step": 6624
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.742682231886184e-07,
+      "loss": 0.7834,
+      "step": 6625
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.738638375958879e-07,
+      "loss": 1.0096,
+      "step": 6626
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.734595607521822e-07,
+      "loss": 0.8443,
+      "step": 6627
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.730553927049734e-07,
+      "loss": 0.9626,
+      "step": 6628
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.7265133350172e-07,
+      "loss": 0.9422,
+      "step": 6629
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.722473831898665e-07,
+      "loss": 0.9313,
+      "step": 6630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.718435418168464e-07,
+      "loss": 0.9742,
+      "step": 6631
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.714398094300802e-07,
+      "loss": 1.0311,
+      "step": 6632
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.710361860769738e-07,
+      "loss": 0.7591,
+      "step": 6633
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.706326718049228e-07,
+      "loss": 0.963,
+      "step": 6634
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.702292666613078e-07,
+      "loss": 1.0482,
+      "step": 6635
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.698259706934985e-07,
+      "loss": 0.7216,
+      "step": 6636
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.694227839488497e-07,
+      "loss": 1.0699,
+      "step": 6637
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.69019706474705e-07,
+      "loss": 0.9313,
+      "step": 6638
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.68616738318395e-07,
+      "loss": 1.2297,
+      "step": 6639
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.682138795272374e-07,
+      "loss": 0.8746,
+      "step": 6640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.678111301485359e-07,
+      "loss": 0.7744,
+      "step": 6641
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.674084902295831e-07,
+      "loss": 0.8652,
+      "step": 6642
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.67005959817657e-07,
+      "loss": 0.9353,
+      "step": 6643
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.666035389600247e-07,
+      "loss": 0.808,
+      "step": 6644
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.662012277039378e-07,
+      "loss": 0.8753,
+      "step": 6645
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.657990260966385e-07,
+      "loss": 1.2135,
+      "step": 6646
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.653969341853528e-07,
+      "loss": 0.9469,
+      "step": 6647
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.649949520172965e-07,
+      "loss": 0.7446,
+      "step": 6648
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.645930796396701e-07,
+      "loss": 0.8081,
+      "step": 6649
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.641913170996633e-07,
+      "loss": 0.9327,
+      "step": 6650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.63789664444451e-07,
+      "loss": 0.7342,
+      "step": 6651
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.633881217211971e-07,
+      "loss": 1.2369,
+      "step": 6652
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.629866889770507e-07,
+      "loss": 1.0942,
+      "step": 6653
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.625853662591497e-07,
+      "loss": 1.0137,
+      "step": 6654
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.62184153614618e-07,
+      "loss": 1.0531,
+      "step": 6655
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.617830510905676e-07,
+      "loss": 0.8828,
+      "step": 6656
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.613820587340959e-07,
+      "loss": 1.1156,
+      "step": 6657
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.609811765922892e-07,
+      "loss": 1.0926,
+      "step": 6658
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.60580404712219e-07,
+      "loss": 0.8144,
+      "step": 6659
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.601797431409453e-07,
+      "loss": 1.2064,
+      "step": 6660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.597791919255156e-07,
+      "loss": 1.0661,
+      "step": 6661
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.593787511129619e-07,
+      "loss": 0.8748,
+      "step": 6662
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.589784207503064e-07,
+      "loss": 0.877,
+      "step": 6663
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.585782008845557e-07,
+      "loss": 0.8054,
+      "step": 6664
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.581780915627053e-07,
+      "loss": 1.0534,
+      "step": 6665
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.577780928317363e-07,
+      "loss": 0.7029,
+      "step": 6666
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.573782047386181e-07,
+      "loss": 1.0319,
+      "step": 6667
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.569784273303052e-07,
+      "loss": 1.1198,
+      "step": 6668
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.565787606537423e-07,
+      "loss": 1.4661,
+      "step": 6669
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.561792047558577e-07,
+      "loss": 1.0039,
+      "step": 6670
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.557797596835692e-07,
+      "loss": 1.0713,
+      "step": 6671
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.553804254837795e-07,
+      "loss": 0.5375,
+      "step": 6672
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.549812022033804e-07,
+      "loss": 1.0477,
+      "step": 6673
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.545820898892484e-07,
+      "loss": 0.8886,
+      "step": 6674
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.541830885882489e-07,
+      "loss": 0.8701,
+      "step": 6675
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.537841983472337e-07,
+      "loss": 0.8144,
+      "step": 6676
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.533854192130415e-07,
+      "loss": 0.8335,
+      "step": 6677
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.529867512324971e-07,
+      "loss": 0.9749,
+      "step": 6678
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.525881944524139e-07,
+      "loss": 0.8123,
+      "step": 6679
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.521897489195905e-07,
+      "loss": 0.8737,
+      "step": 6680
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.517914146808141e-07,
+      "loss": 1.2291,
+      "step": 6681
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.513931917828572e-07,
+      "loss": 0.8525,
+      "step": 6682
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.509950802724804e-07,
+      "loss": 0.9895,
+      "step": 6683
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.505970801964309e-07,
+      "loss": 0.9287,
+      "step": 6684
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.501991916014434e-07,
+      "loss": 0.7347,
+      "step": 6685
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.498014145342376e-07,
+      "loss": 0.9282,
+      "step": 6686
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.494037490415229e-07,
+      "loss": 0.9136,
+      "step": 6687
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.490061951699925e-07,
+      "loss": 0.9592,
+      "step": 6688
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.486087529663293e-07,
+      "loss": 1.2893,
+      "step": 6689
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.48211422477201e-07,
+      "loss": 1.3141,
+      "step": 6690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.478142037492636e-07,
+      "loss": 0.9035,
+      "step": 6691
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.474170968291596e-07,
+      "loss": 0.8267,
+      "step": 6692
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.470201017635175e-07,
+      "loss": 0.8622,
+      "step": 6693
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.466232185989544e-07,
+      "loss": 0.6549,
+      "step": 6694
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.46226447382072e-07,
+      "loss": 1.1745,
+      "step": 6695
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.458297881594611e-07,
+      "loss": 1.0481,
+      "step": 6696
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.45433240977697e-07,
+      "loss": 0.8917,
+      "step": 6697
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.450368058833452e-07,
+      "loss": 1.0141,
+      "step": 6698
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.446404829229542e-07,
+      "loss": 0.9496,
+      "step": 6699
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.442442721430626e-07,
+      "loss": 0.8455,
+      "step": 6700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.43848173590193e-07,
+      "loss": 1.1289,
+      "step": 6701
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.434521873108573e-07,
+      "loss": 0.8126,
+      "step": 6702
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.430563133515524e-07,
+      "loss": 0.8972,
+      "step": 6703
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.426605517587633e-07,
+      "loss": 0.9698,
+      "step": 6704
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.422649025789598e-07,
+      "loss": 0.8235,
+      "step": 6705
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.418693658586022e-07,
+      "loss": 0.6997,
+      "step": 6706
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.414739416441336e-07,
+      "loss": 1.3812,
+      "step": 6707
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.410786299819867e-07,
+      "loss": 1.2818,
+      "step": 6708
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.406834309185789e-07,
+      "loss": 0.9557,
+      "step": 6709
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.402883445003162e-07,
+      "loss": 0.9007,
+      "step": 6710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.398933707735899e-07,
+      "loss": 0.8003,
+      "step": 6711
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.39498509784779e-07,
+      "loss": 0.9896,
+      "step": 6712
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.391037615802488e-07,
+      "loss": 0.9956,
+      "step": 6713
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.387091262063525e-07,
+      "loss": 0.7331,
+      "step": 6714
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.383146037094277e-07,
+      "loss": 0.963,
+      "step": 6715
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.379201941358012e-07,
+      "loss": 0.9099,
+      "step": 6716
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.375258975317848e-07,
+      "loss": 1.1426,
+      "step": 6717
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.371317139436782e-07,
+      "loss": 0.8337,
+      "step": 6718
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.367376434177668e-07,
+      "loss": 0.8163,
+      "step": 6719
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.363436860003233e-07,
+      "loss": 0.8717,
+      "step": 6720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.359498417376079e-07,
+      "loss": 1.0425,
+      "step": 6721
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.355561106758656e-07,
+      "loss": 0.8522,
+      "step": 6722
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.351624928613303e-07,
+      "loss": 0.9776,
+      "step": 6723
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.347689883402204e-07,
+      "loss": 0.9036,
+      "step": 6724
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.343755971587428e-07,
+      "loss": 0.9006,
+      "step": 6725
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.339823193630904e-07,
+      "loss": 0.7296,
+      "step": 6726
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.335891549994422e-07,
+      "loss": 0.9245,
+      "step": 6727
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.33196104113965e-07,
+      "loss": 0.9016,
+      "step": 6728
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.328031667528119e-07,
+      "loss": 1.3822,
+      "step": 6729
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.324103429621218e-07,
+      "loss": 1.0795,
+      "step": 6730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.320176327880219e-07,
+      "loss": 1.2789,
+      "step": 6731
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.316250362766239e-07,
+      "loss": 0.7353,
+      "step": 6732
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.312325534740287e-07,
+      "loss": 1.0117,
+      "step": 6733
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.308401844263215e-07,
+      "loss": 0.8736,
+      "step": 6734
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.304479291795754e-07,
+      "loss": 0.8886,
+      "step": 6735
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.300557877798501e-07,
+      "loss": 0.8625,
+      "step": 6736
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.296637602731922e-07,
+      "loss": 0.8127,
+      "step": 6737
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.292718467056336e-07,
+      "loss": 0.8307,
+      "step": 6738
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.288800471231945e-07,
+      "loss": 0.4992,
+      "step": 6739
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.284883615718798e-07,
+      "loss": 1.0425,
+      "step": 6740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.280967900976833e-07,
+      "loss": 0.7193,
+      "step": 6741
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.27705332746583e-07,
+      "loss": 1.2148,
+      "step": 6742
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.273139895645462e-07,
+      "loss": 1.3616,
+      "step": 6743
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.26922760597524e-07,
+      "loss": 0.9569,
+      "step": 6744
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.265316458914565e-07,
+      "loss": 1.0759,
+      "step": 6745
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.261406454922681e-07,
+      "loss": 1.3272,
+      "step": 6746
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.257497594458723e-07,
+      "loss": 0.8707,
+      "step": 6747
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.253589877981663e-07,
+      "loss": 0.9871,
+      "step": 6748
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.24968330595037e-07,
+      "loss": 1.0083,
+      "step": 6749
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.245777878823549e-07,
+      "loss": 0.9802,
+      "step": 6750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.241873597059788e-07,
+      "loss": 1.062,
+      "step": 6751
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.237970461117544e-07,
+      "loss": 0.6993,
+      "step": 6752
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.234068471455122e-07,
+      "loss": 0.8015,
+      "step": 6753
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.230167628530709e-07,
+      "loss": 0.728,
+      "step": 6754
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.226267932802352e-07,
+      "loss": 0.8624,
+      "step": 6755
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.222369384727955e-07,
+      "loss": 1.1083,
+      "step": 6756
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.218471984765298e-07,
+      "loss": 1.1479,
+      "step": 6757
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.214575733372029e-07,
+      "loss": 0.8115,
+      "step": 6758
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.210680631005644e-07,
+      "loss": 0.8289,
+      "step": 6759
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.206786678123524e-07,
+      "loss": 1.0693,
+      "step": 6760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.202893875182899e-07,
+      "loss": 0.7835,
+      "step": 6761
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.199002222640877e-07,
+      "loss": 0.8986,
+      "step": 6762
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.195111720954417e-07,
+      "loss": 0.6207,
+      "step": 6763
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.191222370580354e-07,
+      "loss": 1.1123,
+      "step": 6764
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.187334171975387e-07,
+      "loss": 0.7692,
+      "step": 6765
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.183447125596078e-07,
+      "loss": 0.9488,
+      "step": 6766
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.179561231898847e-07,
+      "loss": 1.0254,
+      "step": 6767
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.175676491339992e-07,
+      "loss": 0.6845,
+      "step": 6768
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.17179290437566e-07,
+      "loss": 1.1208,
+      "step": 6769
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.167910471461879e-07,
+      "loss": 1.1642,
+      "step": 6770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.164029193054525e-07,
+      "loss": 0.8239,
+      "step": 6771
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.160149069609349e-07,
+      "loss": 1.3108,
+      "step": 6772
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.156270101581966e-07,
+      "loss": 0.7765,
+      "step": 6773
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.152392289427857e-07,
+      "loss": 1.026,
+      "step": 6774
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.148515633602356e-07,
+      "loss": 0.9479,
+      "step": 6775
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.144640134560675e-07,
+      "loss": 0.7622,
+      "step": 6776
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.140765792757878e-07,
+      "loss": 0.9292,
+      "step": 6777
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.136892608648906e-07,
+      "loss": 0.7639,
+      "step": 6778
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.133020582688548e-07,
+      "loss": 0.8791,
+      "step": 6779
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.129149715331472e-07,
+      "loss": 0.8654,
+      "step": 6780
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.125280007032203e-07,
+      "loss": 1.0054,
+      "step": 6781
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.121411458245136e-07,
+      "loss": 1.0754,
+      "step": 6782
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.117544069424515e-07,
+      "loss": 0.9301,
+      "step": 6783
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.113677841024468e-07,
+      "loss": 1.0822,
+      "step": 6784
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.109812773498967e-07,
+      "loss": 0.8768,
+      "step": 6785
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.105948867301867e-07,
+      "loss": 0.9711,
+      "step": 6786
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.102086122886865e-07,
+      "loss": 1.0114,
+      "step": 6787
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.098224540707539e-07,
+      "loss": 1.0124,
+      "step": 6788
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.094364121217329e-07,
+      "loss": 0.8331,
+      "step": 6789
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.090504864869528e-07,
+      "loss": 0.7554,
+      "step": 6790
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.086646772117306e-07,
+      "loss": 0.9992,
+      "step": 6791
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.082789843413678e-07,
+      "loss": 0.9784,
+      "step": 6792
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.078934079211545e-07,
+      "loss": 0.3653,
+      "step": 6793
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.075079479963645e-07,
+      "loss": 0.7908,
+      "step": 6794
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.071226046122612e-07,
+      "loss": 1.0422,
+      "step": 6795
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.06737377814091e-07,
+      "loss": 0.6614,
+      "step": 6796
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.063522676470894e-07,
+      "loss": 1.2836,
+      "step": 6797
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.059672741564757e-07,
+      "loss": 0.9922,
+      "step": 6798
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.055823973874577e-07,
+      "loss": 0.9041,
+      "step": 6799
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.051976373852277e-07,
+      "loss": 1.0102,
+      "step": 6800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.048129941949657e-07,
+      "loss": 1.1083,
+      "step": 6801
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.044284678618365e-07,
+      "loss": 0.8899,
+      "step": 6802
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.040440584309935e-07,
+      "loss": 0.5971,
+      "step": 6803
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.036597659475736e-07,
+      "loss": 1.1725,
+      "step": 6804
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.032755904567024e-07,
+      "loss": 0.9862,
+      "step": 6805
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.028915320034895e-07,
+      "loss": 1.0762,
+      "step": 6806
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.025075906330332e-07,
+      "loss": 1.0328,
+      "step": 6807
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.021237663904155e-07,
+      "loss": 0.9143,
+      "step": 6808
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.017400593207066e-07,
+      "loss": 0.8756,
+      "step": 6809
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.013564694689622e-07,
+      "loss": 1.0568,
+      "step": 6810
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.009729968802246e-07,
+      "loss": 1.003,
+      "step": 6811
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.005896415995213e-07,
+      "loss": 0.9674,
+      "step": 6812
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.002064036718678e-07,
+      "loss": 0.9427,
+      "step": 6813
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.998232831422635e-07,
+      "loss": 1.065,
+      "step": 6814
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.994402800556964e-07,
+      "loss": 1.1748,
+      "step": 6815
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.990573944571388e-07,
+      "loss": 0.8511,
+      "step": 6816
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.9867462639155e-07,
+      "loss": 0.8082,
+      "step": 6817
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.982919759038767e-07,
+      "loss": 0.698,
+      "step": 6818
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.97909443039049e-07,
+      "loss": 0.6419,
+      "step": 6819
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.97527027841986e-07,
+      "loss": 0.7873,
+      "step": 6820
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.971447303575908e-07,
+      "loss": 1.2329,
+      "step": 6821
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.967625506307546e-07,
+      "loss": 0.7233,
+      "step": 6822
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.963804887063529e-07,
+      "loss": 1.1292,
+      "step": 6823
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.959985446292487e-07,
+      "loss": 1.0467,
+      "step": 6824
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.956167184442906e-07,
+      "loss": 0.8471,
+      "step": 6825
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.952350101963145e-07,
+      "loss": 0.866,
+      "step": 6826
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.948534199301399e-07,
+      "loss": 1.0202,
+      "step": 6827
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.944719476905753e-07,
+      "loss": 0.9948,
+      "step": 6828
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.940905935224129e-07,
+      "loss": 1.0247,
+      "step": 6829
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.937093574704333e-07,
+      "loss": 0.8885,
+      "step": 6830
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.933282395794006e-07,
+      "loss": 0.8857,
+      "step": 6831
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.929472398940684e-07,
+      "loss": 1.1904,
+      "step": 6832
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.925663584591732e-07,
+      "loss": 0.7756,
+      "step": 6833
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.921855953194397e-07,
+      "loss": 0.6324,
+      "step": 6834
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.918049505195774e-07,
+      "loss": 0.9053,
+      "step": 6835
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.914244241042833e-07,
+      "loss": 0.8347,
+      "step": 6836
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.910440161182386e-07,
+      "loss": 0.8604,
+      "step": 6837
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.906637266061127e-07,
+      "loss": 0.9569,
+      "step": 6838
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.902835556125585e-07,
+      "loss": 1.0493,
+      "step": 6839
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.899035031822189e-07,
+      "loss": 0.7797,
+      "step": 6840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.895235693597187e-07,
+      "loss": 0.8133,
+      "step": 6841
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.891437541896715e-07,
+      "loss": 1.1558,
+      "step": 6842
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.887640577166753e-07,
+      "loss": 0.8481,
+      "step": 6843
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.883844799853158e-07,
+      "loss": 1.0548,
+      "step": 6844
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.88005021040163e-07,
+      "loss": 1.1777,
+      "step": 6845
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.876256809257744e-07,
+      "loss": 1.0899,
+      "step": 6846
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.872464596866931e-07,
+      "loss": 0.8132,
+      "step": 6847
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.868673573674477e-07,
+      "loss": 0.7671,
+      "step": 6848
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.864883740125538e-07,
+      "loss": 1.1772,
+      "step": 6849
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.861095096665116e-07,
+      "loss": 0.8988,
+      "step": 6850
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.857307643738087e-07,
+      "loss": 0.8394,
+      "step": 6851
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.853521381789189e-07,
+      "loss": 1.2451,
+      "step": 6852
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.849736311263003e-07,
+      "loss": 0.6033,
+      "step": 6853
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.845952432603984e-07,
+      "loss": 0.9944,
+      "step": 6854
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.842169746256451e-07,
+      "loss": 0.7334,
+      "step": 6855
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.838388252664564e-07,
+      "loss": 0.915,
+      "step": 6856
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.834607952272364e-07,
+      "loss": 0.6945,
+      "step": 6857
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.830828845523734e-07,
+      "loss": 0.7072,
+      "step": 6858
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.827050932862436e-07,
+      "loss": 0.883,
+      "step": 6859
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.82327421473207e-07,
+      "loss": 0.9067,
+      "step": 6860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.819498691576112e-07,
+      "loss": 4.7683,
+      "step": 6861
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.815724363837893e-07,
+      "loss": 1.2178,
+      "step": 6862
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.811951231960606e-07,
+      "loss": 0.5719,
+      "step": 6863
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.808179296387294e-07,
+      "loss": 0.8609,
+      "step": 6864
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.804408557560875e-07,
+      "loss": 0.9074,
+      "step": 6865
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.800639015924106e-07,
+      "loss": 1.0529,
+      "step": 6866
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.796870671919628e-07,
+      "loss": 1.0007,
+      "step": 6867
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.793103525989917e-07,
+      "loss": 0.9253,
+      "step": 6868
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.789337578577326e-07,
+      "loss": 1.0018,
+      "step": 6869
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.78557283012406e-07,
+      "loss": 0.6839,
+      "step": 6870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.781809281072189e-07,
+      "loss": 0.8069,
+      "step": 6871
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.778046931863629e-07,
+      "loss": 0.9593,
+      "step": 6872
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.774285782940173e-07,
+      "loss": 0.7618,
+      "step": 6873
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.770525834743453e-07,
+      "loss": 0.9935,
+      "step": 6874
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.766767087714983e-07,
+      "loss": 0.6619,
+      "step": 6875
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.76300954229611e-07,
+      "loss": 0.777,
+      "step": 6876
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7592531989280615e-07,
+      "loss": 1.0431,
+      "step": 6877
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.755498058051915e-07,
+      "loss": 0.9674,
+      "step": 6878
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7517441201086124e-07,
+      "loss": 0.7967,
+      "step": 6879
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.747991385538942e-07,
+      "loss": 0.8273,
+      "step": 6880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7442398547835655e-07,
+      "loss": 0.843,
+      "step": 6881
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.740489528282986e-07,
+      "loss": 1.214,
+      "step": 6882
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.736740406477583e-07,
+      "loss": 0.9618,
+      "step": 6883
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.73299248980759e-07,
+      "loss": 0.9377,
+      "step": 6884
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.729245778713087e-07,
+      "loss": 1.2154,
+      "step": 6885
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.72550027363403e-07,
+      "loss": 1.1449,
+      "step": 6886
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7217559750102163e-07,
+      "loss": 0.8987,
+      "step": 6887
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7180128832813187e-07,
+      "loss": 0.973,
+      "step": 6888
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7142709988868515e-07,
+      "loss": 0.9007,
+      "step": 6889
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.710530322266204e-07,
+      "loss": 0.8879,
+      "step": 6890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.706790853858599e-07,
+      "loss": 0.9737,
+      "step": 6891
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.703052594103156e-07,
+      "loss": 0.6417,
+      "step": 6892
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.6993155434388135e-07,
+      "loss": 0.8494,
+      "step": 6893
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.695579702304394e-07,
+      "loss": 1.1017,
+      "step": 6894
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.691845071138558e-07,
+      "loss": 0.9717,
+      "step": 6895
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.688111650379847e-07,
+      "loss": 1.0119,
+      "step": 6896
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6843794404666356e-07,
+      "loss": 1.0837,
+      "step": 6897
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.680648441837175e-07,
+      "loss": 0.9741,
+      "step": 6898
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.676918654929567e-07,
+      "loss": 0.7183,
+      "step": 6899
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6731900801817747e-07,
+      "loss": 1.0969,
+      "step": 6900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.669462718031608e-07,
+      "loss": 0.9842,
+      "step": 6901
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.665736568916752e-07,
+      "loss": 0.9479,
+      "step": 6902
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6620116332747317e-07,
+      "loss": 0.8808,
+      "step": 6903
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6582879115429435e-07,
+      "loss": 1.1241,
+      "step": 6904
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.654565404158626e-07,
+      "loss": 0.751,
+      "step": 6905
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6508441115588937e-07,
+      "loss": 1.1436,
+      "step": 6906
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6471240341807044e-07,
+      "loss": 0.9666,
+      "step": 6907
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6434051724608855e-07,
+      "loss": 0.5231,
+      "step": 6908
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6396875268361035e-07,
+      "loss": 1.0677,
+      "step": 6909
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6359710977429033e-07,
+      "loss": 0.9249,
+      "step": 6910
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6322558856176663e-07,
+      "loss": 0.9145,
+      "step": 6911
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6285418908966515e-07,
+      "loss": 0.8697,
+      "step": 6912
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.624829114015954e-07,
+      "loss": 1.0297,
+      "step": 6913
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.621117555411543e-07,
+      "loss": 1.3505,
+      "step": 6914
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6174072155192394e-07,
+      "loss": 4.8279,
+      "step": 6915
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.613698094774715e-07,
+      "loss": 0.7519,
+      "step": 6916
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.609990193613509e-07,
+      "loss": 0.9386,
+      "step": 6917
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.6062835124710054e-07,
+      "loss": 1.2308,
+      "step": 6918
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.6025780517824585e-07,
+      "loss": 1.2578,
+      "step": 6919
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.598873811982963e-07,
+      "loss": 1.3408,
+      "step": 6920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.595170793507482e-07,
+      "loss": 0.8953,
+      "step": 6921
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.591468996790836e-07,
+      "loss": 1.0259,
+      "step": 6922
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5877684222677006e-07,
+      "loss": 0.6202,
+      "step": 6923
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5840690703725965e-07,
+      "loss": 1.0515,
+      "step": 6924
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5803709415399216e-07,
+      "loss": 0.9609,
+      "step": 6925
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.576674036203907e-07,
+      "loss": 0.9625,
+      "step": 6926
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5729783547986617e-07,
+      "loss": 0.9294,
+      "step": 6927
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.569283897758127e-07,
+      "loss": 0.9591,
+      "step": 6928
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.565590665516133e-07,
+      "loss": 0.6671,
+      "step": 6929
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.561898658506335e-07,
+      "loss": 4.9669,
+      "step": 6930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.558207877162265e-07,
+      "loss": 0.798,
+      "step": 6931
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.554518321917293e-07,
+      "loss": 0.3776,
+      "step": 6932
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5508299932046655e-07,
+      "loss": 0.7503,
+      "step": 6933
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.547142891457464e-07,
+      "loss": 0.8741,
+      "step": 6934
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5434570171086463e-07,
+      "loss": 0.9987,
+      "step": 6935
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5397723705910027e-07,
+      "loss": 0.7232,
+      "step": 6936
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.536088952337211e-07,
+      "loss": 0.9958,
+      "step": 6937
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5324067627797744e-07,
+      "loss": 1.4185,
+      "step": 6938
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5287258023510696e-07,
+      "loss": 1.1638,
+      "step": 6939
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5250460714833166e-07,
+      "loss": 0.9835,
+      "step": 6940
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.521367570608607e-07,
+      "loss": 1.0868,
+      "step": 6941
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.51769030015887e-07,
+      "loss": 0.8485,
+      "step": 6942
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.514014260565901e-07,
+      "loss": 1.2362,
+      "step": 6943
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510339452261356e-07,
+      "loss": 0.8813,
+      "step": 6944
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5066658756767307e-07,
+      "loss": 0.8689,
+      "step": 6945
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5029935312433934e-07,
+      "loss": 1.3018,
+      "step": 6946
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4993224193925486e-07,
+      "loss": 1.29,
+      "step": 6947
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.495652540555277e-07,
+      "loss": 1.0523,
+      "step": 6948
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4919838951624933e-07,
+      "loss": 0.7968,
+      "step": 6949
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4883164836449863e-07,
+      "loss": 0.8589,
+      "step": 6950
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.484650306433389e-07,
+      "loss": 0.6609,
+      "step": 6951
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4809853639581975e-07,
+      "loss": 0.6932,
+      "step": 6952
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4773216566497504e-07,
+      "loss": 0.916,
+      "step": 6953
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.473659184938255e-07,
+      "loss": 0.7064,
+      "step": 6954
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4699979492537594e-07,
+      "loss": 0.705,
+      "step": 6955
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.466337950026183e-07,
+      "loss": 0.9455,
+      "step": 6956
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4626791876852814e-07,
+      "loss": 1.0578,
+      "step": 6957
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.459021662660682e-07,
+      "loss": 0.8358,
+      "step": 6958
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.455365375381857e-07,
+      "loss": 0.7889,
+      "step": 6959
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4517103262781414e-07,
+      "loss": 0.9533,
+      "step": 6960
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 1.2546870708465576,
+      "eval_runtime": 32.2613,
+      "eval_samples_per_second": 38.591,
+      "eval_steps_per_second": 19.311,
+      "step": 6960
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4480565157787115e-07,
+      "loss": 0.8309,
+      "step": 6961
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.444403944312613e-07,
+      "loss": 1.1417,
+      "step": 6962
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.440752612308732e-07,
+      "loss": 0.8591,
+      "step": 6963
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4371025201958256e-07,
+      "loss": 0.7682,
+      "step": 6964
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4334536684024793e-07,
+      "loss": 0.8456,
+      "step": 6965
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4298060573571713e-07,
+      "loss": 1.0005,
+      "step": 6966
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4261596874881965e-07,
+      "loss": 1.0765,
+      "step": 6967
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.422514559223731e-07,
+      "loss": 0.9345,
+      "step": 6968
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.418870672991784e-07,
+      "loss": 0.9,
+      "step": 6969
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4152280292202363e-07,
+      "loss": 0.7287,
+      "step": 6970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4115866283368093e-07,
+      "loss": 1.0336,
+      "step": 6971
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.407946470769093e-07,
+      "loss": 0.9769,
+      "step": 6972
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.404307556944512e-07,
+      "loss": 1.0006,
+      "step": 6973
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.400669887290362e-07,
+      "loss": 1.0737,
+      "step": 6974
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.397033462233793e-07,
+      "loss": 0.9135,
+      "step": 6975
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3933982822017883e-07,
+      "loss": 0.9992,
+      "step": 6976
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.389764347621207e-07,
+      "loss": 0.5135,
+      "step": 6977
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.386131658918759e-07,
+      "loss": 0.9621,
+      "step": 6978
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3825002165209913e-07,
+      "loss": 0.7173,
+      "step": 6979
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.378870020854324e-07,
+      "loss": 0.8771,
+      "step": 6980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3752410723450243e-07,
+      "loss": 1.1786,
+      "step": 6981
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3716133714192043e-07,
+      "loss": 0.887,
+      "step": 6982
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3679869185028466e-07,
+      "loss": 0.9901,
+      "step": 6983
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3643617140217674e-07,
+      "loss": 0.4793,
+      "step": 6984
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3607377584016546e-07,
+      "loss": 0.9207,
+      "step": 6985
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.357115052068034e-07,
+      "loss": 0.9965,
+      "step": 6986
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.353493595446301e-07,
+      "loss": 1.0216,
+      "step": 6987
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3498733889616804e-07,
+      "loss": 0.9301,
+      "step": 6988
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.346254433039285e-07,
+      "loss": 0.8731,
+      "step": 6989
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3426367281040455e-07,
+      "loss": 0.8104,
+      "step": 6990
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.339020274580771e-07,
+      "loss": 0.761,
+      "step": 6991
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3354050728941066e-07,
+      "loss": 0.6836,
+      "step": 6992
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.331791123468563e-07,
+      "loss": 1.2374,
+      "step": 6993
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.328178426728492e-07,
+      "loss": 1.6467,
+      "step": 6994
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3245669830981065e-07,
+      "loss": 0.8184,
+      "step": 6995
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.320956793001475e-07,
+      "loss": 1.1349,
+      "step": 6996
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.317347856862514e-07,
+      "loss": 0.7786,
+      "step": 6997
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.313740175104987e-07,
+      "loss": 0.4402,
+      "step": 6998
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.310133748152525e-07,
+      "loss": 0.774,
+      "step": 6999
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.306528576428592e-07,
+      "loss": 1.0482,
+      "step": 7000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.302924660356527e-07,
+      "loss": 0.903,
+      "step": 7001
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.299322000359499e-07,
+      "loss": 0.8251,
+      "step": 7002
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.295720596860545e-07,
+      "loss": 0.7062,
+      "step": 7003
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.292120450282552e-07,
+      "loss": 0.8488,
+      "step": 7004
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.2885215610482606e-07,
+      "loss": 1.0314,
+      "step": 7005
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.284923929580253e-07,
+      "loss": 0.9207,
+      "step": 7006
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.281327556300978e-07,
+      "loss": 1.3392,
+      "step": 7007
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.277732441632722e-07,
+      "loss": 0.8139,
+      "step": 7008
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.274138585997642e-07,
+      "loss": 0.9496,
+      "step": 7009
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.270545989817725e-07,
+      "loss": 1.1424,
+      "step": 7010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2669546535148277e-07,
+      "loss": 0.7149,
+      "step": 7011
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2633645775106597e-07,
+      "loss": 1.0945,
+      "step": 7012
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2597757622267623e-07,
+      "loss": 0.7971,
+      "step": 7013
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2561882080845566e-07,
+      "loss": 1.0284,
+      "step": 7014
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.252601915505288e-07,
+      "loss": 1.1977,
+      "step": 7015
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.249016884910079e-07,
+      "loss": 0.9251,
+      "step": 7016
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.245433116719876e-07,
+      "loss": 1.31,
+      "step": 7017
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2418506113555144e-07,
+      "loss": 0.9515,
+      "step": 7018
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2382693692376435e-07,
+      "loss": 0.9706,
+      "step": 7019
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2346893907867934e-07,
+      "loss": 0.8645,
+      "step": 7020
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.231110676423321e-07,
+      "loss": 0.8859,
+      "step": 7021
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.227533226567458e-07,
+      "loss": 0.6999,
+      "step": 7022
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2239570416392677e-07,
+      "loss": 1.2721,
+      "step": 7023
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2203821220586835e-07,
+      "loss": 0.8732,
+      "step": 7024
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.216808468245464e-07,
+      "loss": 0.8686,
+      "step": 7025
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2132360806192576e-07,
+      "loss": 1.135,
+      "step": 7026
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2096649595995264e-07,
+      "loss": 0.7927,
+      "step": 7027
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2060951056056093e-07,
+      "loss": 0.9046,
+      "step": 7028
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2025265190566775e-07,
+      "loss": 0.9919,
+      "step": 7029
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.198959200371773e-07,
+      "loss": 1.087,
+      "step": 7030
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1953931499697674e-07,
+      "loss": 0.7309,
+      "step": 7031
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1918283682694004e-07,
+      "loss": 0.9941,
+      "step": 7032
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1882648556892557e-07,
+      "loss": 0.8628,
+      "step": 7033
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.184702612647773e-07,
+      "loss": 1.0564,
+      "step": 7034
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1811416395632343e-07,
+      "loss": 0.9334,
+      "step": 7035
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.17758193685378e-07,
+      "loss": 0.9389,
+      "step": 7036
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.174023504937395e-07,
+      "loss": 1.239,
+      "step": 7037
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1704663442319244e-07,
+      "loss": 1.1324,
+      "step": 7038
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1669104551550503e-07,
+      "loss": 0.9081,
+      "step": 7039
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.163355838124316e-07,
+      "loss": 1.1504,
+      "step": 7040
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.159802493557121e-07,
+      "loss": 0.8452,
+      "step": 7041
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.156250421870694e-07,
+      "loss": 0.8259,
+      "step": 7042
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.152699623482139e-07,
+      "loss": 1.2047,
+      "step": 7043
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1491500988083884e-07,
+      "loss": 0.8886,
+      "step": 7044
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1456018482662454e-07,
+      "loss": 0.9526,
+      "step": 7045
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1420548722723463e-07,
+      "loss": 1.4035,
+      "step": 7046
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1385091712431856e-07,
+      "loss": 0.9971,
+      "step": 7047
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1349647455951107e-07,
+      "loss": 0.8782,
+      "step": 7048
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.13142159574432e-07,
+      "loss": 0.8282,
+      "step": 7049
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.127879722106848e-07,
+      "loss": 0.7883,
+      "step": 7050
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1243391250986e-07,
+      "loss": 0.7581,
+      "step": 7051
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1207998051353135e-07,
+      "loss": 1.1007,
+      "step": 7052
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1172617626325895e-07,
+      "loss": 0.9232,
+      "step": 7053
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1137249980058634e-07,
+      "loss": 0.5891,
+      "step": 7054
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1101895116704383e-07,
+      "loss": 0.9319,
+      "step": 7055
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.106655304041458e-07,
+      "loss": 1.1118,
+      "step": 7056
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.103122375533919e-07,
+      "loss": 0.817,
+      "step": 7057
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0995907265626606e-07,
+      "loss": 1.2434,
+      "step": 7058
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.096060357542384e-07,
+      "loss": 0.7479,
+      "step": 7059
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.092531268887625e-07,
+      "loss": 0.8913,
+      "step": 7060
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0890034610127856e-07,
+      "loss": 0.7498,
+      "step": 7061
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.085476934332096e-07,
+      "loss": 1.1044,
+      "step": 7062
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0819516892596653e-07,
+      "loss": 1.0082,
+      "step": 7063
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.078427726209425e-07,
+      "loss": 0.5687,
+      "step": 7064
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0749050455951755e-07,
+      "loss": 1.4371,
+      "step": 7065
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.071383647830547e-07,
+      "loss": 0.6716,
+      "step": 7066
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0678635333290413e-07,
+      "loss": 0.97,
+      "step": 7067
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0643447025039875e-07,
+      "loss": 0.5756,
+      "step": 7068
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.060827155768585e-07,
+      "loss": 0.8065,
+      "step": 7069
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0573108935358633e-07,
+      "loss": 0.8872,
+      "step": 7070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0537959162187136e-07,
+      "loss": 0.8898,
+      "step": 7071
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0502822242298764e-07,
+      "loss": 1.2115,
+      "step": 7072
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0467698179819314e-07,
+      "loss": 1.0875,
+      "step": 7073
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.04325869788732e-07,
+      "loss": 1.0968,
+      "step": 7074
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0397488643583156e-07,
+      "loss": 1.2735,
+      "step": 7075
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.0362403178070595e-07,
+      "loss": 0.7674,
+      "step": 7076
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.03273305864553e-07,
+      "loss": 0.888,
+      "step": 7077
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.029227087285562e-07,
+      "loss": 1.2581,
+      "step": 7078
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.025722404138827e-07,
+      "loss": 0.6221,
+      "step": 7079
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.022219009616862e-07,
+      "loss": 1.1298,
+      "step": 7080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.018716904131035e-07,
+      "loss": 0.8424,
+      "step": 7081
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.015216088092577e-07,
+      "loss": 0.5938,
+      "step": 7082
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.011716561912557e-07,
+      "loss": 0.4395,
+      "step": 7083
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.008218326001899e-07,
+      "loss": 1.2372,
+      "step": 7084
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.004721380771375e-07,
+      "loss": 0.7909,
+      "step": 7085
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.0012257266316086e-07,
+      "loss": 1.1099,
+      "step": 7086
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9977313639930603e-07,
+      "loss": 1.0076,
+      "step": 7087
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.994238293266051e-07,
+      "loss": 1.0125,
+      "step": 7088
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9907465148607396e-07,
+      "loss": 1.1196,
+      "step": 7089
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.987256029187147e-07,
+      "loss": 0.7196,
+      "step": 7090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9837668366551245e-07,
+      "loss": 0.5466,
+      "step": 7091
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.980278937674385e-07,
+      "loss": 1.4137,
+      "step": 7092
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9767923326544855e-07,
+      "loss": 1.0832,
+      "step": 7093
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9733070220048383e-07,
+      "loss": 0.8553,
+      "step": 7094
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9698230061346835e-07,
+      "loss": 0.9838,
+      "step": 7095
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.966340285453135e-07,
+      "loss": 1.1555,
+      "step": 7096
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.962858860369131e-07,
+      "loss": 1.0464,
+      "step": 7097
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.959378731291477e-07,
+      "loss": 1.006,
+      "step": 7098
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9558998986288085e-07,
+      "loss": 0.9583,
+      "step": 7099
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.952422362789624e-07,
+      "loss": 0.8246,
+      "step": 7100
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9489461241822676e-07,
+      "loss": 0.7815,
+      "step": 7101
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.945471183214917e-07,
+      "loss": 0.7236,
+      "step": 7102
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9419975402956153e-07,
+      "loss": 0.9182,
+      "step": 7103
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9385251958322467e-07,
+      "loss": 0.5944,
+      "step": 7104
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.935054150232534e-07,
+      "loss": 0.9532,
+      "step": 7105
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.9315844039040645e-07,
+      "loss": 1.1119,
+      "step": 7106
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.9281159572542543e-07,
+      "loss": 0.7649,
+      "step": 7107
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.9246488106903825e-07,
+      "loss": 1.0248,
+      "step": 7108
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.921182964619571e-07,
+      "loss": 0.8789,
+      "step": 7109
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.917718419448781e-07,
+      "loss": 1.2343,
+      "step": 7110
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.914255175584836e-07,
+      "loss": 1.0363,
+      "step": 7111
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.910793233434387e-07,
+      "loss": 0.9098,
+      "step": 7112
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.907332593403954e-07,
+      "loss": 1.1203,
+      "step": 7113
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.9038732558998796e-07,
+      "loss": 0.7445,
+      "step": 7114
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.900415221328384e-07,
+      "loss": 0.9516,
+      "step": 7115
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8969584900955054e-07,
+      "loss": 1.0394,
+      "step": 7116
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8935030626071486e-07,
+      "loss": 1.0695,
+      "step": 7117
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8900489392690505e-07,
+      "loss": 0.9421,
+      "step": 7118
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.88659612048681e-07,
+      "loss": 0.8724,
+      "step": 7119
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8831446066658554e-07,
+      "loss": 0.8568,
+      "step": 7120
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8796943982114823e-07,
+      "loss": 0.8138,
+      "step": 7121
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8762454955288086e-07,
+      "loss": 0.9169,
+      "step": 7122
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.872797899022828e-07,
+      "loss": 0.7437,
+      "step": 7123
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.869351609098352e-07,
+      "loss": 0.8132,
+      "step": 7124
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8659066261600623e-07,
+      "loss": 0.7215,
+      "step": 7125
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.862462950612468e-07,
+      "loss": 0.8908,
+      "step": 7126
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.8590205828599415e-07,
+      "loss": 0.8629,
+      "step": 7127
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.855579523306683e-07,
+      "loss": 0.8665,
+      "step": 7128
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8521397723567577e-07,
+      "loss": 1.0349,
+      "step": 7129
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8487013304140645e-07,
+      "loss": 0.9945,
+      "step": 7130
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.845264197882361e-07,
+      "loss": 1.1145,
+      "step": 7131
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8418283751652345e-07,
+      "loss": 0.8219,
+      "step": 7132
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8383938626661333e-07,
+      "loss": 0.586,
+      "step": 7133
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8349606607883376e-07,
+      "loss": 0.9367,
+      "step": 7134
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8315287699349923e-07,
+      "loss": 0.8355,
+      "step": 7135
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8280981905090684e-07,
+      "loss": 0.8473,
+      "step": 7136
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.824668922913396e-07,
+      "loss": 1.1131,
+      "step": 7137
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.821240967550652e-07,
+      "loss": 1.2671,
+      "step": 7138
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8178143248233473e-07,
+      "loss": 1.0078,
+      "step": 7139
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.814388995133854e-07,
+      "loss": 1.0689,
+      "step": 7140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8109649788843723e-07,
+      "loss": 0.7297,
+      "step": 7141
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.807542276476968e-07,
+      "loss": 0.5562,
+      "step": 7142
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.8041208883135353e-07,
+      "loss": 0.8843,
+      "step": 7143
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.800700814795822e-07,
+      "loss": 0.7336,
+      "step": 7144
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.797282056325424e-07,
+      "loss": 0.9311,
+      "step": 7145
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.7938646133037846e-07,
+      "loss": 1.0979,
+      "step": 7146
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.790448486132176e-07,
+      "loss": 0.9622,
+      "step": 7147
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.787033675211739e-07,
+      "loss": 1.0827,
+      "step": 7148
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.783620180943438e-07,
+      "loss": 1.0876,
+      "step": 7149
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.780208003728103e-07,
+      "loss": 0.8863,
+      "step": 7150
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.776797143966388e-07,
+      "loss": 0.8329,
+      "step": 7151
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.773387602058818e-07,
+      "loss": 0.8935,
+      "step": 7152
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.769979378405739e-07,
+      "loss": 0.8353,
+      "step": 7153
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7665724734073605e-07,
+      "loss": 0.9788,
+      "step": 7154
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.763166887463719e-07,
+      "loss": 0.9135,
+      "step": 7155
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.759762620974716e-07,
+      "loss": 0.9134,
+      "step": 7156
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.756359674340079e-07,
+      "loss": 0.863,
+      "step": 7157
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7529580479594e-07,
+      "loss": 1.0555,
+      "step": 7158
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.749557742232091e-07,
+      "loss": 1.1639,
+      "step": 7159
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7461587575574426e-07,
+      "loss": 0.7876,
+      "step": 7160
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7427610943345567e-07,
+      "loss": 1.0751,
+      "step": 7161
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7393647529624035e-07,
+      "loss": 0.9068,
+      "step": 7162
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.735969733839782e-07,
+      "loss": 0.7223,
+      "step": 7163
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7325760373653494e-07,
+      "loss": 1.0147,
+      "step": 7164
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.729183663937596e-07,
+      "loss": 1.0347,
+      "step": 7165
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.725792613954863e-07,
+      "loss": 1.2146,
+      "step": 7166
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.722402887815341e-07,
+      "loss": 1.032,
+      "step": 7167
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.719014485917051e-07,
+      "loss": 0.648,
+      "step": 7168
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7156274086578743e-07,
+      "loss": 1.0672,
+      "step": 7169
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.712241656435523e-07,
+      "loss": 0.815,
+      "step": 7170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7088572296475676e-07,
+      "loss": 1.1196,
+      "step": 7171
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.705474128691405e-07,
+      "loss": 1.2071,
+      "step": 7172
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.7020923539642916e-07,
+      "loss": 0.9434,
+      "step": 7173
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.698711905863324e-07,
+      "loss": 0.8667,
+      "step": 7174
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.6953327847854464e-07,
+      "loss": 1.1001,
+      "step": 7175
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.691954991127434e-07,
+      "loss": 0.8655,
+      "step": 7176
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6885785252859237e-07,
+      "loss": 0.8112,
+      "step": 7177
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.685203387657378e-07,
+      "loss": 0.8756,
+      "step": 7178
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.681829578638125e-07,
+      "loss": 0.6979,
+      "step": 7179
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6784570986243127e-07,
+      "loss": 1.0362,
+      "step": 7180
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6750859480119525e-07,
+      "loss": 0.9641,
+      "step": 7181
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.671716127196891e-07,
+      "loss": 1.2524,
+      "step": 7182
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6683476365748257e-07,
+      "loss": 0.9888,
+      "step": 7183
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.664980476541283e-07,
+      "loss": 1.1268,
+      "step": 7184
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.661614647491652e-07,
+      "loss": 1.0001,
+      "step": 7185
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.658250149821146e-07,
+      "loss": 1.0148,
+      "step": 7186
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.654886983924842e-07,
+      "loss": 0.8452,
+      "step": 7187
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6515251501976406e-07,
+      "loss": 0.8587,
+      "step": 7188
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.648164649034303e-07,
+      "loss": 0.7215,
+      "step": 7189
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.644805480829424e-07,
+      "loss": 1.2756,
+      "step": 7190
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6414476459774496e-07,
+      "loss": 1.2135,
+      "step": 7191
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6380911448726586e-07,
+      "loss": 0.9795,
+      "step": 7192
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6347359779091843e-07,
+      "loss": 1.0023,
+      "step": 7193
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6313821454809907e-07,
+      "loss": 0.9191,
+      "step": 7194
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.628029647981901e-07,
+      "loss": 1.0902,
+      "step": 7195
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.624678485805566e-07,
+      "loss": 1.065,
+      "step": 7196
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.621328659345488e-07,
+      "loss": 0.8873,
+      "step": 7197
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.617980168995018e-07,
+      "loss": 0.9202,
+      "step": 7198
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.614633015147335e-07,
+      "loss": 0.7443,
+      "step": 7199
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.611287198195473e-07,
+      "loss": 0.71,
+      "step": 7200
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.607942718532308e-07,
+      "loss": 1.0092,
+      "step": 7201
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.60459957655055e-07,
+      "loss": 1.004,
+      "step": 7202
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.6012577726427625e-07,
+      "loss": 0.7442,
+      "step": 7203
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.597917307201352e-07,
+      "loss": 0.9006,
+      "step": 7204
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.5945781806185547e-07,
+      "loss": 0.9727,
+      "step": 7205
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.591240393286466e-07,
+      "loss": 0.7197,
+      "step": 7206
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.58790394559701e-07,
+      "loss": 1.0828,
+      "step": 7207
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.5845688379419685e-07,
+      "loss": 0.727,
+      "step": 7208
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.581235070712947e-07,
+      "loss": 0.8528,
+      "step": 7209
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.577902644301413e-07,
+      "loss": 0.7785,
+      "step": 7210
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.574571559098656e-07,
+      "loss": 0.9216,
+      "step": 7211
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.571241815495837e-07,
+      "loss": 0.7535,
+      "step": 7212
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.567913413883928e-07,
+      "loss": 0.7824,
+      "step": 7213
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.5645863546537666e-07,
+      "loss": 0.8234,
+      "step": 7214
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.561260638196015e-07,
+      "loss": 1.1688,
+      "step": 7215
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.557936264901194e-07,
+      "loss": 1.2359,
+      "step": 7216
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.554613235159652e-07,
+      "loss": 0.9709,
+      "step": 7217
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.5512915493615916e-07,
+      "loss": 0.7463,
+      "step": 7218
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.5479712078970524e-07,
+      "loss": 0.9461,
+      "step": 7219
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.544652211155919e-07,
+      "loss": 0.9293,
+      "step": 7220
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.541334559527908e-07,
+      "loss": 0.839,
+      "step": 7221
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.538018253402596e-07,
+      "loss": 1.1413,
+      "step": 7222
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.534703293169379e-07,
+      "loss": 1.096,
+      "step": 7223
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.53138967921752e-07,
+      "loss": 1.1495,
+      "step": 7224
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.528077411936099e-07,
+      "loss": 0.9751,
+      "step": 7225
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.524766491714056e-07,
+      "loss": 0.8088,
+      "step": 7226
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.521456918940171e-07,
+      "loss": 0.7884,
+      "step": 7227
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.518148694003053e-07,
+      "loss": 0.9742,
+      "step": 7228
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.514841817291164e-07,
+      "loss": 0.7222,
+      "step": 7229
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.511536289192812e-07,
+      "loss": 0.9019,
+      "step": 7230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.50823211009613e-07,
+      "loss": 0.8985,
+      "step": 7231
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.5049292803891115e-07,
+      "loss": 0.9393,
+      "step": 7232
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.501627800459571e-07,
+      "loss": 1.2988,
+      "step": 7233
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4983276706951836e-07,
+      "loss": 0.9556,
+      "step": 7234
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4950288914834607e-07,
+      "loss": 1.1742,
+      "step": 7235
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4917314632117427e-07,
+      "loss": 1.1775,
+      "step": 7236
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4884353862672313e-07,
+      "loss": 1.0488,
+      "step": 7237
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.485140661036952e-07,
+      "loss": 1.2657,
+      "step": 7238
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.481847287907786e-07,
+      "loss": 0.8859,
+      "step": 7239
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47855526726644e-07,
+      "loss": 1.0517,
+      "step": 7240
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4752645994994746e-07,
+      "loss": 1.1472,
+      "step": 7241
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471975284993288e-07,
+      "loss": 0.6131,
+      "step": 7242
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4686873241341236e-07,
+      "loss": 1.0728,
+      "step": 7243
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4654007173080523e-07,
+      "loss": 0.9697,
+      "step": 7244
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.462115464901004e-07,
+      "loss": 0.8248,
+      "step": 7245
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4588315672987324e-07,
+      "loss": 0.6841,
+      "step": 7246
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4555490248868477e-07,
+      "loss": 1.0568,
+      "step": 7247
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.452267838050781e-07,
+      "loss": 1.3941,
+      "step": 7248
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4489880071758346e-07,
+      "loss": 0.7144,
+      "step": 7249
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.44570953264712e-07,
+      "loss": 0.9946,
+      "step": 7250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.442432414849611e-07,
+      "loss": 1.1179,
+      "step": 7251
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4391566541681085e-07,
+      "loss": 1.0678,
+      "step": 7252
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.435882250987266e-07,
+      "loss": 0.9199,
+      "step": 7253
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.432609205691563e-07,
+      "loss": 0.8929,
+      "step": 7254
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4293375186653367e-07,
+      "loss": 0.7139,
+      "step": 7255
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4260671902927455e-07,
+      "loss": 0.9888,
+      "step": 7256
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.422798220957813e-07,
+      "loss": 1.187,
+      "step": 7257
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.419530611044376e-07,
+      "loss": 1.174,
+      "step": 7258
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4162643609361356e-07,
+      "loss": 1.1774,
+      "step": 7259
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4129994710166136e-07,
+      "loss": 1.0128,
+      "step": 7260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4097359416691867e-07,
+      "loss": 1.069,
+      "step": 7261
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4064737732770597e-07,
+      "loss": 0.8847,
+      "step": 7262
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4032129662232883e-07,
+      "loss": 0.9272,
+      "step": 7263
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.3999535208907674e-07,
+      "loss": 1.1425,
+      "step": 7264
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.3966954376622193e-07,
+      "loss": 0.8544,
+      "step": 7265
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.3934387169202267e-07,
+      "loss": 1.1999,
+      "step": 7266
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.3901833590471904e-07,
+      "loss": 0.8837,
+      "step": 7267
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3869293644253706e-07,
+      "loss": 0.898,
+      "step": 7268
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3836767334368506e-07,
+      "loss": 0.8582,
+      "step": 7269
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.380425466463565e-07,
+      "loss": 1.0364,
+      "step": 7270
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3771755638872874e-07,
+      "loss": 0.9657,
+      "step": 7271
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.373927026089632e-07,
+      "loss": 1.0329,
+      "step": 7272
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.37067985345204e-07,
+      "loss": 0.8765,
+      "step": 7273
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.367434046355811e-07,
+      "loss": 0.8196,
+      "step": 7274
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3641896051820674e-07,
+      "loss": 1.2308,
+      "step": 7275
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3609465303117873e-07,
+      "loss": 0.6798,
+      "step": 7276
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.35770482212577e-07,
+      "loss": 0.9585,
+      "step": 7277
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.354464481004669e-07,
+      "loss": 0.6663,
+      "step": 7278
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.351225507328974e-07,
+      "loss": 1.2012,
+      "step": 7279
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.347987901479016e-07,
+      "loss": 1.1851,
+      "step": 7280
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.344751663834953e-07,
+      "loss": 1.2644,
+      "step": 7281
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3415167947767996e-07,
+      "loss": 0.9897,
+      "step": 7282
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.338283294684395e-07,
+      "loss": 0.855,
+      "step": 7283
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.335051163937431e-07,
+      "loss": 1.3566,
+      "step": 7284
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3318204029154187e-07,
+      "loss": 1.2254,
+      "step": 7285
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3285910119977415e-07,
+      "loss": 0.8776,
+      "step": 7286
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.325362991563584e-07,
+      "loss": 0.9406,
+      "step": 7287
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.322136341992002e-07,
+      "loss": 1.2794,
+      "step": 7288
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3189110636618635e-07,
+      "loss": 1.0418,
+      "step": 7289
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3156871569518974e-07,
+      "loss": 1.0389,
+      "step": 7290
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.3124646222406545e-07,
+      "loss": 0.9033,
+      "step": 7291
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.3092434599065404e-07,
+      "loss": 0.7777,
+      "step": 7292
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.306023670327782e-07,
+      "loss": 1.1271,
+      "step": 7293
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.3028052538824604e-07,
+      "loss": 0.7942,
+      "step": 7294
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.299588210948492e-07,
+      "loss": 0.766,
+      "step": 7295
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.296372541903621e-07,
+      "loss": 0.9764,
+      "step": 7296
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2931582471254474e-07,
+      "loss": 0.8541,
+      "step": 7297
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2899453269913937e-07,
+      "loss": 0.8617,
+      "step": 7298
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2867337818787304e-07,
+      "loss": 0.8897,
+      "step": 7299
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2835236121645666e-07,
+      "loss": 0.7878,
+      "step": 7300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.28031481822585e-07,
+      "loss": 0.8284,
+      "step": 7301
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.277107400439359e-07,
+      "loss": 0.9927,
+      "step": 7302
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2739013591817205e-07,
+      "loss": 1.078,
+      "step": 7303
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2706966948293885e-07,
+      "loss": 0.9296,
+      "step": 7304
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.267493407758672e-07,
+      "loss": 1.0628,
+      "step": 7305
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2642914983456984e-07,
+      "loss": 1.0299,
+      "step": 7306
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2610909669664516e-07,
+      "loss": 1.0436,
+      "step": 7307
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2578918139967334e-07,
+      "loss": 0.7752,
+      "step": 7308
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.254694039812211e-07,
+      "loss": 1.0819,
+      "step": 7309
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2514976447883626e-07,
+      "loss": 0.8634,
+      "step": 7310
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.248302629300525e-07,
+      "loss": 0.5487,
+      "step": 7311
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2451089937238556e-07,
+      "loss": 0.977,
+      "step": 7312
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2419167384333663e-07,
+      "loss": 0.5685,
+      "step": 7313
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.238725863803891e-07,
+      "loss": 0.8352,
+      "step": 7314
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2355363702101136e-07,
+      "loss": 0.9634,
+      "step": 7315
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2323482580265525e-07,
+      "loss": 0.978,
+      "step": 7316
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.229161527627566e-07,
+      "loss": 6.0708,
+      "step": 7317
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2259761793873403e-07,
+      "loss": 1.1588,
+      "step": 7318
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.222792213679913e-07,
+      "loss": 0.7984,
+      "step": 7319
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2196096308791467e-07,
+      "loss": 0.8839,
+      "step": 7320
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2164284313587536e-07,
+      "loss": 0.9659,
+      "step": 7321
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2132486154922694e-07,
+      "loss": 0.694,
+      "step": 7322
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.21007018365308e-07,
+      "loss": 0.9098,
+      "step": 7323
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2068931362144075e-07,
+      "loss": 0.7856,
+      "step": 7324
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2037174735493024e-07,
+      "loss": 1.0832,
+      "step": 7325
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2005431960306597e-07,
+      "loss": 0.8401,
+      "step": 7326
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.1973703040312146e-07,
+      "loss": 1.0849,
+      "step": 7327
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.1941987979235276e-07,
+      "loss": 1.0232,
+      "step": 7328
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.191028678080014e-07,
+      "loss": 0.9755,
+      "step": 7329
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.1878599448729063e-07,
+      "loss": 0.9081,
+      "step": 7330
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.184692598674288e-07,
+      "loss": 1.1505,
+      "step": 7331
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.181526639856084e-07,
+      "loss": 0.7928,
+      "step": 7332
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.1783620687900346e-07,
+      "loss": 0.7815,
+      "step": 7333
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.175198885847744e-07,
+      "loss": 0.7952,
+      "step": 7334
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.1720370914006295e-07,
+      "loss": 0.84,
+      "step": 7335
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.1688766858199654e-07,
+      "loss": 0.7622,
+      "step": 7336
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.165717669476841e-07,
+      "loss": 1.0231,
+      "step": 7337
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.162560042742213e-07,
+      "loss": 0.7887,
+      "step": 7338
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1594038059868417e-07,
+      "loss": 1.0479,
+      "step": 7339
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.156248959581353e-07,
+      "loss": 0.82,
+      "step": 7340
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.153095503896182e-07,
+      "loss": 0.8513,
+      "step": 7341
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1499434393016264e-07,
+      "loss": 1.074,
+      "step": 7342
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1467927661678004e-07,
+      "loss": 0.8624,
+      "step": 7343
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.143643484864672e-07,
+      "loss": 0.8758,
+      "step": 7344
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1404955957620213e-07,
+      "loss": 1.0432,
+      "step": 7345
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.137349099229502e-07,
+      "loss": 0.7827,
+      "step": 7346
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1342039956365657e-07,
+      "loss": 0.986,
+      "step": 7347
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.13106028535253e-07,
+      "loss": 0.9081,
+      "step": 7348
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1279179687465256e-07,
+      "loss": 0.829,
+      "step": 7349
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.12477704618754e-07,
+      "loss": 0.987,
+      "step": 7350
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1216375180443795e-07,
+      "loss": 0.8925,
+      "step": 7351
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.118499384685697e-07,
+      "loss": 1.0997,
+      "step": 7352
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.115362646479982e-07,
+      "loss": 0.8787,
+      "step": 7353
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1122273037955595e-07,
+      "loss": 1.1614,
+      "step": 7354
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1090933570005806e-07,
+      "loss": 0.9957,
+      "step": 7355
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.10596080646305e-07,
+      "loss": 0.7799,
+      "step": 7356
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.10282965255079e-07,
+      "loss": 0.5159,
+      "step": 7357
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.0996998956314745e-07,
+      "loss": 1.1068,
+      "step": 7358
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.096571536072601e-07,
+      "loss": 0.9753,
+      "step": 7359
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.093444574241511e-07,
+      "loss": 1.2716,
+      "step": 7360
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0903190105053845e-07,
+      "loss": 0.7074,
+      "step": 7361
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.087194845231224e-07,
+      "loss": 0.9847,
+      "step": 7362
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.084072078785885e-07,
+      "loss": 0.9282,
+      "step": 7363
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.08095071153604e-07,
+      "loss": 0.7394,
+      "step": 7364
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.077830743848217e-07,
+      "loss": 0.5314,
+      "step": 7365
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.07471217608876e-07,
+      "loss": 0.995,
+      "step": 7366
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0715950086238647e-07,
+      "loss": 0.9944,
+      "step": 7367
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0684792418195524e-07,
+      "loss": 0.6678,
+      "step": 7368
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.065364876041691e-07,
+      "loss": 0.7653,
+      "step": 7369
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.062251911655966e-07,
+      "loss": 0.8536,
+      "step": 7370
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.059140349027919e-07,
+      "loss": 0.5791,
+      "step": 7371
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.056030188522908e-07,
+      "loss": 0.7742,
+      "step": 7372
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0529214305061424e-07,
+      "loss": 1.085,
+      "step": 7373
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0498140753426523e-07,
+      "loss": 0.9201,
+      "step": 7374
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.046708123397314e-07,
+      "loss": 1.0111,
+      "step": 7375
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.043603575034835e-07,
+      "loss": 0.7976,
+      "step": 7376
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0405004306197647e-07,
+      "loss": 0.7923,
+      "step": 7377
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0373986905164714e-07,
+      "loss": 0.9266,
+      "step": 7378
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0342983550891767e-07,
+      "loss": 0.7963,
+      "step": 7379
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0311994247019215e-07,
+      "loss": 0.8669,
+      "step": 7380
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0281018997185976e-07,
+      "loss": 0.8425,
+      "step": 7381
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.0250057805029115e-07,
+      "loss": 0.8663,
+      "step": 7382
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.021911067418431e-07,
+      "loss": 0.7914,
+      "step": 7383
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.0188177608285357e-07,
+      "loss": 0.8502,
+      "step": 7384
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.0157258610964516e-07,
+      "loss": 1.0957,
+      "step": 7385
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.0126353685852343e-07,
+      "loss": 0.8717,
+      "step": 7386
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.009546283657781e-07,
+      "loss": 0.8786,
+      "step": 7387
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.0064586066768113e-07,
+      "loss": 1.1177,
+      "step": 7388
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.0033723380048973e-07,
+      "loss": 0.8436,
+      "step": 7389
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.000287478004424e-07,
+      "loss": 0.7981,
+      "step": 7390
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.997204027037631e-07,
+      "loss": 1.1309,
+      "step": 7391
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9941219854665864e-07,
+      "loss": 0.8661,
+      "step": 7392
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.991041353653182e-07,
+      "loss": 0.7818,
+      "step": 7393
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.987962131959161e-07,
+      "loss": 1.0278,
+      "step": 7394
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.984884320746085e-07,
+      "loss": 0.7167,
+      "step": 7395
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.981807920375366e-07,
+      "loss": 0.9541,
+      "step": 7396
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9787329312082267e-07,
+      "loss": 1.053,
+      "step": 7397
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9756593536057595e-07,
+      "loss": 0.7128,
+      "step": 7398
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9725871879288574e-07,
+      "loss": 0.924,
+      "step": 7399
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9695164345382697e-07,
+      "loss": 0.803,
+      "step": 7400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9664470937945617e-07,
+      "loss": 0.8193,
+      "step": 7401
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9633791660581525e-07,
+      "loss": 0.5434,
+      "step": 7402
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9603126516892756e-07,
+      "loss": 0.8321,
+      "step": 7403
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.957247551048011e-07,
+      "loss": 0.9652,
+      "step": 7404
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.954183864494274e-07,
+      "loss": 1.2945,
+      "step": 7405
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.9511215923878097e-07,
+      "loss": 1.0169,
+      "step": 7406
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.9480607350881907e-07,
+      "loss": 1.1074,
+      "step": 7407
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.945001292954837e-07,
+      "loss": 1.102,
+      "step": 7408
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.9419432663469887e-07,
+      "loss": 0.9127,
+      "step": 7409
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.938886655623733e-07,
+      "loss": 1.0111,
+      "step": 7410
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.935831461143978e-07,
+      "loss": 0.9784,
+      "step": 7411
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.932777683266473e-07,
+      "loss": 1.0001,
+      "step": 7412
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.9297253223498005e-07,
+      "loss": 1.0063,
+      "step": 7413
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.926674378752382e-07,
+      "loss": 1.1471,
+      "step": 7414
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.923624852832454e-07,
+      "loss": 1.3365,
+      "step": 7415
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.92057674494811e-07,
+      "loss": 0.9989,
+      "step": 7416
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.9175300554572564e-07,
+      "loss": 0.6455,
+      "step": 7417
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.9144847847176513e-07,
+      "loss": 0.8849,
+      "step": 7418
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.9114409330868705e-07,
+      "loss": 1.3811,
+      "step": 7419
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.90839850092233e-07,
+      "loss": 0.8489,
+      "step": 7420
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.9053574885812875e-07,
+      "loss": 1.1506,
+      "step": 7421
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.902317896420816e-07,
+      "loss": 0.815,
+      "step": 7422
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.899279724797837e-07,
+      "loss": 0.8662,
+      "step": 7423
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.8962429740690954e-07,
+      "loss": 0.7802,
+      "step": 7424
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.893207644591176e-07,
+      "loss": 0.9741,
+      "step": 7425
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.890173736720497e-07,
+      "loss": 0.8853,
+      "step": 7426
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.8871412508132996e-07,
+      "loss": 0.5437,
+      "step": 7427
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.884110187225669e-07,
+      "loss": 1.0144,
+      "step": 7428
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.8810805463135246e-07,
+      "loss": 0.8941,
+      "step": 7429
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.8780523284326047e-07,
+      "loss": 0.8556,
+      "step": 7430
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.8750255339384983e-07,
+      "loss": 0.6888,
+      "step": 7431
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.872000163186611e-07,
+      "loss": 0.7589,
+      "step": 7432
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.868976216532194e-07,
+      "loss": 1.2277,
+      "step": 7433
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.865953694330317e-07,
+      "loss": 0.8952,
+      "step": 7434
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.862932596935906e-07,
+      "loss": 0.9484,
+      "step": 7435
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.859912924703693e-07,
+      "loss": 0.7182,
+      "step": 7436
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.856894677988265e-07,
+      "loss": 0.718,
+      "step": 7437
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.853877857144021e-07,
+      "loss": 1.1217,
+      "step": 7438
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.850862462525212e-07,
+      "loss": 0.7071,
+      "step": 7439
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.8478484944859034e-07,
+      "loss": 0.7888,
+      "step": 7440
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.844835953380012e-07,
+      "loss": 0.9426,
+      "step": 7441
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.8418248395612635e-07,
+      "loss": 1.0965,
+      "step": 7442
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.838815153383247e-07,
+      "loss": 1.0606,
+      "step": 7443
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.835806895199354e-07,
+      "loss": 1.0091,
+      "step": 7444
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.832800065362832e-07,
+      "loss": 0.6676,
+      "step": 7445
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.829794664226737e-07,
+      "loss": 0.9129,
+      "step": 7446
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.826790692143981e-07,
+      "loss": 0.8108,
+      "step": 7447
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.8237881494672873e-07,
+      "loss": 0.7962,
+      "step": 7448
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.820787036549228e-07,
+      "loss": 1.1456,
+      "step": 7449
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.8177873537422036e-07,
+      "loss": 1.0475,
+      "step": 7450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.8147891013984357e-07,
+      "loss": 1.2534,
+      "step": 7451
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.811792279869989e-07,
+      "loss": 0.7828,
+      "step": 7452
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.808796889508761e-07,
+      "loss": 0.8364,
+      "step": 7453
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.805802930666471e-07,
+      "loss": 1.0615,
+      "step": 7454
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.802810403694684e-07,
+      "loss": 0.913,
+      "step": 7455
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7998193089447797e-07,
+      "loss": 0.9898,
+      "step": 7456
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7968296467679863e-07,
+      "loss": 0.8838,
+      "step": 7457
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7938414175153564e-07,
+      "loss": 0.9753,
+      "step": 7458
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7908546215377724e-07,
+      "loss": 1.1546,
+      "step": 7459
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.787869259185953e-07,
+      "loss": 0.845,
+      "step": 7460
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.784885330810443e-07,
+      "loss": 0.8364,
+      "step": 7461
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.781902836761625e-07,
+      "loss": 0.7679,
+      "step": 7462
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.778921777389708e-07,
+      "loss": 0.9797,
+      "step": 7463
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.775942153044735e-07,
+      "loss": 0.8044,
+      "step": 7464
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7729639640765825e-07,
+      "loss": 1.1904,
+      "step": 7465
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7699872108349577e-07,
+      "loss": 0.6366,
+      "step": 7466
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7670118936693913e-07,
+      "loss": 1.0162,
+      "step": 7467
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7640380129292595e-07,
+      "loss": 1.1685,
+      "step": 7468
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.761065568963756e-07,
+      "loss": 0.9234,
+      "step": 7469
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.758094562121917e-07,
+      "loss": 1.0015,
+      "step": 7470
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7551249927525947e-07,
+      "loss": 0.9087,
+      "step": 7471
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.752156861204499e-07,
+      "loss": 1.3042,
+      "step": 7472
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.74919016782614e-07,
+      "loss": 0.9274,
+      "step": 7473
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.746224912965886e-07,
+      "loss": 0.769,
+      "step": 7474
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.7432610969719125e-07,
+      "loss": 1.0215,
+      "step": 7475
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.740298720192245e-07,
+      "loss": 0.7865,
+      "step": 7476
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.7373377829747294e-07,
+      "loss": 1.0333,
+      "step": 7477
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.734378285667048e-07,
+      "loss": 0.6943,
+      "step": 7478
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.7314202286167036e-07,
+      "loss": 0.4168,
+      "step": 7479
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.728463612171053e-07,
+      "loss": 1.1084,
+      "step": 7480
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.7255084366772547e-07,
+      "loss": 0.9401,
+      "step": 7481
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.722554702482324e-07,
+      "loss": 0.7227,
+      "step": 7482
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.719602409933084e-07,
+      "loss": 0.7148,
+      "step": 7483
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.71665155937621e-07,
+      "loss": 1.0019,
+      "step": 7484
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.7137021511581875e-07,
+      "loss": 0.7828,
+      "step": 7485
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.7107541856253477e-07,
+      "loss": 0.9759,
+      "step": 7486
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.7078076631238524e-07,
+      "loss": 1.1339,
+      "step": 7487
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.704862583999678e-07,
+      "loss": 0.7025,
+      "step": 7488
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.701918948598655e-07,
+      "loss": 0.9326,
+      "step": 7489
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.6989767572664197e-07,
+      "loss": 0.8093,
+      "step": 7490
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.696036010348461e-07,
+      "loss": 0.9501,
+      "step": 7491
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.6930967081900795e-07,
+      "loss": 1.1307,
+      "step": 7492
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.690158851136422e-07,
+      "loss": 0.8539,
+      "step": 7493
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.687222439532449e-07,
+      "loss": 1.0939,
+      "step": 7494
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.684287473722974e-07,
+      "loss": 0.6054,
+      "step": 7495
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.681353954052615e-07,
+      "loss": 1.1502,
+      "step": 7496
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.6784218808658433e-07,
+      "loss": 0.9988,
+      "step": 7497
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.6754912545069404e-07,
+      "loss": 0.9449,
+      "step": 7498
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.672562075320035e-07,
+      "loss": 0.9298,
+      "step": 7499
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6696343436490695e-07,
+      "loss": 1.5664,
+      "step": 7500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.66670805983783e-07,
+      "loss": 1.3422,
+      "step": 7501
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.663783224229927e-07,
+      "loss": 0.8463,
+      "step": 7502
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.660859837168805e-07,
+      "loss": 0.8303,
+      "step": 7503
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6579378989977264e-07,
+      "loss": 1.2484,
+      "step": 7504
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.655017410059801e-07,
+      "loss": 0.9457,
+      "step": 7505
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.65209837069795e-07,
+      "loss": 0.7097,
+      "step": 7506
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.649180781254944e-07,
+      "loss": 0.9016,
+      "step": 7507
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6462646420733624e-07,
+      "loss": 0.9781,
+      "step": 7508
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.643349953495629e-07,
+      "loss": 1.0054,
+      "step": 7509
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.640436715863996e-07,
+      "loss": 0.7396,
+      "step": 7510
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.637524929520544e-07,
+      "loss": 0.6949,
+      "step": 7511
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.634614594807173e-07,
+      "loss": 0.881,
+      "step": 7512
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6317057120656317e-07,
+      "loss": 1.0148,
+      "step": 7513
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.628798281637478e-07,
+      "loss": 0.8084,
+      "step": 7514
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6258923038641165e-07,
+      "loss": 0.8296,
+      "step": 7515
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.622987779086768e-07,
+      "loss": 0.9997,
+      "step": 7516
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6200847076464916e-07,
+      "loss": 1.1379,
+      "step": 7517
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.617183089884175e-07,
+      "loss": 1.0612,
+      "step": 7518
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.614282926140526e-07,
+      "loss": 1.4014,
+      "step": 7519
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6113842167560964e-07,
+      "loss": 0.8354,
+      "step": 7520
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.608486962071251e-07,
+      "loss": 0.8313,
+      "step": 7521
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.6055911624261974e-07,
+      "loss": 1.0798,
+      "step": 7522
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.6026968181609655e-07,
+      "loss": 0.7322,
+      "step": 7523
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5998039296154203e-07,
+      "loss": 1.0577,
+      "step": 7524
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.596912497129243e-07,
+      "loss": 0.5635,
+      "step": 7525
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5940225210419603e-07,
+      "loss": 1.4021,
+      "step": 7526
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.591134001692911e-07,
+      "loss": 0.8749,
+      "step": 7527
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5882469394212824e-07,
+      "loss": 1.0518,
+      "step": 7528
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5853613345660697e-07,
+      "loss": 0.6427,
+      "step": 7529
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.582477187466117e-07,
+      "loss": 1.1726,
+      "step": 7530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5795944984600727e-07,
+      "loss": 0.8814,
+      "step": 7531
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.576713267886447e-07,
+      "loss": 1.0634,
+      "step": 7532
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5738334960835486e-07,
+      "loss": 0.7134,
+      "step": 7533
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5709551833895324e-07,
+      "loss": 0.8878,
+      "step": 7534
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.568078330142371e-07,
+      "loss": 0.9821,
+      "step": 7535
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.565202936679878e-07,
+      "loss": 1.0691,
+      "step": 7536
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5623290033396815e-07,
+      "loss": 0.8444,
+      "step": 7537
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5594565304592477e-07,
+      "loss": 1.1119,
+      "step": 7538
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.556585518375871e-07,
+      "loss": 1.0352,
+      "step": 7539
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.553715967426676e-07,
+      "loss": 0.6969,
+      "step": 7540
+    },
+    {
+      "epoch": 3.19,
+      "eval_loss": 1.2546535730361938,
+      "eval_runtime": 32.1006,
+      "eval_samples_per_second": 38.784,
+      "eval_steps_per_second": 19.408,
+      "step": 7540
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5508478779486014e-07,
+      "loss": 0.9152,
+      "step": 7541
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.547981250278434e-07,
+      "loss": 1.0758,
+      "step": 7542
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.5451160847527747e-07,
+      "loss": 0.7136,
+      "step": 7543
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.542252381708062e-07,
+      "loss": 0.7828,
+      "step": 7544
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.539390141480552e-07,
+      "loss": 0.6967,
+      "step": 7545
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5365293644063385e-07,
+      "loss": 0.9652,
+      "step": 7546
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5336700508213465e-07,
+      "loss": 0.75,
+      "step": 7547
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5308122010613145e-07,
+      "loss": 0.9706,
+      "step": 7548
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.52795581546182e-07,
+      "loss": 0.7823,
+      "step": 7549
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5251008943582734e-07,
+      "loss": 0.6237,
+      "step": 7550
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.522247438085895e-07,
+      "loss": 1.1665,
+      "step": 7551
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5193954469797527e-07,
+      "loss": 0.8981,
+      "step": 7552
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.516544921374726e-07,
+      "loss": 1.0448,
+      "step": 7553
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.513695861605537e-07,
+      "loss": 0.8896,
+      "step": 7554
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5108482680067285e-07,
+      "loss": 1.0887,
+      "step": 7555
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5080021409126656e-07,
+      "loss": 1.044,
+      "step": 7556
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.505157480657557e-07,
+      "loss": 0.689,
+      "step": 7557
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.5023142875754156e-07,
+      "loss": 1.0388,
+      "step": 7558
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.4994725620001093e-07,
+      "loss": 1.1037,
+      "step": 7559
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.49663230426531e-07,
+      "loss": 1.2524,
+      "step": 7560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.49379351470453e-07,
+      "loss": 1.015,
+      "step": 7561
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.4909561936511087e-07,
+      "loss": 1.1157,
+      "step": 7562
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.488120341438212e-07,
+      "loss": 1.3802,
+      "step": 7563
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.485285958398828e-07,
+      "loss": 1.0257,
+      "step": 7564
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.4824530448657805e-07,
+      "loss": 0.7533,
+      "step": 7565
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.47962160117171e-07,
+      "loss": 1.4272,
+      "step": 7566
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.476791627649102e-07,
+      "loss": 1.246,
+      "step": 7567
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.473963124630241e-07,
+      "loss": 0.7513,
+      "step": 7568
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4711360924472773e-07,
+      "loss": 1.0546,
+      "step": 7569
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4683105314321533e-07,
+      "loss": 1.06,
+      "step": 7570
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.465486441916661e-07,
+      "loss": 0.8629,
+      "step": 7571
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.462663824232402e-07,
+      "loss": 0.9371,
+      "step": 7572
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.459842678710825e-07,
+      "loss": 0.8568,
+      "step": 7573
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.457023005683186e-07,
+      "loss": 0.8848,
+      "step": 7574
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.454204805480586e-07,
+      "loss": 0.8142,
+      "step": 7575
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4513880784339367e-07,
+      "loss": 0.8029,
+      "step": 7576
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.448572824873988e-07,
+      "loss": 1.1417,
+      "step": 7577
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.445759045131313e-07,
+      "loss": 1.062,
+      "step": 7578
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4429467395363166e-07,
+      "loss": 0.6206,
+      "step": 7579
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.44013590841922e-07,
+      "loss": 0.9858,
+      "step": 7580
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4373265521100805e-07,
+      "loss": 1.107,
+      "step": 7581
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.434518670938777e-07,
+      "loss": 0.9413,
+      "step": 7582
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.431712265235019e-07,
+      "loss": 1.1618,
+      "step": 7583
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.428907335328343e-07,
+      "loss": 0.775,
+      "step": 7584
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4261038815481063e-07,
+      "loss": 0.7833,
+      "step": 7585
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4233019042235003e-07,
+      "loss": 0.9643,
+      "step": 7586
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4205014036835346e-07,
+      "loss": 0.9129,
+      "step": 7587
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.417702380257058e-07,
+      "loss": 1.1049,
+      "step": 7588
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4149048342727304e-07,
+      "loss": 0.9063,
+      "step": 7589
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.412108766059049e-07,
+      "loss": 0.8736,
+      "step": 7590
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.409314175944335e-07,
+      "loss": 0.8798,
+      "step": 7591
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.4065210642567386e-07,
+      "loss": 0.9517,
+      "step": 7592
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.4037294313242276e-07,
+      "loss": 1.0034,
+      "step": 7593
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.400939277474609e-07,
+      "loss": 1.0759,
+      "step": 7594
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3981506030355e-07,
+      "loss": 0.9476,
+      "step": 7595
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.395363408334361e-07,
+      "loss": 0.7996,
+      "step": 7596
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.392577693698466e-07,
+      "loss": 0.9911,
+      "step": 7597
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.38979345945492e-07,
+      "loss": 0.9162,
+      "step": 7598
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3870107059306574e-07,
+      "loss": 0.7322,
+      "step": 7599
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3842294334524356e-07,
+      "loss": 0.9729,
+      "step": 7600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3814496423468334e-07,
+      "loss": 0.7074,
+      "step": 7601
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3786713329402666e-07,
+      "loss": 0.8074,
+      "step": 7602
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3758945055589638e-07,
+      "loss": 0.6567,
+      "step": 7603
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3731191605289926e-07,
+      "loss": 0.8344,
+      "step": 7604
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3703452981762308e-07,
+      "loss": 0.8017,
+      "step": 7605
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.367572918826406e-07,
+      "loss": 1.0886,
+      "step": 7606
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3648020228050465e-07,
+      "loss": 0.9802,
+      "step": 7607
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3620326104375217e-07,
+      "loss": 0.8661,
+      "step": 7608
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.35926468204902e-07,
+      "loss": 0.868,
+      "step": 7609
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3564982379645612e-07,
+      "loss": 0.7041,
+      "step": 7610
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3537332785089822e-07,
+      "loss": 0.8298,
+      "step": 7611
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3509698040069567e-07,
+      "loss": 0.7485,
+      "step": 7612
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3482078147829717e-07,
+      "loss": 1.1059,
+      "step": 7613
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3454473111613512e-07,
+      "loss": 0.8314,
+      "step": 7614
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3426882934662407e-07,
+      "loss": 0.7044,
+      "step": 7615
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.339930762021605e-07,
+      "loss": 0.7201,
+      "step": 7616
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3371747171512476e-07,
+      "loss": 0.928,
+      "step": 7617
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.33442015917878e-07,
+      "loss": 1.101,
+      "step": 7618
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3316670884276587e-07,
+      "loss": 1.1643,
+      "step": 7619
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.328915505221143e-07,
+      "loss": 1.2187,
+      "step": 7620
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3261654098823432e-07,
+      "loss": 1.0825,
+      "step": 7621
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.323416802734174e-07,
+      "loss": 1.1945,
+      "step": 7622
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.320669684099389e-07,
+      "loss": 0.9019,
+      "step": 7623
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3179240543005536e-07,
+      "loss": 0.8306,
+      "step": 7624
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3151799136600748e-07,
+      "loss": 0.6773,
+      "step": 7625
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3124372625001666e-07,
+      "loss": 0.9079,
+      "step": 7626
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3096961011428846e-07,
+      "loss": 0.9514,
+      "step": 7627
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3069564299100916e-07,
+      "loss": 1.1397,
+      "step": 7628
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3042182491235014e-07,
+      "loss": 0.804,
+      "step": 7629
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3014815591046273e-07,
+      "loss": 1.1485,
+      "step": 7630
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2987463601748238e-07,
+      "loss": 0.5981,
+      "step": 7631
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2960126526552556e-07,
+      "loss": 0.9455,
+      "step": 7632
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2932804368669308e-07,
+      "loss": 1.0951,
+      "step": 7633
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.290549713130663e-07,
+      "loss": 1.1738,
+      "step": 7634
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2878204817671056e-07,
+      "loss": 1.1329,
+      "step": 7635
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2850927430967288e-07,
+      "loss": 1.217,
+      "step": 7636
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.282366497439835e-07,
+      "loss": 0.975,
+      "step": 7637
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2796417451165397e-07,
+      "loss": 0.7474,
+      "step": 7638
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.276918486446797e-07,
+      "loss": 1.1474,
+      "step": 7639
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2741967217503672e-07,
+      "loss": 0.6718,
+      "step": 7640
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2714764513468573e-07,
+      "loss": 0.9536,
+      "step": 7641
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2687576755556793e-07,
+      "loss": 1.4192,
+      "step": 7642
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2660403946960816e-07,
+      "loss": 1.2356,
+      "step": 7643
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2633246090871375e-07,
+      "loss": 0.7799,
+      "step": 7644
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2606103190477329e-07,
+      "loss": 1.0731,
+      "step": 7645
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2578975248965917e-07,
+      "loss": 0.9542,
+      "step": 7646
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2551862269522522e-07,
+      "loss": 0.8245,
+      "step": 7647
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2524764255330848e-07,
+      "loss": 1.1267,
+      "step": 7648
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.24976812095728e-07,
+      "loss": 1.1093,
+      "step": 7649
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.247061313542852e-07,
+      "loss": 0.8866,
+      "step": 7650
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2443560036076376e-07,
+      "loss": 0.9435,
+      "step": 7651
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2416521914693083e-07,
+      "loss": 0.9249,
+      "step": 7652
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2389498774453448e-07,
+      "loss": 1.0655,
+      "step": 7653
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2362490618530635e-07,
+      "loss": 1.0289,
+      "step": 7654
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2335497450095955e-07,
+      "loss": 0.7965,
+      "step": 7655
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2308519272319056e-07,
+      "loss": 0.81,
+      "step": 7656
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2281556088367688e-07,
+      "loss": 0.8593,
+      "step": 7657
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2254607901408087e-07,
+      "loss": 1.1984,
+      "step": 7658
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2227674714604434e-07,
+      "loss": 0.5978,
+      "step": 7659
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2200756531119355e-07,
+      "loss": 1.107,
+      "step": 7660
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.21738533541136e-07,
+      "loss": 0.9381,
+      "step": 7661
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.214696518674626e-07,
+      "loss": 1.2114,
+      "step": 7662
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2120092032174528e-07,
+      "loss": 0.9194,
+      "step": 7663
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2093233893553994e-07,
+      "loss": 1.1659,
+      "step": 7664
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2066390774038287e-07,
+      "loss": 0.7986,
+      "step": 7665
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.203956267677952e-07,
+      "loss": 0.8132,
+      "step": 7666
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2012749604927833e-07,
+      "loss": 1.0578,
+      "step": 7667
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1985951561631719e-07,
+      "loss": 0.8051,
+      "step": 7668
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1959168550037817e-07,
+      "loss": 1.1134,
+      "step": 7669
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.19324005732911e-07,
+      "loss": 0.807,
+      "step": 7670
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1905647634534654e-07,
+      "loss": 0.9693,
+      "step": 7671
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.187890973690993e-07,
+      "loss": 1.2451,
+      "step": 7672
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1852186883556568e-07,
+      "loss": 1.1671,
+      "step": 7673
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1825479077612377e-07,
+      "loss": 0.8776,
+      "step": 7674
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1798786322213443e-07,
+      "loss": 1.0055,
+      "step": 7675
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.177210862049418e-07,
+      "loss": 0.8972,
+      "step": 7676
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.174544597558703e-07,
+      "loss": 0.7192,
+      "step": 7677
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1718798390622858e-07,
+      "loss": 0.9927,
+      "step": 7678
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.169216586873064e-07,
+      "loss": 0.9392,
+      "step": 7679
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1665548413037644e-07,
+      "loss": 0.7644,
+      "step": 7680
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1638946026669388e-07,
+      "loss": 1.0637,
+      "step": 7681
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1612358712749502e-07,
+      "loss": 1.0915,
+      "step": 7682
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1585786474400027e-07,
+      "loss": 0.9008,
+      "step": 7683
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1559229314741065e-07,
+      "loss": 0.8537,
+      "step": 7684
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1532687236891053e-07,
+      "loss": 0.7831,
+      "step": 7685
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1506160243966599e-07,
+      "loss": 0.8515,
+      "step": 7686
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1479648339082563e-07,
+      "loss": 0.9662,
+      "step": 7687
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.145315152535204e-07,
+      "loss": 1.0999,
+      "step": 7688
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.14266698058864e-07,
+      "loss": 0.9513,
+      "step": 7689
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1400203183795097e-07,
+      "loss": 0.7429,
+      "step": 7690
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1373751662185974e-07,
+      "loss": 1.0961,
+      "step": 7691
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1347315244164983e-07,
+      "loss": 1.1434,
+      "step": 7692
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.132089393283639e-07,
+      "loss": 0.6452,
+      "step": 7693
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1294487731302575e-07,
+      "loss": 0.8367,
+      "step": 7694
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.126809664266428e-07,
+      "loss": 1.1132,
+      "step": 7695
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1241720670020375e-07,
+      "loss": 0.9089,
+      "step": 7696
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1215359816468034e-07,
+      "loss": 0.9565,
+      "step": 7697
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1189014085102542e-07,
+      "loss": 0.9147,
+      "step": 7698
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1162683479017546e-07,
+      "loss": 0.9875,
+      "step": 7699
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1136368001304774e-07,
+      "loss": 0.8108,
+      "step": 7700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1110067655054317e-07,
+      "loss": 0.9262,
+      "step": 7701
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.1083782443354304e-07,
+      "loss": 0.9677,
+      "step": 7702
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.105751236929137e-07,
+      "loss": 0.7984,
+      "step": 7703
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.103125743595008e-07,
+      "loss": 1.3464,
+      "step": 7704
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.100501764641345e-07,
+      "loss": 0.7139,
+      "step": 7705
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0978793003762497e-07,
+      "loss": 0.8297,
+      "step": 7706
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0952583511076695e-07,
+      "loss": 0.7353,
+      "step": 7707
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.092638917143353e-07,
+      "loss": 0.9696,
+      "step": 7708
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.090020998790884e-07,
+      "loss": 0.9179,
+      "step": 7709
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.087404596357669e-07,
+      "loss": 0.9716,
+      "step": 7710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.084789710150921e-07,
+      "loss": 0.7933,
+      "step": 7711
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.082176340477699e-07,
+      "loss": 0.8765,
+      "step": 7712
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0795644876448582e-07,
+      "loss": 0.9567,
+      "step": 7713
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.076954151959099e-07,
+      "loss": 1.5061,
+      "step": 7714
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0743453337269257e-07,
+      "loss": 0.9464,
+      "step": 7715
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.071738033254677e-07,
+      "loss": 1.0644,
+      "step": 7716
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0691322508484977e-07,
+      "loss": 1.0706,
+      "step": 7717
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.06652798681438e-07,
+      "loss": 0.8969,
+      "step": 7718
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0639252414581107e-07,
+      "loss": 0.9422,
+      "step": 7719
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0613240150853185e-07,
+      "loss": 1.0832,
+      "step": 7720
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0587243080014377e-07,
+      "loss": 0.9321,
+      "step": 7721
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0561261205117394e-07,
+      "loss": 1.0866,
+      "step": 7722
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0535294529213005e-07,
+      "loss": 1.0527,
+      "step": 7723
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.050934305535031e-07,
+      "loss": 1.3642,
+      "step": 7724
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.048340678657662e-07,
+      "loss": 1.2019,
+      "step": 7725
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.045748572593743e-07,
+      "loss": 1.1572,
+      "step": 7726
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0431579876476392e-07,
+      "loss": 0.9913,
+      "step": 7727
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0405689241235503e-07,
+      "loss": 0.7349,
+      "step": 7728
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0379813823254832e-07,
+      "loss": 0.7648,
+      "step": 7729
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0353953625572803e-07,
+      "loss": 1.0844,
+      "step": 7730
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0328108651225895e-07,
+      "loss": 1.1918,
+      "step": 7731
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0302278903248938e-07,
+      "loss": 1.0385,
+      "step": 7732
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0276464384674904e-07,
+      "loss": 0.9911,
+      "step": 7733
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.025066509853505e-07,
+      "loss": 1.0524,
+      "step": 7734
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0224881047858685e-07,
+      "loss": 0.6003,
+      "step": 7735
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0199112235673522e-07,
+      "loss": 0.5008,
+      "step": 7736
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0173358665005327e-07,
+      "loss": 0.8045,
+      "step": 7737
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0147620338878192e-07,
+      "loss": 0.6127,
+      "step": 7738
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.012189726031432e-07,
+      "loss": 0.7775,
+      "step": 7739
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0096189432334195e-07,
+      "loss": 0.9713,
+      "step": 7740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.007049685795655e-07,
+      "loss": 0.7212,
+      "step": 7741
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0044819540198155e-07,
+      "loss": 0.9819,
+      "step": 7742
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.00191574820742e-07,
+      "loss": 1.0307,
+      "step": 7743
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.999351068659791e-07,
+      "loss": 1.14,
+      "step": 7744
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9967879156780854e-07,
+      "loss": 0.735,
+      "step": 7745
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.994226289563268e-07,
+      "loss": 1.1611,
+      "step": 7746
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9916661906161328e-07,
+      "loss": 0.9952,
+      "step": 7747
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9891076191372941e-07,
+      "loss": 0.9138,
+      "step": 7748
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9865505754271873e-07,
+      "loss": 1.225,
+      "step": 7749
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9839950597860602e-07,
+      "loss": 0.7343,
+      "step": 7750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9814410725139943e-07,
+      "loss": 1.0085,
+      "step": 7751
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.978888613910878e-07,
+      "loss": 0.7681,
+      "step": 7752
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9763376842764336e-07,
+      "loss": 1.102,
+      "step": 7753
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9737882839101862e-07,
+      "loss": 1.0768,
+      "step": 7754
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9712404131115065e-07,
+      "loss": 0.6712,
+      "step": 7755
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9686940721795599e-07,
+      "loss": 0.7851,
+      "step": 7756
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9661492614133524e-07,
+      "loss": 0.8053,
+      "step": 7757
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9636059811116917e-07,
+      "loss": 0.9982,
+      "step": 7758
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9610642315732257e-07,
+      "loss": 0.8877,
+      "step": 7759
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9585240130964038e-07,
+      "loss": 0.9482,
+      "step": 7760
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9559853259795124e-07,
+      "loss": 0.9663,
+      "step": 7761
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.953448170520638e-07,
+      "loss": 1.2529,
+      "step": 7762
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.950912547017714e-07,
+      "loss": 0.9481,
+      "step": 7763
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9483784557684676e-07,
+      "loss": 1.1254,
+      "step": 7764
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9458458970704652e-07,
+      "loss": 0.7823,
+      "step": 7765
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9433148712210807e-07,
+      "loss": 1.0761,
+      "step": 7766
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9407853785175168e-07,
+      "loss": 0.6202,
+      "step": 7767
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.938257419256787e-07,
+      "loss": 1.0522,
+      "step": 7768
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9357309937357326e-07,
+      "loss": 1.0871,
+      "step": 7769
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.933206102251016e-07,
+      "loss": 1.1515,
+      "step": 7770
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9306827450991088e-07,
+      "loss": 0.7357,
+      "step": 7771
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.928160922576317e-07,
+      "loss": 0.7912,
+      "step": 7772
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9256406349787519e-07,
+      "loss": 0.9305,
+      "step": 7773
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.923121882602354e-07,
+      "loss": 0.7881,
+      "step": 7774
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.920604665742886e-07,
+      "loss": 1.0865,
+      "step": 7775
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9180889846959154e-07,
+      "loss": 0.9795,
+      "step": 7776
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9155748397568458e-07,
+      "loss": 0.6481,
+      "step": 7777
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.9130622312208955e-07,
+      "loss": 1.0503,
+      "step": 7778
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.9105511593830928e-07,
+      "loss": 0.7461,
+      "step": 7779
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.9080416245383043e-07,
+      "loss": 1.0927,
+      "step": 7780
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.9055336269811946e-07,
+      "loss": 0.9116,
+      "step": 7781
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.9030271670062665e-07,
+      "loss": 0.8665,
+      "step": 7782
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.900522244907828e-07,
+      "loss": 0.936,
+      "step": 7783
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.898018860980017e-07,
+      "loss": 0.8502,
+      "step": 7784
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8955170155167833e-07,
+      "loss": 0.8866,
+      "step": 7785
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8930167088119053e-07,
+      "loss": 0.7336,
+      "step": 7786
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8905179411589686e-07,
+      "loss": 0.5508,
+      "step": 7787
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8880207128513894e-07,
+      "loss": 1.2583,
+      "step": 7788
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8855250241823917e-07,
+      "loss": 1.9309,
+      "step": 7789
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.883030875445033e-07,
+      "loss": 0.8679,
+      "step": 7790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.880538266932169e-07,
+      "loss": 0.8237,
+      "step": 7791
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8780471989365022e-07,
+      "loss": 0.9747,
+      "step": 7792
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.87555767175053e-07,
+      "loss": 0.9681,
+      "step": 7793
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8730696856665857e-07,
+      "loss": 0.7892,
+      "step": 7794
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8705832409768054e-07,
+      "loss": 1.0649,
+      "step": 7795
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.868098337973162e-07,
+      "loss": 0.8009,
+      "step": 7796
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8656149769474307e-07,
+      "loss": 0.9314,
+      "step": 7797
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8631331581912198e-07,
+      "loss": 0.9895,
+      "step": 7798
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8606528819959445e-07,
+      "loss": 0.9818,
+      "step": 7799
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8581741486528486e-07,
+      "loss": 0.9077,
+      "step": 7800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8556969584529877e-07,
+      "loss": 1.4166,
+      "step": 7801
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8532213116872438e-07,
+      "loss": 0.6896,
+      "step": 7802
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.850747208646308e-07,
+      "loss": 0.9352,
+      "step": 7803
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8482746496206993e-07,
+      "loss": 1.0138,
+      "step": 7804
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8458036349007473e-07,
+      "loss": 0.7694,
+      "step": 7805
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.843334164776605e-07,
+      "loss": 0.9135,
+      "step": 7806
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.840866239538247e-07,
+      "loss": 1.0946,
+      "step": 7807
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8383998594754564e-07,
+      "loss": 1.1897,
+      "step": 7808
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8359350248778485e-07,
+      "loss": 1.0511,
+      "step": 7809
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.833471736034843e-07,
+      "loss": 0.8821,
+      "step": 7810
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8310099932356922e-07,
+      "loss": 0.9536,
+      "step": 7811
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8285497967694498e-07,
+      "loss": 0.895,
+      "step": 7812
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.826091146925008e-07,
+      "loss": 0.7992,
+      "step": 7813
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.823634043991056e-07,
+      "loss": 1.0853,
+      "step": 7814
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8211784882561227e-07,
+      "loss": 0.8322,
+      "step": 7815
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.818724480008541e-07,
+      "loss": 0.8096,
+      "step": 7816
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8162720195364674e-07,
+      "loss": 0.7947,
+      "step": 7817
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8138211071278726e-07,
+      "loss": 0.6545,
+      "step": 7818
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8113717430705517e-07,
+      "loss": 1.1699,
+      "step": 7819
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8089239276521096e-07,
+      "loss": 1.1955,
+      "step": 7820
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8064776611599797e-07,
+      "loss": 1.0531,
+      "step": 7821
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.804032943881404e-07,
+      "loss": 0.9692,
+      "step": 7822
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8015897761034543e-07,
+      "loss": 0.8523,
+      "step": 7823
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.799148158113003e-07,
+      "loss": 0.8505,
+      "step": 7824
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7967080901967593e-07,
+      "loss": 0.9465,
+      "step": 7825
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7942695726412333e-07,
+      "loss": 0.8717,
+      "step": 7826
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7918326057327684e-07,
+      "loss": 0.9151,
+      "step": 7827
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7893971897575135e-07,
+      "loss": 0.9544,
+      "step": 7828
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7869633250014417e-07,
+      "loss": 0.8638,
+      "step": 7829
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7845310117503445e-07,
+      "loss": 0.7304,
+      "step": 7830
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7821002502898331e-07,
+      "loss": 0.9023,
+      "step": 7831
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7796710409053263e-07,
+      "loss": 0.8953,
+      "step": 7832
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.777243383882074e-07,
+      "loss": 1.2373,
+      "step": 7833
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7748172795051315e-07,
+      "loss": 0.9928,
+      "step": 7834
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.772392728059381e-07,
+      "loss": 0.843,
+      "step": 7835
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7699697298295146e-07,
+      "loss": 1.1955,
+      "step": 7836
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.767548285100052e-07,
+      "loss": 1.0128,
+      "step": 7837
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.765128394155323e-07,
+      "loss": 0.8396,
+      "step": 7838
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.762710057279473e-07,
+      "loss": 0.9126,
+      "step": 7839
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7602932747564742e-07,
+      "loss": 1.1781,
+      "step": 7840
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7578780468701068e-07,
+      "loss": 0.7456,
+      "step": 7841
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7554643739039748e-07,
+      "loss": 0.9478,
+      "step": 7842
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.753052256141491e-07,
+      "loss": 1.0908,
+      "step": 7843
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7506416938659025e-07,
+      "loss": 0.9832,
+      "step": 7844
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7482326873602538e-07,
+      "loss": 0.846,
+      "step": 7845
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7458252369074236e-07,
+      "loss": 0.7781,
+      "step": 7846
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.743419342790092e-07,
+      "loss": 0.9739,
+      "step": 7847
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7410150052907715e-07,
+      "loss": 0.9493,
+      "step": 7848
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7386122246917796e-07,
+      "loss": 1.2734,
+      "step": 7849
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7362110012752625e-07,
+      "loss": 0.7431,
+      "step": 7850
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7338113353231676e-07,
+      "loss": 1.0869,
+      "step": 7851
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7314132271172817e-07,
+      "loss": 1.035,
+      "step": 7852
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7290166769391857e-07,
+      "loss": 0.9567,
+      "step": 7853
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.726621685070295e-07,
+      "loss": 0.7649,
+      "step": 7854
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.724228251791829e-07,
+      "loss": 0.8619,
+      "step": 7855
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.721836377384835e-07,
+      "loss": 0.9262,
+      "step": 7856
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.71944606213017e-07,
+      "loss": 0.8448,
+      "step": 7857
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7170573063085094e-07,
+      "loss": 0.9319,
+      "step": 7858
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.714670110200347e-07,
+      "loss": 0.8572,
+      "step": 7859
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7122844740859973e-07,
+      "loss": 1.1619,
+      "step": 7860
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7099003982455807e-07,
+      "loss": 1.2611,
+      "step": 7861
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7075178829590455e-07,
+      "loss": 0.7603,
+      "step": 7862
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7051369285061474e-07,
+      "loss": 0.6734,
+      "step": 7863
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.70275753516647e-07,
+      "loss": 0.8191,
+      "step": 7864
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7003797032194007e-07,
+      "loss": 0.8532,
+      "step": 7865
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6980034329441518e-07,
+      "loss": 1.3224,
+      "step": 7866
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6956287246197532e-07,
+      "loss": 0.8211,
+      "step": 7867
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6932555785250454e-07,
+      "loss": 1.0797,
+      "step": 7868
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6908839949386918e-07,
+      "loss": 0.5585,
+      "step": 7869
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.688513974139163e-07,
+      "loss": 1.3269,
+      "step": 7870
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6861455164047585e-07,
+      "loss": 0.7302,
+      "step": 7871
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6837786220135887e-07,
+      "loss": 0.8973,
+      "step": 7872
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.681413291243573e-07,
+      "loss": 1.0564,
+      "step": 7873
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.679049524372458e-07,
+      "loss": 1.1376,
+      "step": 7874
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.676687321677806e-07,
+      "loss": 0.6165,
+      "step": 7875
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6743266834369857e-07,
+      "loss": 0.9048,
+      "step": 7876
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6719676099271953e-07,
+      "loss": 1.1764,
+      "step": 7877
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6696101014254362e-07,
+      "loss": 0.6437,
+      "step": 7878
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.667254158208537e-07,
+      "loss": 0.8255,
+      "step": 7879
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6648997805531335e-07,
+      "loss": 1.1793,
+      "step": 7880
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.662546968735686e-07,
+      "loss": 0.7802,
+      "step": 7881
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6601957230324649e-07,
+      "loss": 0.9388,
+      "step": 7882
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.657846043719563e-07,
+      "loss": 0.8966,
+      "step": 7883
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6554979310728796e-07,
+      "loss": 0.933,
+      "step": 7884
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.653151385368139e-07,
+      "loss": 0.846,
+      "step": 7885
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6508064068808744e-07,
+      "loss": 1.098,
+      "step": 7886
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.648462995886445e-07,
+      "loss": 0.8146,
+      "step": 7887
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6461211526600078e-07,
+      "loss": 0.8357,
+      "step": 7888
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6437808774765612e-07,
+      "loss": 0.7843,
+      "step": 7889
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6414421706108967e-07,
+      "loss": 0.7567,
+      "step": 7890
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.639105032337635e-07,
+      "loss": 1.3676,
+      "step": 7891
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6367694629312047e-07,
+      "loss": 0.7344,
+      "step": 7892
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6344354626658582e-07,
+      "loss": 0.9019,
+      "step": 7893
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6321030318156528e-07,
+      "loss": 0.9017,
+      "step": 7894
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.629772170654476e-07,
+      "loss": 1.1831,
+      "step": 7895
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6274428794560143e-07,
+      "loss": 0.7066,
+      "step": 7896
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6251151584937818e-07,
+      "loss": 0.9971,
+      "step": 7897
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6227890080411085e-07,
+      "loss": 0.7607,
+      "step": 7898
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6204644283711306e-07,
+      "loss": 1.1828,
+      "step": 7899
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6181414197568084e-07,
+      "loss": 1.157,
+      "step": 7900
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6158199824709168e-07,
+      "loss": 0.8061,
+      "step": 7901
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6135001167860398e-07,
+      "loss": 0.7012,
+      "step": 7902
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6111818229745838e-07,
+      "loss": 0.9063,
+      "step": 7903
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6088651013087718e-07,
+      "loss": 0.6908,
+      "step": 7904
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.606549952060631e-07,
+      "loss": 0.9362,
+      "step": 7905
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6042363755020167e-07,
+      "loss": 0.7385,
+      "step": 7906
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6019243719045923e-07,
+      "loss": 0.9222,
+      "step": 7907
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5996139415398404e-07,
+      "loss": 0.5897,
+      "step": 7908
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5973050846790537e-07,
+      "loss": 0.7545,
+      "step": 7909
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.594997801593346e-07,
+      "loss": 0.8159,
+      "step": 7910
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.592692092553643e-07,
+      "loss": 0.8693,
+      "step": 7911
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.590387957830689e-07,
+      "loss": 0.9337,
+      "step": 7912
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5880853976950354e-07,
+      "loss": 1.0244,
+      "step": 7913
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.585784412417061e-07,
+      "loss": 0.8769,
+      "step": 7914
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.583485002266945e-07,
+      "loss": 0.9834,
+      "step": 7915
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5811871675146978e-07,
+      "loss": 0.9407,
+      "step": 7916
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5788909084301266e-07,
+      "loss": 0.8789,
+      "step": 7917
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5765962252828696e-07,
+      "loss": 0.7385,
+      "step": 7918
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.574303118342373e-07,
+      "loss": 0.727,
+      "step": 7919
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5720115878779024e-07,
+      "loss": 1.0819,
+      "step": 7920
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5697216341585262e-07,
+      "loss": 0.9634,
+      "step": 7921
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.567433257453143e-07,
+      "loss": 0.9435,
+      "step": 7922
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5651464580304542e-07,
+      "loss": 0.9255,
+      "step": 7923
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5628612361589867e-07,
+      "loss": 0.9436,
+      "step": 7924
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5605775921070692e-07,
+      "loss": 0.6028,
+      "step": 7925
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.558295526142857e-07,
+      "loss": 0.7808,
+      "step": 7926
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5560150385343164e-07,
+      "loss": 0.9744,
+      "step": 7927
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5537361295492276e-07,
+      "loss": 0.936,
+      "step": 7928
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.55145879945518e-07,
+      "loss": 0.9424,
+      "step": 7929
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5491830485195895e-07,
+      "loss": 0.9176,
+      "step": 7930
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5469088770096763e-07,
+      "loss": 0.6694,
+      "step": 7931
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.544636285192481e-07,
+      "loss": 0.7809,
+      "step": 7932
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.542365273334851e-07,
+      "loss": 0.8829,
+      "step": 7933
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5400958417034604e-07,
+      "loss": 0.8717,
+      "step": 7934
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5378279905647884e-07,
+      "loss": 0.6681,
+      "step": 7935
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.53556172018513e-07,
+      "loss": 1.1642,
+      "step": 7936
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.533297030830601e-07,
+      "loss": 1.2575,
+      "step": 7937
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5310339227671183e-07,
+      "loss": 1.3045,
+      "step": 7938
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5287723962604294e-07,
+      "loss": 0.6028,
+      "step": 7939
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.526512451576077e-07,
+      "loss": 0.6855,
+      "step": 7940
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5242540889794438e-07,
+      "loss": 0.9682,
+      "step": 7941
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.521997308735699e-07,
+      "loss": 0.9013,
+      "step": 7942
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.519742111109848e-07,
+      "loss": 0.9563,
+      "step": 7943
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5174884963666935e-07,
+      "loss": 0.9573,
+      "step": 7944
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5152364647708672e-07,
+      "loss": 1.0676,
+      "step": 7945
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5129860165868026e-07,
+      "loss": 1.069,
+      "step": 7946
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5107371520787567e-07,
+      "loss": 0.9867,
+      "step": 7947
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5084898715107863e-07,
+      "loss": 1.1435,
+      "step": 7948
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5062441751467876e-07,
+      "loss": 0.6999,
+      "step": 7949
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.504000063250444e-07,
+      "loss": 0.9751,
+      "step": 7950
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5017575360852704e-07,
+      "loss": 1.1282,
+      "step": 7951
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4995165939145855e-07,
+      "loss": 1.0114,
+      "step": 7952
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.497277237001529e-07,
+      "loss": 1.0713,
+      "step": 7953
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.495039465609046e-07,
+      "loss": 0.8355,
+      "step": 7954
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4928032799999057e-07,
+      "loss": 0.8878,
+      "step": 7955
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4905686804366853e-07,
+      "loss": 1.392,
+      "step": 7956
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4883356671817778e-07,
+      "loss": 0.6994,
+      "step": 7957
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4861042404973834e-07,
+      "loss": 0.95,
+      "step": 7958
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4838744006455296e-07,
+      "loss": 0.781,
+      "step": 7959
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.48164614788804e-07,
+      "loss": 0.7813,
+      "step": 7960
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4794194824865704e-07,
+      "loss": 1.2841,
+      "step": 7961
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4771944047025732e-07,
+      "loss": 1.2945,
+      "step": 7962
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4749709147973246e-07,
+      "loss": 1.0451,
+      "step": 7963
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4727490130319138e-07,
+      "loss": 0.7272,
+      "step": 7964
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.470528699667239e-07,
+      "loss": 0.6447,
+      "step": 7965
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4683099749640182e-07,
+      "loss": 1.1081,
+      "step": 7966
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.466092839182775e-07,
+      "loss": 1.12,
+      "step": 7967
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4638772925838539e-07,
+      "loss": 0.8831,
+      "step": 7968
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.461663335427404e-07,
+      "loss": 0.6824,
+      "step": 7969
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4594509679733986e-07,
+      "loss": 0.7898,
+      "step": 7970
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.457240190481617e-07,
+      "loss": 0.877,
+      "step": 7971
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4550310032116559e-07,
+      "loss": 1.0059,
+      "step": 7972
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4528234064229201e-07,
+      "loss": 0.6747,
+      "step": 7973
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4506174003746336e-07,
+      "loss": 1.063,
+      "step": 7974
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.448412985325826e-07,
+      "loss": 0.9229,
+      "step": 7975
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.44621016153535e-07,
+      "loss": 0.7075,
+      "step": 7976
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.444008929261859e-07,
+      "loss": 0.9574,
+      "step": 7977
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.441809288763838e-07,
+      "loss": 0.9809,
+      "step": 7978
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4396112402995633e-07,
+      "loss": 0.7384,
+      "step": 7979
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4374147841271419e-07,
+      "loss": 1.1511,
+      "step": 7980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4352199205044807e-07,
+      "loss": 1.094,
+      "step": 7981
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4330266496893125e-07,
+      "loss": 0.9599,
+      "step": 7982
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4308349719391688e-07,
+      "loss": 0.9538,
+      "step": 7983
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4286448875114077e-07,
+      "loss": 0.9659,
+      "step": 7984
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.426456396663185e-07,
+      "loss": 1.3859,
+      "step": 7985
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.42426949965149e-07,
+      "loss": 1.1459,
+      "step": 7986
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4220841967331055e-07,
+      "loss": 0.719,
+      "step": 7987
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4199004881646398e-07,
+      "loss": 1.0305,
+      "step": 7988
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4177183742025029e-07,
+      "loss": 0.9348,
+      "step": 7989
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4155378551029297e-07,
+      "loss": 0.9006,
+      "step": 7990
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4133589311219552e-07,
+      "loss": 0.8565,
+      "step": 7991
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4111816025154367e-07,
+      "loss": 1.083,
+      "step": 7992
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4090058695390445e-07,
+      "loss": 0.6796,
+      "step": 7993
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4068317324482543e-07,
+      "loss": 0.8858,
+      "step": 7994
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4046591914983603e-07,
+      "loss": 0.9702,
+      "step": 7995
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4024882469444634e-07,
+      "loss": 0.9228,
+      "step": 7996
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.400318899041483e-07,
+      "loss": 1.0762,
+      "step": 7997
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3981511480441533e-07,
+      "loss": 1.1555,
+      "step": 7998
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3959849942070098e-07,
+      "loss": 0.9356,
+      "step": 7999
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3938204377844098e-07,
+      "loss": 1.0155,
+      "step": 8000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3916574790305242e-07,
+      "loss": 1.1785,
+      "step": 8001
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.389496118199326e-07,
+      "loss": 0.8158,
+      "step": 8002
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3873363555446123e-07,
+      "loss": 1.1857,
+      "step": 8003
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3851781913199833e-07,
+      "loss": 0.9302,
+      "step": 8004
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3830216257788603e-07,
+      "loss": 0.9597,
+      "step": 8005
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3808666591744668e-07,
+      "loss": 0.8226,
+      "step": 8006
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.378713291759847e-07,
+      "loss": 0.7165,
+      "step": 8007
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3765615237878527e-07,
+      "loss": 1.0618,
+      "step": 8008
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3744113555111543e-07,
+      "loss": 1.0055,
+      "step": 8009
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3722627871822218e-07,
+      "loss": 1.3128,
+      "step": 8010
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3701158190533536e-07,
+      "loss": 0.8826,
+      "step": 8011
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3679704513766428e-07,
+      "loss": 1.0959,
+      "step": 8012
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3658266844040114e-07,
+      "loss": 1.033,
+      "step": 8013
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3636845183871777e-07,
+      "loss": 1.1643,
+      "step": 8014
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3615439535776842e-07,
+      "loss": 1.4258,
+      "step": 8015
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3594049902268796e-07,
+      "loss": 1.0888,
+      "step": 8016
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3572676285859298e-07,
+      "loss": 1.0858,
+      "step": 8017
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3551318689058022e-07,
+      "loss": 0.9911,
+      "step": 8018
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3529977114372883e-07,
+      "loss": 0.7481,
+      "step": 8019
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3508651564309805e-07,
+      "loss": 1.0003,
+      "step": 8020
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3487342041372952e-07,
+      "loss": 0.9401,
+      "step": 8021
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3466048548064459e-07,
+      "loss": 1.1057,
+      "step": 8022
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3444771086884687e-07,
+      "loss": 0.9958,
+      "step": 8023
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3423509660332096e-07,
+      "loss": 0.892,
+      "step": 8024
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.340226427090328e-07,
+      "loss": 1.3051,
+      "step": 8025
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.338103492109285e-07,
+      "loss": 0.9122,
+      "step": 8026
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3359821613393673e-07,
+      "loss": 0.9943,
+      "step": 8027
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3338624350296598e-07,
+      "loss": 0.6807,
+      "step": 8028
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3317443134290707e-07,
+      "loss": 0.8192,
+      "step": 8029
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3296277967863135e-07,
+      "loss": 0.6444,
+      "step": 8030
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3275128853499123e-07,
+      "loss": 0.8807,
+      "step": 8031
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3253995793682106e-07,
+      "loss": 1.1574,
+      "step": 8032
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3232878790893492e-07,
+      "loss": 0.8336,
+      "step": 8033
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3211777847612954e-07,
+      "loss": 1.0038,
+      "step": 8034
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3190692966318174e-07,
+      "loss": 1.0252,
+      "step": 8035
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3169624149485038e-07,
+      "loss": 1.1972,
+      "step": 8036
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3148571399587382e-07,
+      "loss": 0.8595,
+      "step": 8037
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3127534719097417e-07,
+      "loss": 1.056,
+      "step": 8038
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3106514110485212e-07,
+      "loss": 1.0117,
+      "step": 8039
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3085509576219129e-07,
+      "loss": 0.8734,
+      "step": 8040
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3064521118765492e-07,
+      "loss": 0.8636,
+      "step": 8041
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3043548740588867e-07,
+      "loss": 0.955,
+      "step": 8042
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.302259244415185e-07,
+      "loss": 0.9361,
+      "step": 8043
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3001652231915186e-07,
+      "loss": 1.0534,
+      "step": 8044
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2980728106337726e-07,
+      "loss": 1.1102,
+      "step": 8045
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.295982006987647e-07,
+      "loss": 0.8247,
+      "step": 8046
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2938928124986393e-07,
+      "loss": 0.7202,
+      "step": 8047
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2918052274120778e-07,
+      "loss": 0.7492,
+      "step": 8048
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2897192519730832e-07,
+      "loss": 0.892,
+      "step": 8049
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2876348864266012e-07,
+      "loss": 0.8475,
+      "step": 8050
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.285552131017378e-07,
+      "loss": 1.2068,
+      "step": 8051
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2834709859899773e-07,
+      "loss": 0.7742,
+      "step": 8052
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2813914515887726e-07,
+      "loss": 0.8081,
+      "step": 8053
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2793135280579498e-07,
+      "loss": 0.8405,
+      "step": 8054
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2772372156414992e-07,
+      "loss": 0.8941,
+      "step": 8055
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2751625145832302e-07,
+      "loss": 1.1363,
+      "step": 8056
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.273089425126755e-07,
+      "loss": 0.6687,
+      "step": 8057
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2710179475155054e-07,
+      "loss": 1.2116,
+      "step": 8058
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.268948081992712e-07,
+      "loss": 0.9217,
+      "step": 8059
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2668798288014282e-07,
+      "loss": 1.0386,
+      "step": 8060
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2648131881845138e-07,
+      "loss": 0.7362,
+      "step": 8061
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2627481603846343e-07,
+      "loss": 0.676,
+      "step": 8062
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2606847456442738e-07,
+      "loss": 0.7274,
+      "step": 8063
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2586229442057205e-07,
+      "loss": 1.2941,
+      "step": 8064
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2565627563110793e-07,
+      "loss": 0.63,
+      "step": 8065
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2545041822022564e-07,
+      "loss": 1.1426,
+      "step": 8066
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2524472221209782e-07,
+      "loss": 0.9484,
+      "step": 8067
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2503918763087784e-07,
+      "loss": 0.7854,
+      "step": 8068
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2483381450070008e-07,
+      "loss": 0.8502,
+      "step": 8069
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.246286028456794e-07,
+      "loss": 1.2446,
+      "step": 8070
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.24423552689913e-07,
+      "loss": 0.9766,
+      "step": 8071
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2421866405747768e-07,
+      "loss": 0.6298,
+      "step": 8072
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.240139369724323e-07,
+      "loss": 0.7616,
+      "step": 8073
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2380937145881582e-07,
+      "loss": 1.18,
+      "step": 8074
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.236049675406497e-07,
+      "loss": 0.9602,
+      "step": 8075
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2340072524193475e-07,
+      "loss": 1.0047,
+      "step": 8076
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2319664458665425e-07,
+      "loss": 0.8572,
+      "step": 8077
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2299272559877106e-07,
+      "loss": 0.9973,
+      "step": 8078
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2278896830223053e-07,
+      "loss": 0.8168,
+      "step": 8079
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2258537272095766e-07,
+      "loss": 0.8257,
+      "step": 8080
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2238193887885968e-07,
+      "loss": 1.1881,
+      "step": 8081
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.221786667998233e-07,
+      "loss": 0.9671,
+      "step": 8082
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2197555650771863e-07,
+      "loss": 4.1985,
+      "step": 8083
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2177260802639423e-07,
+      "loss": 0.8514,
+      "step": 8084
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2156982137968136e-07,
+      "loss": 0.6914,
+      "step": 8085
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2136719659139134e-07,
+      "loss": 0.8743,
+      "step": 8086
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.211647336853171e-07,
+      "loss": 0.7517,
+      "step": 8087
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.20962432685232e-07,
+      "loss": 0.9856,
+      "step": 8088
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2076029361489066e-07,
+      "loss": 1.2793,
+      "step": 8089
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.205583164980293e-07,
+      "loss": 1.0495,
+      "step": 8090
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.203565013583639e-07,
+      "loss": 0.7665,
+      "step": 8091
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2015484821959238e-07,
+      "loss": 1.3569,
+      "step": 8092
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1995335710539312e-07,
+      "loss": 1.1374,
+      "step": 8093
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1975202803942604e-07,
+      "loss": 0.7829,
+      "step": 8094
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1955086104533104e-07,
+      "loss": 0.929,
+      "step": 8095
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1934985614673006e-07,
+      "loss": 1.0568,
+      "step": 8096
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1914901336722539e-07,
+      "loss": 0.8082,
+      "step": 8097
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1894833273040068e-07,
+      "loss": 0.8163,
+      "step": 8098
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1874781425981989e-07,
+      "loss": 0.95,
+      "step": 8099
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1854745797902906e-07,
+      "loss": 1.1132,
+      "step": 8100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1834726391155365e-07,
+      "loss": 0.8702,
+      "step": 8101
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1814723208090156e-07,
+      "loss": 1.0357,
+      "step": 8102
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1794736251056033e-07,
+      "loss": 0.9606,
+      "step": 8103
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1774765522399966e-07,
+      "loss": 0.7962,
+      "step": 8104
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1754811024466932e-07,
+      "loss": 1.2055,
+      "step": 8105
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1734872759600073e-07,
+      "loss": 1.0344,
+      "step": 8106
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1714950730140533e-07,
+      "loss": 1.0368,
+      "step": 8107
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1695044938427673e-07,
+      "loss": 0.455,
+      "step": 8108
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1675155386798791e-07,
+      "loss": 0.7086,
+      "step": 8109
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1655282077589435e-07,
+      "loss": 0.9569,
+      "step": 8110
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1635425013133089e-07,
+      "loss": 0.8321,
+      "step": 8111
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1615584195761537e-07,
+      "loss": 1.1471,
+      "step": 8112
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1595759627804431e-07,
+      "loss": 0.6129,
+      "step": 8113
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1575951311589677e-07,
+      "loss": 1.0654,
+      "step": 8114
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1556159249443166e-07,
+      "loss": 0.9586,
+      "step": 8115
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1536383443688969e-07,
+      "loss": 0.7489,
+      "step": 8116
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1516623896649165e-07,
+      "loss": 0.9566,
+      "step": 8117
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1496880610644012e-07,
+      "loss": 1.1852,
+      "step": 8118
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1477153587991773e-07,
+      "loss": 0.7887,
+      "step": 8119
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1457442831008846e-07,
+      "loss": 0.6204,
+      "step": 8120
+    },
+    {
+      "epoch": 3.44,
+      "eval_loss": 1.254706621170044,
+      "eval_runtime": 32.1622,
+      "eval_samples_per_second": 38.71,
+      "eval_steps_per_second": 19.371,
+      "step": 8120
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1437748342009746e-07,
+      "loss": 1.1067,
+      "step": 8121
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.141807012330699e-07,
+      "loss": 0.6542,
+      "step": 8122
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1398408177211278e-07,
+      "loss": 0.6791,
+      "step": 8123
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1378762506031382e-07,
+      "loss": 0.7254,
+      "step": 8124
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1359133112074072e-07,
+      "loss": 0.3727,
+      "step": 8125
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1339519997644321e-07,
+      "loss": 1.1956,
+      "step": 8126
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1319923165045171e-07,
+      "loss": 1.0384,
+      "step": 8127
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1300342616577681e-07,
+      "loss": 1.2343,
+      "step": 8128
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1280778354541077e-07,
+      "loss": 0.9882,
+      "step": 8129
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.126123038123259e-07,
+      "loss": 0.8681,
+      "step": 8130
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.124169869894765e-07,
+      "loss": 0.9228,
+      "step": 8131
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1222183309979656e-07,
+      "loss": 0.7246,
+      "step": 8132
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1202684216620208e-07,
+      "loss": 0.7973,
+      "step": 8133
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.118320142115884e-07,
+      "loss": 0.8379,
+      "step": 8134
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1163734925883374e-07,
+      "loss": 1.0494,
+      "step": 8135
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1144284733079546e-07,
+      "loss": 1.3523,
+      "step": 8136
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1124850845031298e-07,
+      "loss": 1.1052,
+      "step": 8137
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.110543326402052e-07,
+      "loss": 0.926,
+      "step": 8138
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1086031992327338e-07,
+      "loss": 1.1024,
+      "step": 8139
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1066647032229865e-07,
+      "loss": 1.0253,
+      "step": 8140
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1047278386004311e-07,
+      "loss": 0.8023,
+      "step": 8141
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1027926055925008e-07,
+      "loss": 1.1553,
+      "step": 8142
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1008590044264371e-07,
+      "loss": 0.7504,
+      "step": 8143
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0989270353292851e-07,
+      "loss": 1.1319,
+      "step": 8144
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0969966985279018e-07,
+      "loss": 0.9663,
+      "step": 8145
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0950679942489506e-07,
+      "loss": 0.9502,
+      "step": 8146
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.093140922718907e-07,
+      "loss": 1.083,
+      "step": 8147
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0912154841640487e-07,
+      "loss": 0.7661,
+      "step": 8148
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0892916788104662e-07,
+      "loss": 0.8722,
+      "step": 8149
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0873695068840589e-07,
+      "loss": 1.0154,
+      "step": 8150
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0854489686105329e-07,
+      "loss": 1.2771,
+      "step": 8151
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0835300642153978e-07,
+      "loss": 0.8993,
+      "step": 8152
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0816127939239834e-07,
+      "loss": 0.9617,
+      "step": 8153
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0796971579614112e-07,
+      "loss": 0.6933,
+      "step": 8154
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0777831565526264e-07,
+      "loss": 0.8127,
+      "step": 8155
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0758707899223708e-07,
+      "loss": 1.5703,
+      "step": 8156
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0739600582951997e-07,
+      "loss": 0.9663,
+      "step": 8157
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0720509618954804e-07,
+      "loss": 0.8402,
+      "step": 8158
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0701435009473753e-07,
+      "loss": 1.1326,
+      "step": 8159
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0682376756748719e-07,
+      "loss": 0.7199,
+      "step": 8160
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0663334863017477e-07,
+      "loss": 1.001,
+      "step": 8161
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0644309330516039e-07,
+      "loss": 0.8094,
+      "step": 8162
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0625300161478352e-07,
+      "loss": 1.3247,
+      "step": 8163
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0606307358136613e-07,
+      "loss": 1.0221,
+      "step": 8164
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0587330922720922e-07,
+      "loss": 0.6949,
+      "step": 8165
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.056837085745958e-07,
+      "loss": 0.9218,
+      "step": 8166
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.054942716457889e-07,
+      "loss": 0.8835,
+      "step": 8167
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0530499846303304e-07,
+      "loss": 0.6671,
+      "step": 8168
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0511588904855247e-07,
+      "loss": 1.1282,
+      "step": 8169
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0492694342455356e-07,
+      "loss": 0.7004,
+      "step": 8170
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0473816161322191e-07,
+      "loss": 1.0867,
+      "step": 8171
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0454954363672564e-07,
+      "loss": 0.6482,
+      "step": 8172
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0436108951721218e-07,
+      "loss": 1.0055,
+      "step": 8173
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.041727992768105e-07,
+      "loss": 1.0353,
+      "step": 8174
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0398467293762958e-07,
+      "loss": 0.8871,
+      "step": 8175
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0379671052176026e-07,
+      "loss": 0.7796,
+      "step": 8176
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0360891205127304e-07,
+      "loss": 1.062,
+      "step": 8177
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0342127754821978e-07,
+      "loss": 0.9882,
+      "step": 8178
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0323380703463303e-07,
+      "loss": 0.7338,
+      "step": 8179
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0304650053252634e-07,
+      "loss": 0.883,
+      "step": 8180
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0285935806389296e-07,
+      "loss": 0.9882,
+      "step": 8181
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0267237965070813e-07,
+      "loss": 0.9235,
+      "step": 8182
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0248556531492682e-07,
+      "loss": 1.0763,
+      "step": 8183
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0229891507848577e-07,
+      "loss": 0.7156,
+      "step": 8184
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0211242896330148e-07,
+      "loss": 0.9095,
+      "step": 8185
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0192610699127142e-07,
+      "loss": 0.871,
+      "step": 8186
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.017399491842746e-07,
+      "loss": 0.7414,
+      "step": 8187
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0155395556416936e-07,
+      "loss": 0.9183,
+      "step": 8188
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0136812615279605e-07,
+      "loss": 0.739,
+      "step": 8189
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0118246097197475e-07,
+      "loss": 0.8104,
+      "step": 8190
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0099696004350717e-07,
+      "loss": 1.0694,
+      "step": 8191
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0081162338917471e-07,
+      "loss": 1.0035,
+      "step": 8192
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0062645103074031e-07,
+      "loss": 1.0413,
+      "step": 8193
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0044144298994723e-07,
+      "loss": 1.1288,
+      "step": 8194
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0025659928851993e-07,
+      "loss": 0.9712,
+      "step": 8195
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0007191994816272e-07,
+      "loss": 1.0131,
+      "step": 8196
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.988740499056126e-08,
+      "loss": 0.8924,
+      "step": 8197
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.970305443738153e-08,
+      "loss": 0.7484,
+      "step": 8198
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.951886831027091e-08,
+      "loss": 0.8974,
+      "step": 8199
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.933484663085624e-08,
+      "loss": 1.2936,
+      "step": 8200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.915098942074608e-08,
+      "loss": 0.747,
+      "step": 8201
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.896729670152949e-08,
+      "loss": 0.8742,
+      "step": 8202
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.878376849477622e-08,
+      "loss": 0.8539,
+      "step": 8203
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.860040482203603e-08,
+      "loss": 0.8113,
+      "step": 8204
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.841720570484052e-08,
+      "loss": 0.9297,
+      "step": 8205
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.823417116470085e-08,
+      "loss": 0.9455,
+      "step": 8206
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.805130122310979e-08,
+      "loss": 1.0083,
+      "step": 8207
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.786859590153952e-08,
+      "loss": 0.8489,
+      "step": 8208
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.768605522144486e-08,
+      "loss": 1.0737,
+      "step": 8209
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.750367920425935e-08,
+      "loss": 0.9969,
+      "step": 8210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.73214678713985e-08,
+      "loss": 0.9395,
+      "step": 8211
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.713942124425757e-08,
+      "loss": 0.6742,
+      "step": 8212
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.695753934421325e-08,
+      "loss": 0.9259,
+      "step": 8213
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.677582219262216e-08,
+      "loss": 1.2741,
+      "step": 8214
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.65942698108222e-08,
+      "loss": 0.9391,
+      "step": 8215
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.641288222013134e-08,
+      "loss": 0.8847,
+      "step": 8216
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.623165944184887e-08,
+      "loss": 0.6805,
+      "step": 8217
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.605060149725458e-08,
+      "loss": 0.7404,
+      "step": 8218
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.5869708407608e-08,
+      "loss": 1.0781,
+      "step": 8219
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.568898019415079e-08,
+      "loss": 1.0506,
+      "step": 8220
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.55084168781038e-08,
+      "loss": 1.1552,
+      "step": 8221
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.532801848066958e-08,
+      "loss": 1.1232,
+      "step": 8222
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.514778502303068e-08,
+      "loss": 0.9735,
+      "step": 8223
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.4967716526351e-08,
+      "loss": 0.9105,
+      "step": 8224
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.478781301177397e-08,
+      "loss": 0.7769,
+      "step": 8225
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.460807450042485e-08,
+      "loss": 1.0494,
+      "step": 8226
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.442850101340845e-08,
+      "loss": 0.7772,
+      "step": 8227
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.424909257181119e-08,
+      "loss": 1.13,
+      "step": 8228
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.406984919669892e-08,
+      "loss": 0.7286,
+      "step": 8229
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.389077090911946e-08,
+      "loss": 1.1268,
+      "step": 8230
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.371185773010016e-08,
+      "loss": 1.1577,
+      "step": 8231
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.353310968065005e-08,
+      "loss": 0.8939,
+      "step": 8232
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.335452678175732e-08,
+      "loss": 0.7775,
+      "step": 8233
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.317610905439222e-08,
+      "loss": 1.0009,
+      "step": 8234
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.299785651950449e-08,
+      "loss": 0.9931,
+      "step": 8235
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.281976919802555e-08,
+      "loss": 0.8316,
+      "step": 8236
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.264184711086599e-08,
+      "loss": 1.0255,
+      "step": 8237
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.246409027891844e-08,
+      "loss": 0.983,
+      "step": 8238
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.228649872305523e-08,
+      "loss": 0.9173,
+      "step": 8239
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.210907246413015e-08,
+      "loss": 1.1321,
+      "step": 8240
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.193181152297603e-08,
+      "loss": 0.8036,
+      "step": 8241
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.175471592040808e-08,
+      "loss": 0.9015,
+      "step": 8242
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.15777856772208e-08,
+      "loss": 1.0628,
+      "step": 8243
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.140102081419011e-08,
+      "loss": 0.7828,
+      "step": 8244
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.122442135207171e-08,
+      "loss": 0.7388,
+      "step": 8245
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.104798731160252e-08,
+      "loss": 0.8806,
+      "step": 8246
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.087171871349997e-08,
+      "loss": 0.9359,
+      "step": 8247
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.069561557846168e-08,
+      "loss": 0.9073,
+      "step": 8248
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.051967792716626e-08,
+      "loss": 0.6792,
+      "step": 8249
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.034390578027269e-08,
+      "loss": 0.8466,
+      "step": 8250
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.016829915842046e-08,
+      "loss": 0.6195,
+      "step": 8251
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.99928580822299e-08,
+      "loss": 0.6766,
+      "step": 8252
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.981758257230122e-08,
+      "loss": 0.7432,
+      "step": 8253
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.96424726492161e-08,
+      "loss": 0.9625,
+      "step": 8254
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.946752833353644e-08,
+      "loss": 0.8259,
+      "step": 8255
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.929274964580414e-08,
+      "loss": 0.6902,
+      "step": 8256
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.91181366065426e-08,
+      "loss": 0.8434,
+      "step": 8257
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.894368923625479e-08,
+      "loss": 1.1924,
+      "step": 8258
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.876940755542528e-08,
+      "loss": 1.0956,
+      "step": 8259
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.85952915845179e-08,
+      "loss": 1.363,
+      "step": 8260
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.842134134397861e-08,
+      "loss": 0.9451,
+      "step": 8261
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.824755685423225e-08,
+      "loss": 0.8321,
+      "step": 8262
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.807393813568582e-08,
+      "loss": 0.9343,
+      "step": 8263
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.790048520872518e-08,
+      "loss": 0.9016,
+      "step": 8264
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.77271980937182e-08,
+      "loss": 0.8423,
+      "step": 8265
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.755407681101213e-08,
+      "loss": 1.0726,
+      "step": 8266
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.738112138093569e-08,
+      "loss": 0.6999,
+      "step": 8267
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.720833182379728e-08,
+      "loss": 0.5363,
+      "step": 8268
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.703570815988688e-08,
+      "loss": 0.6706,
+      "step": 8269
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.686325040947373e-08,
+      "loss": 0.7862,
+      "step": 8270
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.669095859280867e-08,
+      "loss": 1.1383,
+      "step": 8271
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.651883273012217e-08,
+      "loss": 0.9364,
+      "step": 8272
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.634687284162623e-08,
+      "loss": 0.9066,
+      "step": 8273
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.617507894751203e-08,
+      "loss": 0.8042,
+      "step": 8274
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.600345106795244e-08,
+      "loss": 0.7071,
+      "step": 8275
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.583198922310031e-08,
+      "loss": 1.0182,
+      "step": 8276
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.566069343308941e-08,
+      "loss": 0.9526,
+      "step": 8277
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.548956371803312e-08,
+      "loss": 1.1871,
+      "step": 8278
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.531860009802638e-08,
+      "loss": 1.1514,
+      "step": 8279
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.514780259314381e-08,
+      "loss": 0.8522,
+      "step": 8280
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.497717122344101e-08,
+      "loss": 0.8229,
+      "step": 8281
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.480670600895363e-08,
+      "loss": 0.8172,
+      "step": 8282
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.463640696969815e-08,
+      "loss": 0.9212,
+      "step": 8283
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.446627412567209e-08,
+      "loss": 1.027,
+      "step": 8284
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.429630749685197e-08,
+      "loss": 0.9272,
+      "step": 8285
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.412650710319631e-08,
+      "loss": 1.2681,
+      "step": 8286
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.395687296464282e-08,
+      "loss": 1.026,
+      "step": 8287
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.37874051011111e-08,
+      "loss": 0.7641,
+      "step": 8288
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.361810353249988e-08,
+      "loss": 1.3275,
+      "step": 8289
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.344896827868893e-08,
+      "loss": 0.7674,
+      "step": 8290
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.327999935953884e-08,
+      "loss": 0.9414,
+      "step": 8291
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.311119679489027e-08,
+      "loss": 0.9586,
+      "step": 8292
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.294256060456418e-08,
+      "loss": 0.9437,
+      "step": 8293
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.27740908083624e-08,
+      "loss": 0.9089,
+      "step": 8294
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.260578742606694e-08,
+      "loss": 1.2341,
+      "step": 8295
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.243765047744051e-08,
+      "loss": 0.8261,
+      "step": 8296
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.226967998222562e-08,
+      "loss": 1.0203,
+      "step": 8297
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.210187596014668e-08,
+      "loss": 0.7324,
+      "step": 8298
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.193423843090674e-08,
+      "loss": 0.7473,
+      "step": 8299
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.176676741419093e-08,
+      "loss": 0.8964,
+      "step": 8300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.159946292966331e-08,
+      "loss": 1.1637,
+      "step": 8301
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.143232499696984e-08,
+      "loss": 1.2423,
+      "step": 8302
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.126535363573567e-08,
+      "loss": 1.0031,
+      "step": 8303
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.109854886556761e-08,
+      "loss": 0.8102,
+      "step": 8304
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.093191070605116e-08,
+      "loss": 0.6537,
+      "step": 8305
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.076543917675467e-08,
+      "loss": 0.7528,
+      "step": 8306
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.059913429722482e-08,
+      "loss": 0.9831,
+      "step": 8307
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.043299608698984e-08,
+      "loss": 1.0314,
+      "step": 8308
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.026702456555779e-08,
+      "loss": 0.7549,
+      "step": 8309
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.010121975241758e-08,
+      "loss": 1.4158,
+      "step": 8310
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.99355816670383e-08,
+      "loss": 0.8292,
+      "step": 8311
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.977011032886955e-08,
+      "loss": 0.7809,
+      "step": 8312
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.960480575734163e-08,
+      "loss": 0.8593,
+      "step": 8313
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.943966797186453e-08,
+      "loss": 0.8476,
+      "step": 8314
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.927469699182955e-08,
+      "loss": 0.7128,
+      "step": 8315
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.910989283660752e-08,
+      "loss": 1.4293,
+      "step": 8316
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.894525552555065e-08,
+      "loss": 0.8513,
+      "step": 8317
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.878078507799046e-08,
+      "loss": 0.8266,
+      "step": 8318
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.861648151323969e-08,
+      "loss": 0.8882,
+      "step": 8319
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.845234485059138e-08,
+      "loss": 0.7134,
+      "step": 8320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.82883751093188e-08,
+      "loss": 0.9921,
+      "step": 8321
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.812457230867537e-08,
+      "loss": 0.9772,
+      "step": 8322
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.79609364678957e-08,
+      "loss": 1.1085,
+      "step": 8323
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.779746760619361e-08,
+      "loss": 0.8301,
+      "step": 8324
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.763416574276472e-08,
+      "loss": 0.8964,
+      "step": 8325
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.74710308967837e-08,
+      "loss": 0.8491,
+      "step": 8326
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.73080630874064e-08,
+      "loss": 0.9973,
+      "step": 8327
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.714526233376884e-08,
+      "loss": 0.8597,
+      "step": 8328
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.69826286549879e-08,
+      "loss": 1.0851,
+      "step": 8329
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.682016207015979e-08,
+      "loss": 0.8887,
+      "step": 8330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.66578625983621e-08,
+      "loss": 0.9078,
+      "step": 8331
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.649573025865226e-08,
+      "loss": 1.1603,
+      "step": 8332
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.633376507006818e-08,
+      "loss": 1.0197,
+      "step": 8333
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.617196705162816e-08,
+      "loss": 1.247,
+      "step": 8334
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.6010336222331e-08,
+      "loss": 1.079,
+      "step": 8335
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.58488726011557e-08,
+      "loss": 1.0963,
+      "step": 8336
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.568757620706191e-08,
+      "loss": 0.7698,
+      "step": 8337
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.552644705898898e-08,
+      "loss": 1.2486,
+      "step": 8338
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.536548517585745e-08,
+      "loss": 0.9333,
+      "step": 8339
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.520469057656754e-08,
+      "loss": 0.9126,
+      "step": 8340
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.504406328000063e-08,
+      "loss": 1.0968,
+      "step": 8341
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.488360330501715e-08,
+      "loss": 1.157,
+      "step": 8342
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.472331067045917e-08,
+      "loss": 0.739,
+      "step": 8343
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.456318539514884e-08,
+      "loss": 1.0083,
+      "step": 8344
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.440322749788791e-08,
+      "loss": 0.7684,
+      "step": 8345
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.424343699745923e-08,
+      "loss": 1.2133,
+      "step": 8346
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.408381391262592e-08,
+      "loss": 0.9487,
+      "step": 8347
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.3924358262131e-08,
+      "loss": 0.7651,
+      "step": 8348
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.376507006469835e-08,
+      "loss": 1.278,
+      "step": 8349
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.360594933903197e-08,
+      "loss": 0.9506,
+      "step": 8350
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.344699610381594e-08,
+      "loss": 0.9858,
+      "step": 8351
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.328821037771516e-08,
+      "loss": 0.6992,
+      "step": 8352
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.312959217937438e-08,
+      "loss": 0.877,
+      "step": 8353
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.297114152741918e-08,
+      "loss": 0.8475,
+      "step": 8354
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.281285844045488e-08,
+      "loss": 0.8568,
+      "step": 8355
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.265474293706792e-08,
+      "loss": 0.8147,
+      "step": 8356
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.249679503582379e-08,
+      "loss": 0.9619,
+      "step": 8357
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.233901475527e-08,
+      "loss": 0.9014,
+      "step": 8358
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.21814021139327e-08,
+      "loss": 0.7109,
+      "step": 8359
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.202395713031995e-08,
+      "loss": 1.1054,
+      "step": 8360
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.186667982291844e-08,
+      "loss": 1.0332,
+      "step": 8361
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.170957021019659e-08,
+      "loss": 0.702,
+      "step": 8362
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.155262831060243e-08,
+      "loss": 0.726,
+      "step": 8363
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.139585414256411e-08,
+      "loss": 0.8488,
+      "step": 8364
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.123924772449087e-08,
+      "loss": 0.8573,
+      "step": 8365
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.108280907477183e-08,
+      "loss": 0.8022,
+      "step": 8366
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.092653821177597e-08,
+      "loss": 0.7375,
+      "step": 8367
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.077043515385346e-08,
+      "loss": 1.013,
+      "step": 8368
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.061449991933377e-08,
+      "loss": 0.8811,
+      "step": 8369
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.045873252652746e-08,
+      "loss": 0.8237,
+      "step": 8370
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.030313299372487e-08,
+      "loss": 0.7908,
+      "step": 8371
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.014770133919707e-08,
+      "loss": 0.8425,
+      "step": 8372
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.999243758119545e-08,
+      "loss": 0.906,
+      "step": 8373
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.983734173795076e-08,
+      "loss": 1.0608,
+      "step": 8374
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.96824138276751e-08,
+      "loss": 0.7869,
+      "step": 8375
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.952765386856075e-08,
+      "loss": 0.8237,
+      "step": 8376
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.937306187877934e-08,
+      "loss": 0.8772,
+      "step": 8377
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.921863787648403e-08,
+      "loss": 0.9472,
+      "step": 8378
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.906438187980712e-08,
+      "loss": 0.6658,
+      "step": 8379
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.891029390686198e-08,
+      "loss": 1.2975,
+      "step": 8380
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.875637397574197e-08,
+      "loss": 0.7941,
+      "step": 8381
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.860262210452062e-08,
+      "loss": 0.8725,
+      "step": 8382
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.844903831125198e-08,
+      "loss": 1.1382,
+      "step": 8383
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.829562261396999e-08,
+      "loss": 0.5901,
+      "step": 8384
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.814237503068938e-08,
+      "loss": 0.8521,
+      "step": 8385
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.798929557940442e-08,
+      "loss": 0.7259,
+      "step": 8386
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.783638427809041e-08,
+      "loss": 1.0752,
+      "step": 8387
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.768364114470232e-08,
+      "loss": 0.8838,
+      "step": 8388
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.753106619717581e-08,
+      "loss": 0.8939,
+      "step": 8389
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.737865945342636e-08,
+      "loss": 1.0583,
+      "step": 8390
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.722642093135e-08,
+      "loss": 0.8955,
+      "step": 8391
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.707435064882294e-08,
+      "loss": 1.0727,
+      "step": 8392
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.692244862370189e-08,
+      "loss": 0.9509,
+      "step": 8393
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.677071487382275e-08,
+      "loss": 1.2099,
+      "step": 8394
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.661914941700326e-08,
+      "loss": 0.9167,
+      "step": 8395
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.646775227104018e-08,
+      "loss": 0.9606,
+      "step": 8396
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.63165234537113e-08,
+      "loss": 1.1433,
+      "step": 8397
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.61654629827736e-08,
+      "loss": 0.8918,
+      "step": 8398
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.601457087596552e-08,
+      "loss": 1.1584,
+      "step": 8399
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.58638471510049e-08,
+      "loss": 0.9001,
+      "step": 8400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.571329182559011e-08,
+      "loss": 0.6961,
+      "step": 8401
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.556290491739947e-08,
+      "loss": 0.7747,
+      "step": 8402
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.54126864440922e-08,
+      "loss": 0.9507,
+      "step": 8403
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.526263642330704e-08,
+      "loss": 1.02,
+      "step": 8404
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.511275487266339e-08,
+      "loss": 0.9469,
+      "step": 8405
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.496304180976031e-08,
+      "loss": 0.9928,
+      "step": 8406
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.481349725217795e-08,
+      "loss": 0.6779,
+      "step": 8407
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.466412121747573e-08,
+      "loss": 0.9964,
+      "step": 8408
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.451491372319379e-08,
+      "loss": 0.9912,
+      "step": 8409
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.436587478685279e-08,
+      "loss": 0.8917,
+      "step": 8410
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.421700442595274e-08,
+      "loss": 0.9676,
+      "step": 8411
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.406830265797481e-08,
+      "loss": 1.2023,
+      "step": 8412
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.391976950037937e-08,
+      "loss": 0.9244,
+      "step": 8413
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.377140497060797e-08,
+      "loss": 0.6054,
+      "step": 8414
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.362320908608166e-08,
+      "loss": 0.8692,
+      "step": 8415
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.347518186420187e-08,
+      "loss": 0.9078,
+      "step": 8416
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.332732332235053e-08,
+      "loss": 1.1512,
+      "step": 8417
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.317963347788958e-08,
+      "loss": 0.6742,
+      "step": 8418
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.303211234816097e-08,
+      "loss": 0.8164,
+      "step": 8419
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.288475995048703e-08,
+      "loss": 1.0405,
+      "step": 8420
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.273757630216992e-08,
+      "loss": 0.7927,
+      "step": 8421
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.259056142049264e-08,
+      "loss": 0.9438,
+      "step": 8422
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.244371532271787e-08,
+      "loss": 0.8863,
+      "step": 8423
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.229703802608849e-08,
+      "loss": 1.115,
+      "step": 8424
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.21505295478279e-08,
+      "loss": 1.0427,
+      "step": 8425
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.20041899051395e-08,
+      "loss": 0.7749,
+      "step": 8426
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.185801911520655e-08,
+      "loss": 1.0596,
+      "step": 8427
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.171201719519314e-08,
+      "loss": 0.9208,
+      "step": 8428
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.156618416224257e-08,
+      "loss": 1.1832,
+      "step": 8429
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.142052003347965e-08,
+      "loss": 1.2347,
+      "step": 8430
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.127502482600766e-08,
+      "loss": 0.7118,
+      "step": 8431
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.112969855691197e-08,
+      "loss": 0.8141,
+      "step": 8432
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.098454124325642e-08,
+      "loss": 1.0483,
+      "step": 8433
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.083955290208621e-08,
+      "loss": 0.6485,
+      "step": 8434
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.069473355042571e-08,
+      "loss": 0.6876,
+      "step": 8435
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.055008320528032e-08,
+      "loss": 0.9207,
+      "step": 8436
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.040560188363497e-08,
+      "loss": 0.7843,
+      "step": 8437
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.026128960245508e-08,
+      "loss": 0.9024,
+      "step": 8438
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.01171463786861e-08,
+      "loss": 1.0193,
+      "step": 8439
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.99731722292538e-08,
+      "loss": 0.873,
+      "step": 8440
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.982936717106386e-08,
+      "loss": 1.1974,
+      "step": 8441
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.968573122100207e-08,
+      "loss": 0.8163,
+      "step": 8442
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.95422643959348e-08,
+      "loss": 0.9197,
+      "step": 8443
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.9398966712707727e-08,
+      "loss": 1.2268,
+      "step": 8444
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.925583818814773e-08,
+      "loss": 1.076,
+      "step": 8445
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.9112878839061026e-08,
+      "loss": 1.2022,
+      "step": 8446
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8970088682234356e-08,
+      "loss": 0.8291,
+      "step": 8447
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.882746773443432e-08,
+      "loss": 0.9796,
+      "step": 8448
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8685016012408e-08,
+      "loss": 0.8986,
+      "step": 8449
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.8542733532882034e-08,
+      "loss": 1.0445,
+      "step": 8450
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.8400620312564044e-08,
+      "loss": 0.7261,
+      "step": 8451
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.825867636814086e-08,
+      "loss": 1.1364,
+      "step": 8452
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.811690171628031e-08,
+      "loss": 0.614,
+      "step": 8453
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.797529637362925e-08,
+      "loss": 1.1693,
+      "step": 8454
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.7833860356815874e-08,
+      "loss": 1.1905,
+      "step": 8455
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.769259368244773e-08,
+      "loss": 0.7882,
+      "step": 8456
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.755149636711271e-08,
+      "loss": 0.8654,
+      "step": 8457
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.741056842737874e-08,
+      "loss": 0.8335,
+      "step": 8458
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.726980987979391e-08,
+      "loss": 1.188,
+      "step": 8459
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.712922074088634e-08,
+      "loss": 1.1148,
+      "step": 8460
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6988801027164473e-08,
+      "loss": 0.9229,
+      "step": 8461
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.684855075511647e-08,
+      "loss": 0.8249,
+      "step": 8462
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.670846994121132e-08,
+      "loss": 1.3732,
+      "step": 8463
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6568558601897026e-08,
+      "loss": 1.1985,
+      "step": 8464
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.642881675360295e-08,
+      "loss": 0.9308,
+      "step": 8465
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.628924441273714e-08,
+      "loss": 0.9985,
+      "step": 8466
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.614984159568931e-08,
+      "loss": 0.9986,
+      "step": 8467
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6010608318827706e-08,
+      "loss": 0.5144,
+      "step": 8468
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.587154459850175e-08,
+      "loss": 0.7761,
+      "step": 8469
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.573265045104087e-08,
+      "loss": 0.7002,
+      "step": 8470
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5593925892754026e-08,
+      "loss": 0.6506,
+      "step": 8471
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5455370939930526e-08,
+      "loss": 0.6842,
+      "step": 8472
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.531698560884002e-08,
+      "loss": 0.8559,
+      "step": 8473
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.517876991573184e-08,
+      "loss": 0.8934,
+      "step": 8474
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.504072387683584e-08,
+      "loss": 1.2282,
+      "step": 8475
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.490284750836155e-08,
+      "loss": 0.5205,
+      "step": 8476
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.4765140826498525e-08,
+      "loss": 1.0146,
+      "step": 8477
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.4627603847417e-08,
+      "loss": 0.8513,
+      "step": 8478
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.449023658726654e-08,
+      "loss": 0.9266,
+      "step": 8479
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.4353039062177426e-08,
+      "loss": 0.5982,
+      "step": 8480
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.421601128825943e-08,
+      "loss": 0.8326,
+      "step": 8481
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.407915328160284e-08,
+      "loss": 1.0999,
+      "step": 8482
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.394246505827749e-08,
+      "loss": 0.8421,
+      "step": 8483
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.380594663433419e-08,
+      "loss": 0.841,
+      "step": 8484
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.3669598025802954e-08,
+      "loss": 1.0148,
+      "step": 8485
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.3533419248694146e-08,
+      "loss": 0.855,
+      "step": 8486
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.3397410318998143e-08,
+      "loss": 1.3072,
+      "step": 8487
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.3261571252685495e-08,
+      "loss": 1.0213,
+      "step": 8488
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.312590206570678e-08,
+      "loss": 0.7135,
+      "step": 8489
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.2990402773992584e-08,
+      "loss": 1.0423,
+      "step": 8490
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.285507339345319e-08,
+      "loss": 0.7976,
+      "step": 8491
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.271991393997988e-08,
+      "loss": 0.9148,
+      "step": 8492
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.258492442944279e-08,
+      "loss": 1.0979,
+      "step": 8493
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.245010487769325e-08,
+      "loss": 1.3426,
+      "step": 8494
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.231545530056175e-08,
+      "loss": 0.7344,
+      "step": 8495
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.218097571385932e-08,
+      "loss": 1.1662,
+      "step": 8496
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.204666613337666e-08,
+      "loss": 1.2613,
+      "step": 8497
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.1912526574884654e-08,
+      "loss": 1.1096,
+      "step": 8498
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.1778557054134535e-08,
+      "loss": 0.7957,
+      "step": 8499
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.164475758685738e-08,
+      "loss": 1.0213,
+      "step": 8500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.151112818876397e-08,
+      "loss": 0.944,
+      "step": 8501
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.137766887554557e-08,
+      "loss": 0.8483,
+      "step": 8502
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.1244379662872987e-08,
+      "loss": 1.2818,
+      "step": 8503
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.11112605663977e-08,
+      "loss": 0.9958,
+      "step": 8504
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.097831160175054e-08,
+      "loss": 0.9647,
+      "step": 8505
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.084553278454268e-08,
+      "loss": 0.714,
+      "step": 8506
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.071292413036582e-08,
+      "loss": 1.1546,
+      "step": 8507
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.058048565479051e-08,
+      "loss": 1.149,
+      "step": 8508
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.0448217373368466e-08,
+      "loss": 1.1446,
+      "step": 8509
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.03161193016306e-08,
+      "loss": 0.8027,
+      "step": 8510
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.0184191455088514e-08,
+      "loss": 0.8564,
+      "step": 8511
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.0052433849233135e-08,
+      "loss": 0.8631,
+      "step": 8512
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.9920846499535766e-08,
+      "loss": 0.7727,
+      "step": 8513
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.978942942144787e-08,
+      "loss": 0.7541,
+      "step": 8514
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.965818263040078e-08,
+      "loss": 0.7335,
+      "step": 8515
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.952710614180567e-08,
+      "loss": 0.8358,
+      "step": 8516
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.939619997105388e-08,
+      "loss": 1.0039,
+      "step": 8517
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.9265464133516626e-08,
+      "loss": 1.1221,
+      "step": 8518
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.9134898644545456e-08,
+      "loss": 0.9784,
+      "step": 8519
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.9004503519471276e-08,
+      "loss": 1.0247,
+      "step": 8520
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8874278773605673e-08,
+      "loss": 0.5237,
+      "step": 8521
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8744224422239745e-08,
+      "loss": 0.72,
+      "step": 8522
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.861434048064512e-08,
+      "loss": 0.8081,
+      "step": 8523
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.848462696407258e-08,
+      "loss": 0.8609,
+      "step": 8524
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8355083887753946e-08,
+      "loss": 0.8438,
+      "step": 8525
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.82257112668999e-08,
+      "loss": 0.8844,
+      "step": 8526
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8096509116702105e-08,
+      "loss": 0.7924,
+      "step": 8527
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.796747745233126e-08,
+      "loss": 0.5921,
+      "step": 8528
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.783861628893926e-08,
+      "loss": 0.83,
+      "step": 8529
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.770992564165666e-08,
+      "loss": 0.9513,
+      "step": 8530
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.7581405525595036e-08,
+      "loss": 1.1084,
+      "step": 8531
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.745305595584515e-08,
+      "loss": 0.9385,
+      "step": 8532
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.732487694747861e-08,
+      "loss": 0.7757,
+      "step": 8533
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.719686851554572e-08,
+      "loss": 0.7964,
+      "step": 8534
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.706903067507828e-08,
+      "loss": 0.9851,
+      "step": 8535
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.694136344108696e-08,
+      "loss": 1.2207,
+      "step": 8536
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.681386682856259e-08,
+      "loss": 1.1872,
+      "step": 8537
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.6686540852476536e-08,
+      "loss": 1.1458,
+      "step": 8538
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.655938552777916e-08,
+      "loss": 0.6885,
+      "step": 8539
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.643240086940204e-08,
+      "loss": 1.1081,
+      "step": 8540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.630558689225523e-08,
+      "loss": 1.0812,
+      "step": 8541
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.617894361123015e-08,
+      "loss": 0.9982,
+      "step": 8542
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.605247104119675e-08,
+      "loss": 0.9529,
+      "step": 8543
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5926169197006815e-08,
+      "loss": 1.0071,
+      "step": 8544
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.580003809349015e-08,
+      "loss": 0.6422,
+      "step": 8545
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.56740777454579e-08,
+      "loss": 1.0819,
+      "step": 8546
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.554828816770007e-08,
+      "loss": 0.9572,
+      "step": 8547
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.542266937498768e-08,
+      "loss": 0.8763,
+      "step": 8548
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.529722138207076e-08,
+      "loss": 0.9339,
+      "step": 8549
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5171944203680035e-08,
+      "loss": 1.0138,
+      "step": 8550
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5046837854525566e-08,
+      "loss": 0.9226,
+      "step": 8551
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.492190234929794e-08,
+      "loss": 0.7388,
+      "step": 8552
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.479713770266691e-08,
+      "loss": 0.8799,
+      "step": 8553
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4672543929283105e-08,
+      "loss": 1.3305,
+      "step": 8554
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4548121043776316e-08,
+      "loss": 1.0838,
+      "step": 8555
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.442386906075668e-08,
+      "loss": 1.1169,
+      "step": 8556
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4299787994814046e-08,
+      "loss": 0.9475,
+      "step": 8557
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.417587786051841e-08,
+      "loss": 0.7891,
+      "step": 8558
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4052138672419475e-08,
+      "loss": 0.7194,
+      "step": 8559
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.392857044504728e-08,
+      "loss": 0.6554,
+      "step": 8560
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3805173192910887e-08,
+      "loss": 0.9746,
+      "step": 8561
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3681946930500536e-08,
+      "loss": 1.0498,
+      "step": 8562
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.355889167228533e-08,
+      "loss": 0.7824,
+      "step": 8563
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.343600743271503e-08,
+      "loss": 1.0828,
+      "step": 8564
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.331329422621844e-08,
+      "loss": 0.9139,
+      "step": 8565
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3190752067205354e-08,
+      "loss": 0.9578,
+      "step": 8566
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.306838097006477e-08,
+      "loss": 1.0761,
+      "step": 8567
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.294618094916569e-08,
+      "loss": 0.7457,
+      "step": 8568
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.282415201885731e-08,
+      "loss": 0.9205,
+      "step": 8569
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2702294193468164e-08,
+      "loss": 0.8772,
+      "step": 8570
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2580607487307464e-08,
+      "loss": 0.9424,
+      "step": 8571
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.245909191466413e-08,
+      "loss": 1.0104,
+      "step": 8572
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2337747489806067e-08,
+      "loss": 0.8237,
+      "step": 8573
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.221657422698239e-08,
+      "loss": 1.2628,
+      "step": 8574
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2095572140421537e-08,
+      "loss": 0.8968,
+      "step": 8575
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.197474124433148e-08,
+      "loss": 1.0189,
+      "step": 8576
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1854081552900883e-08,
+      "loss": 1.1108,
+      "step": 8577
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.173359308029756e-08,
+      "loss": 0.982,
+      "step": 8578
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.161327584066987e-08,
+      "loss": 1.0446,
+      "step": 8579
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.149312984814502e-08,
+      "loss": 0.9759,
+      "step": 8580
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.137315511683187e-08,
+      "loss": 1.1413,
+      "step": 8581
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.125335166081734e-08,
+      "loss": 0.8887,
+      "step": 8582
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.113371949416933e-08,
+      "loss": 0.9811,
+      "step": 8583
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1014258630935277e-08,
+      "loss": 1.3866,
+      "step": 8584
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.089496908514262e-08,
+      "loss": 1.0064,
+      "step": 8585
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.0775850870798494e-08,
+      "loss": 0.7738,
+      "step": 8586
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.065690400189021e-08,
+      "loss": 0.8356,
+      "step": 8587
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.053812849238425e-08,
+      "loss": 0.9462,
+      "step": 8588
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.041952435622831e-08,
+      "loss": 0.7446,
+      "step": 8589
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.030109160734841e-08,
+      "loss": 0.9709,
+      "step": 8590
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.018283025965192e-08,
+      "loss": 0.7644,
+      "step": 8591
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.006474032702473e-08,
+      "loss": 0.6742,
+      "step": 8592
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9946821823333594e-08,
+      "loss": 0.9232,
+      "step": 8593
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9829074762424424e-08,
+      "loss": 1.0667,
+      "step": 8594
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.971149915812367e-08,
+      "loss": 0.7803,
+      "step": 8595
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.959409502423744e-08,
+      "loss": 1.0853,
+      "step": 8596
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9476862374551225e-08,
+      "loss": 0.9467,
+      "step": 8597
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.935980122283084e-08,
+      "loss": 0.9332,
+      "step": 8598
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.924291158282211e-08,
+      "loss": 1.1297,
+      "step": 8599
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.912619346825025e-08,
+      "loss": 1.2191,
+      "step": 8600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.900964689282077e-08,
+      "loss": 1.125,
+      "step": 8601
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8893271870218575e-08,
+      "loss": 0.8771,
+      "step": 8602
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8777068414108716e-08,
+      "loss": 0.7738,
+      "step": 8603
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8661036538136465e-08,
+      "loss": 0.9941,
+      "step": 8604
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8545176255925905e-08,
+      "loss": 0.9874,
+      "step": 8605
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8429487581082336e-08,
+      "loss": 0.6857,
+      "step": 8606
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8313970527189546e-08,
+      "loss": 0.6785,
+      "step": 8607
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.819862510781219e-08,
+      "loss": 0.765,
+      "step": 8608
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8083451336494267e-08,
+      "loss": 0.9967,
+      "step": 8609
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.796844922675963e-08,
+      "loss": 0.7444,
+      "step": 8610
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.78536187921123e-08,
+      "loss": 1.2213,
+      "step": 8611
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7738960046035994e-08,
+      "loss": 0.772,
+      "step": 8612
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.762447300199395e-08,
+      "loss": 0.92,
+      "step": 8613
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.751015767342958e-08,
+      "loss": 0.8917,
+      "step": 8614
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.739601407376614e-08,
+      "loss": 1.0915,
+      "step": 8615
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.728204221640658e-08,
+      "loss": 0.7901,
+      "step": 8616
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.716824211473352e-08,
+      "loss": 0.9836,
+      "step": 8617
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.705461378211011e-08,
+      "loss": 0.8665,
+      "step": 8618
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.694115723187835e-08,
+      "loss": 0.8411,
+      "step": 8619
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.682787247736108e-08,
+      "loss": 1.0997,
+      "step": 8620
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6714759531859823e-08,
+      "loss": 0.6286,
+      "step": 8621
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6601818408657286e-08,
+      "loss": 0.9156,
+      "step": 8622
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.648904912101453e-08,
+      "loss": 0.461,
+      "step": 8623
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6376451682173796e-08,
+      "loss": 1.3712,
+      "step": 8624
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6264026105356004e-08,
+      "loss": 1.0395,
+      "step": 8625
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.615177240376294e-08,
+      "loss": 1.0948,
+      "step": 8626
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.603969059057538e-08,
+      "loss": 3.8763,
+      "step": 8627
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5927780678954305e-08,
+      "loss": 0.8341,
+      "step": 8628
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.581604268204036e-08,
+      "loss": 0.7537,
+      "step": 8629
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.570447661295439e-08,
+      "loss": 0.8882,
+      "step": 8630
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5593082484796256e-08,
+      "loss": 0.9126,
+      "step": 8631
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.548186031064632e-08,
+      "loss": 1.3449,
+      "step": 8632
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.537081010356463e-08,
+      "loss": 0.8273,
+      "step": 8633
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.525993187659094e-08,
+      "loss": 0.767,
+      "step": 8634
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.514922564274481e-08,
+      "loss": 1.0965,
+      "step": 8635
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5038691415025357e-08,
+      "loss": 1.0454,
+      "step": 8636
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4928329206412204e-08,
+      "loss": 0.9309,
+      "step": 8637
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.481813902986414e-08,
+      "loss": 0.7822,
+      "step": 8638
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.470812089831982e-08,
+      "loss": 0.9948,
+      "step": 8639
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.459827482469774e-08,
+      "loss": 1.2402,
+      "step": 8640
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.448860082189675e-08,
+      "loss": 0.7544,
+      "step": 8641
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4379098902794725e-08,
+      "loss": 1.2252,
+      "step": 8642
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.42697690802497e-08,
+      "loss": 0.8178,
+      "step": 8643
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.416061136709925e-08,
+      "loss": 1.0499,
+      "step": 8644
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.405162577616111e-08,
+      "loss": 0.7326,
+      "step": 8645
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.394281232023255e-08,
+      "loss": 1.019,
+      "step": 8646
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.383417101209069e-08,
+      "loss": 1.1501,
+      "step": 8647
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.372570186449231e-08,
+      "loss": 0.99,
+      "step": 8648
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.3617404890174564e-08,
+      "loss": 0.8965,
+      "step": 8649
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.350928010185328e-08,
+      "loss": 1.1382,
+      "step": 8650
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.3401327512225464e-08,
+      "loss": 0.8107,
+      "step": 8651
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.3293547133966484e-08,
+      "loss": 1.1449,
+      "step": 8652
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.318593897973255e-08,
+      "loss": 0.3869,
+      "step": 8653
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.307850306215887e-08,
+      "loss": 0.982,
+      "step": 8654
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.2971239393861206e-08,
+      "loss": 0.6652,
+      "step": 8655
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.2864147987434647e-08,
+      "loss": 1.2615,
+      "step": 8656
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.275722885545412e-08,
+      "loss": 0.9486,
+      "step": 8657
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.265048201047394e-08,
+      "loss": 0.8317,
+      "step": 8658
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2543907465029224e-08,
+      "loss": 1.1821,
+      "step": 8659
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.243750523163347e-08,
+      "loss": 1.1185,
+      "step": 8660
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.233127532278135e-08,
+      "loss": 1.0213,
+      "step": 8661
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.222521775094606e-08,
+      "loss": 1.0625,
+      "step": 8662
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.211933252858146e-08,
+      "loss": 0.8399,
+      "step": 8663
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2013619668120774e-08,
+      "loss": 0.9257,
+      "step": 8664
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1908079181976915e-08,
+      "loss": 0.9617,
+      "step": 8665
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.180271108254296e-08,
+      "loss": 0.7213,
+      "step": 8666
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1697515382191177e-08,
+      "loss": 1.268,
+      "step": 8667
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.159249209327403e-08,
+      "loss": 1.1053,
+      "step": 8668
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1487641228123474e-08,
+      "loss": 0.9255,
+      "step": 8669
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.138296279905167e-08,
+      "loss": 0.9623,
+      "step": 8670
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.127845681834979e-08,
+      "loss": 0.9532,
+      "step": 8671
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.11741232982895e-08,
+      "loss": 1.1726,
+      "step": 8672
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.106996225112152e-08,
+      "loss": 1.0634,
+      "step": 8673
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0965973689077054e-08,
+      "loss": 0.7948,
+      "step": 8674
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.086215762436634e-08,
+      "loss": 1.3174,
+      "step": 8675
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0758514069179955e-08,
+      "loss": 0.9613,
+      "step": 8676
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.065504303568767e-08,
+      "loss": 0.9158,
+      "step": 8677
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0551744536039764e-08,
+      "loss": 0.9478,
+      "step": 8678
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.044861858236536e-08,
+      "loss": 0.8364,
+      "step": 8679
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.034566518677412e-08,
+      "loss": 0.7305,
+      "step": 8680
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.024288436135453e-08,
+      "loss": 0.8186,
+      "step": 8681
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.014027611817594e-08,
+      "loss": 1.0671,
+      "step": 8682
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.003784046928637e-08,
+      "loss": 0.7159,
+      "step": 8683
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.993557742671421e-08,
+      "loss": 1.1283,
+      "step": 8684
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9833487002467354e-08,
+      "loss": 0.8611,
+      "step": 8685
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.973156920853387e-08,
+      "loss": 1.3554,
+      "step": 8686
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9629824056880696e-08,
+      "loss": 0.9062,
+      "step": 8687
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9528251559455433e-08,
+      "loss": 0.8483,
+      "step": 8688
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.942685172818438e-08,
+      "loss": 0.8607,
+      "step": 8689
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9325624574974673e-08,
+      "loss": 0.8144,
+      "step": 8690
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.922457011171231e-08,
+      "loss": 0.9007,
+      "step": 8691
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9123688350263466e-08,
+      "loss": 0.8899,
+      "step": 8692
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9022979302473994e-08,
+      "loss": 0.817,
+      "step": 8693
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.8922442980169107e-08,
+      "loss": 1.2113,
+      "step": 8694
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.8822079395154353e-08,
+      "loss": 1.011,
+      "step": 8695
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.8721888559214482e-08,
+      "loss": 0.9732,
+      "step": 8696
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.862187048411391e-08,
+      "loss": 0.8631,
+      "step": 8697
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.8522025181597578e-08,
+      "loss": 1.2667,
+      "step": 8698
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.8422352663388774e-08,
+      "loss": 0.9458,
+      "step": 8699
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.83228529411918e-08,
+      "loss": 0.891,
+      "step": 8700
+    },
+    {
+      "epoch": 3.69,
+      "eval_loss": 1.2547883987426758,
+      "eval_runtime": 32.0246,
+      "eval_samples_per_second": 38.876,
+      "eval_steps_per_second": 19.454,
+      "step": 8700
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.8223526026690315e-08,
+      "loss": 1.4729,
+      "step": 8701
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.812437193154682e-08,
+      "loss": 0.8056,
+      "step": 8702
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.8025390667405006e-08,
+      "loss": 1.1664,
+      "step": 8703
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.7926582245886734e-08,
+      "loss": 1.0833,
+      "step": 8704
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.7827946678594896e-08,
+      "loss": 0.8148,
+      "step": 8705
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.7729483977111226e-08,
+      "loss": 0.9224,
+      "step": 8706
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.7631194152997308e-08,
+      "loss": 0.6786,
+      "step": 8707
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.7533077217794912e-08,
+      "loss": 0.3843,
+      "step": 8708
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.7435133183024987e-08,
+      "loss": 0.8967,
+      "step": 8709
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.733736206018833e-08,
+      "loss": 0.8152,
+      "step": 8710
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.723976386076543e-08,
+      "loss": 0.6947,
+      "step": 8711
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.7142338596216454e-08,
+      "loss": 1.0853,
+      "step": 8712
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.7045086277981578e-08,
+      "loss": 0.8839,
+      "step": 8713
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.694800691747984e-08,
+      "loss": 0.9148,
+      "step": 8714
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6851100526111118e-08,
+      "loss": 0.6174,
+      "step": 8715
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6754367115254142e-08,
+      "loss": 1.235,
+      "step": 8716
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6657806696267496e-08,
+      "loss": 0.9957,
+      "step": 8717
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.656141928048944e-08,
+      "loss": 0.9918,
+      "step": 8718
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6465204879238425e-08,
+      "loss": 0.933,
+      "step": 8719
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6369163503811744e-08,
+      "loss": 1.0581,
+      "step": 8720
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6273295165487045e-08,
+      "loss": 0.8875,
+      "step": 8721
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6177599875520987e-08,
+      "loss": 0.5355,
+      "step": 8722
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6082077645150916e-08,
+      "loss": 0.6705,
+      "step": 8723
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.598672848559286e-08,
+      "loss": 0.8606,
+      "step": 8724
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.5891552408043194e-08,
+      "loss": 0.9774,
+      "step": 8725
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.5796549423677484e-08,
+      "loss": 0.8349,
+      "step": 8726
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.5701719543651468e-08,
+      "loss": 0.9994,
+      "step": 8727
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.560706277909991e-08,
+      "loss": 0.8635,
+      "step": 8728
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.5512579141137913e-08,
+      "loss": 0.9122,
+      "step": 8729
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.5418268640859942e-08,
+      "loss": 0.7019,
+      "step": 8730
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.5324131289340136e-08,
+      "loss": 0.6659,
+      "step": 8731
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.523016709763215e-08,
+      "loss": 0.9469,
+      "step": 8732
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.513637607676966e-08,
+      "loss": 1.194,
+      "step": 8733
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.504275823776586e-08,
+      "loss": 1.021,
+      "step": 8734
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4949313591613287e-08,
+      "loss": 0.859,
+      "step": 8735
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4856042149284665e-08,
+      "loss": 0.8446,
+      "step": 8736
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.476294392173206e-08,
+      "loss": 0.7133,
+      "step": 8737
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4670018919887572e-08,
+      "loss": 0.774,
+      "step": 8738
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.457726715466213e-08,
+      "loss": 1.0088,
+      "step": 8739
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4484688636947194e-08,
+      "loss": 0.9481,
+      "step": 8740
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.43922833776134e-08,
+      "loss": 1.0092,
+      "step": 8741
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4300051387511567e-08,
+      "loss": 0.811,
+      "step": 8742
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4207992677471203e-08,
+      "loss": 0.4486,
+      "step": 8743
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4116107258302323e-08,
+      "loss": 1.0779,
+      "step": 8744
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.4024395140794297e-08,
+      "loss": 1.0318,
+      "step": 8745
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.3932856335716513e-08,
+      "loss": 1.0675,
+      "step": 8746
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.3841490853817038e-08,
+      "loss": 1.0826,
+      "step": 8747
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.3750298705824793e-08,
+      "loss": 0.6607,
+      "step": 8748
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.365927990244737e-08,
+      "loss": 0.8888,
+      "step": 8749
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.3568434454372566e-08,
+      "loss": 1.3776,
+      "step": 8750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.347776237226751e-08,
+      "loss": 0.922,
+      "step": 8751
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.3387263666779513e-08,
+      "loss": 0.8962,
+      "step": 8752
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.3296938348534745e-08,
+      "loss": 0.7629,
+      "step": 8753
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.3206786428139558e-08,
+      "loss": 0.9804,
+      "step": 8754
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.3116807916179816e-08,
+      "loss": 0.9326,
+      "step": 8755
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.3027002823221067e-08,
+      "loss": 0.8927,
+      "step": 8756
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.293737115980821e-08,
+      "loss": 0.8437,
+      "step": 8757
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.284791293646632e-08,
+      "loss": 0.9404,
+      "step": 8758
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.2758628163699334e-08,
+      "loss": 1.0381,
+      "step": 8759
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.26695168519917e-08,
+      "loss": 1.104,
+      "step": 8760
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.258057901180688e-08,
+      "loss": 0.8989,
+      "step": 8761
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.2491814653588182e-08,
+      "loss": 0.7302,
+      "step": 8762
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.2403223787758608e-08,
+      "loss": 0.9285,
+      "step": 8763
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.23148064247205e-08,
+      "loss": 0.7443,
+      "step": 8764
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.2226562574856225e-08,
+      "loss": 1.0955,
+      "step": 8765
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.213849224852732e-08,
+      "loss": 0.6785,
+      "step": 8766
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.2050595456075517e-08,
+      "loss": 0.9187,
+      "step": 8767
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.1962872207821726e-08,
+      "loss": 0.812,
+      "step": 8768
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.187532251406671e-08,
+      "loss": 0.8495,
+      "step": 8769
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.1787946385090406e-08,
+      "loss": 0.8884,
+      "step": 8770
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.1700743831153112e-08,
+      "loss": 0.958,
+      "step": 8771
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.1613714862494305e-08,
+      "loss": 1.022,
+      "step": 8772
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.152685948933297e-08,
+      "loss": 0.8592,
+      "step": 8773
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.1440177721867625e-08,
+      "loss": 0.9802,
+      "step": 8774
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.135366957027729e-08,
+      "loss": 0.8547,
+      "step": 8775
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.1267335044719338e-08,
+      "loss": 0.8152,
+      "step": 8776
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.118117415533183e-08,
+      "loss": 0.6795,
+      "step": 8777
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.1095186912231668e-08,
+      "loss": 0.6975,
+      "step": 8778
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.1009373325515946e-08,
+      "loss": 0.95,
+      "step": 8779
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.092373340526077e-08,
+      "loss": 1.1441,
+      "step": 8780
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.083826716152226e-08,
+      "loss": 1.013,
+      "step": 8781
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.075297460433606e-08,
+      "loss": 0.8519,
+      "step": 8782
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0667855743717656e-08,
+      "loss": 1.1437,
+      "step": 8783
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.058291058966172e-08,
+      "loss": 0.8675,
+      "step": 8784
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0498139152142603e-08,
+      "loss": 0.4889,
+      "step": 8785
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0413541441114515e-08,
+      "loss": 1.0767,
+      "step": 8786
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0329117466511172e-08,
+      "loss": 0.8687,
+      "step": 8787
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.024486723824548e-08,
+      "loss": 0.9775,
+      "step": 8788
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.016079076621069e-08,
+      "loss": 1.1343,
+      "step": 8789
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0076888060279076e-08,
+      "loss": 0.9075,
+      "step": 8790
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.999315913030275e-08,
+      "loss": 0.9202,
+      "step": 8791
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.990960398611319e-08,
+      "loss": 0.7568,
+      "step": 8792
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.9826222637521875e-08,
+      "loss": 1.0949,
+      "step": 8793
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.9743015094319304e-08,
+      "loss": 1.1961,
+      "step": 8794
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.9659981366276336e-08,
+      "loss": 0.6223,
+      "step": 8795
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.9577121463142666e-08,
+      "loss": 1.0281,
+      "step": 8796
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9494435394647846e-08,
+      "loss": 0.9839,
+      "step": 8797
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9411923170501443e-08,
+      "loss": 0.8374,
+      "step": 8798
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9329584800391874e-08,
+      "loss": 0.8973,
+      "step": 8799
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9247420293987572e-08,
+      "loss": 0.9789,
+      "step": 8800
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9165429660936484e-08,
+      "loss": 0.7525,
+      "step": 8801
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9083612910866243e-08,
+      "loss": 0.9012,
+      "step": 8802
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9001970053383832e-08,
+      "loss": 1.1603,
+      "step": 8803
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8920501098076083e-08,
+      "loss": 1.0831,
+      "step": 8804
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8839206054509006e-08,
+      "loss": 0.9672,
+      "step": 8805
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8758084932228803e-08,
+      "loss": 0.9433,
+      "step": 8806
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.867713774076052e-08,
+      "loss": 0.963,
+      "step": 8807
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.859636448960972e-08,
+      "loss": 1.004,
+      "step": 8808
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8515765188260315e-08,
+      "loss": 0.8226,
+      "step": 8809
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8435339846176903e-08,
+      "loss": 0.8429,
+      "step": 8810
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8355088472802928e-08,
+      "loss": 0.9308,
+      "step": 8811
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.827501107756202e-08,
+      "loss": 0.8944,
+      "step": 8812
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.819510766985666e-08,
+      "loss": 0.7818,
+      "step": 8813
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.811537825906967e-08,
+      "loss": 1.0155,
+      "step": 8814
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.80358228545629e-08,
+      "loss": 0.9427,
+      "step": 8815
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7956441465677876e-08,
+      "loss": 1.0431,
+      "step": 8816
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7877234101735807e-08,
+      "loss": 0.9933,
+      "step": 8817
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7798200772037255e-08,
+      "loss": 0.6854,
+      "step": 8818
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7719341485862795e-08,
+      "loss": 0.8021,
+      "step": 8819
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7640656252471853e-08,
+      "loss": 1.2345,
+      "step": 8820
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7562145081104207e-08,
+      "loss": 0.8979,
+      "step": 8821
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7483807980978818e-08,
+      "loss": 0.8496,
+      "step": 8822
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.740564496129382e-08,
+      "loss": 0.8993,
+      "step": 8823
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7327656031227713e-08,
+      "loss": 0.7244,
+      "step": 8824
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7249841199937834e-08,
+      "loss": 1.0966,
+      "step": 8825
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.717220047656154e-08,
+      "loss": 0.7514,
+      "step": 8826
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7094733870215706e-08,
+      "loss": 1.0753,
+      "step": 8827
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7017441389996224e-08,
+      "loss": 1.1359,
+      "step": 8828
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6940323044979334e-08,
+      "loss": 1.0319,
+      "step": 8829
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.686337884422029e-08,
+      "loss": 0.9568,
+      "step": 8830
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.67866087967542e-08,
+      "loss": 0.884,
+      "step": 8831
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6710012911595353e-08,
+      "loss": 0.9724,
+      "step": 8832
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6633591197737887e-08,
+      "loss": 0.8347,
+      "step": 8833
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6557343664155454e-08,
+      "loss": 0.9214,
+      "step": 8834
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.64812703198014e-08,
+      "loss": 0.8635,
+      "step": 8835
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6405371173608074e-08,
+      "loss": 0.9917,
+      "step": 8836
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6329646234488016e-08,
+      "loss": 0.9617,
+      "step": 8837
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6254095511332778e-08,
+      "loss": 0.7923,
+      "step": 8838
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6178719013013932e-08,
+      "loss": 0.81,
+      "step": 8839
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.610351674838223e-08,
+      "loss": 1.1694,
+      "step": 8840
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6028488726267943e-08,
+      "loss": 0.8137,
+      "step": 8841
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5953634955481356e-08,
+      "loss": 0.8672,
+      "step": 8842
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.587895544481194e-08,
+      "loss": 1.0321,
+      "step": 8843
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5804450203028343e-08,
+      "loss": 0.5941,
+      "step": 8844
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.573011923887957e-08,
+      "loss": 0.8329,
+      "step": 8845
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.565596256109364e-08,
+      "loss": 1.0582,
+      "step": 8846
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5581980178378085e-08,
+      "loss": 0.9312,
+      "step": 8847
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.550817209941996e-08,
+      "loss": 0.9768,
+      "step": 8848
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5434538332886328e-08,
+      "loss": 1.1152,
+      "step": 8849
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5361078887423107e-08,
+      "loss": 0.8259,
+      "step": 8850
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.52877937716564e-08,
+      "loss": 1.0252,
+      "step": 8851
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.521468299419132e-08,
+      "loss": 0.8226,
+      "step": 8852
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.514174656361267e-08,
+      "loss": 0.9059,
+      "step": 8853
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5068984488484925e-08,
+      "loss": 0.9211,
+      "step": 8854
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4996396777351763e-08,
+      "loss": 1.0272,
+      "step": 8855
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4923983438736855e-08,
+      "loss": 0.9138,
+      "step": 8856
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4851744481142903e-08,
+      "loss": 1.1869,
+      "step": 8857
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.477967991305279e-08,
+      "loss": 0.6559,
+      "step": 8858
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4707789742927914e-08,
+      "loss": 0.7707,
+      "step": 8859
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4636073979210185e-08,
+      "loss": 0.9688,
+      "step": 8860
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4564532630320537e-08,
+      "loss": 0.7714,
+      "step": 8861
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4493165704659584e-08,
+      "loss": 1.0599,
+      "step": 8862
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.442197321060712e-08,
+      "loss": 0.7922,
+      "step": 8863
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4350955156523127e-08,
+      "loss": 1.0568,
+      "step": 8864
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4280111550746432e-08,
+      "loss": 0.7365,
+      "step": 8865
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4209442401595718e-08,
+      "loss": 1.0556,
+      "step": 8866
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4138947717369177e-08,
+      "loss": 1.1122,
+      "step": 8867
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4068627506344357e-08,
+      "loss": 0.6302,
+      "step": 8868
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3998481776778483e-08,
+      "loss": 0.8148,
+      "step": 8869
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3928510536908135e-08,
+      "loss": 1.0034,
+      "step": 8870
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.385871379494974e-08,
+      "loss": 1.0439,
+      "step": 8871
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3789091559098742e-08,
+      "loss": 1.0322,
+      "step": 8872
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3719643837530437e-08,
+      "loss": 0.9826,
+      "step": 8873
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3650370638399634e-08,
+      "loss": 0.9799,
+      "step": 8874
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3581271969840325e-08,
+      "loss": 1.267,
+      "step": 8875
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3512347839966355e-08,
+      "loss": 1.1773,
+      "step": 8876
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3443598256870914e-08,
+      "loss": 0.9312,
+      "step": 8877
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3375023228626716e-08,
+      "loss": 0.9223,
+      "step": 8878
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3306622763286146e-08,
+      "loss": 0.9347,
+      "step": 8879
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3238396868880952e-08,
+      "loss": 0.9225,
+      "step": 8880
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3170345553422225e-08,
+      "loss": 0.9075,
+      "step": 8881
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3102468824900738e-08,
+      "loss": 1.2204,
+      "step": 8882
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3034766691286614e-08,
+      "loss": 1.0545,
+      "step": 8883
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2967239160529831e-08,
+      "loss": 0.9546,
+      "step": 8884
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2899886240559544e-08,
+      "loss": 0.9969,
+      "step": 8885
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2832707939284426e-08,
+      "loss": 0.9648,
+      "step": 8886
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.276570426459267e-08,
+      "loss": 1.169,
+      "step": 8887
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2698875224352146e-08,
+      "loss": 0.3776,
+      "step": 8888
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.263222082641008e-08,
+      "loss": 0.4853,
+      "step": 8889
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.256574107859304e-08,
+      "loss": 4.5299,
+      "step": 8890
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2499435988707286e-08,
+      "loss": 0.9545,
+      "step": 8891
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2433305564538588e-08,
+      "loss": 1.2266,
+      "step": 8892
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2367349813851902e-08,
+      "loss": 1.0439,
+      "step": 8893
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2301568744392199e-08,
+      "loss": 0.9351,
+      "step": 8894
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2235962363883469e-08,
+      "loss": 1.0812,
+      "step": 8895
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.217053068002938e-08,
+      "loss": 1.0738,
+      "step": 8896
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.210527370051312e-08,
+      "loss": 1.1969,
+      "step": 8897
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2040191432997228e-08,
+      "loss": 1.2292,
+      "step": 8898
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1975283885124089e-08,
+      "loss": 0.7685,
+      "step": 8899
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1910551064514775e-08,
+      "loss": 1.0052,
+      "step": 8900
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.184599297877087e-08,
+      "loss": 1.2447,
+      "step": 8901
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1781609635472645e-08,
+      "loss": 1.2348,
+      "step": 8902
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.171740104218022e-08,
+      "loss": 0.887,
+      "step": 8903
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1653367206433063e-08,
+      "loss": 1.0428,
+      "step": 8904
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1589508135750326e-08,
+      "loss": 0.713,
+      "step": 8905
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1525823837630345e-08,
+      "loss": 0.9692,
+      "step": 8906
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1462314319551303e-08,
+      "loss": 0.8886,
+      "step": 8907
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1398979588970071e-08,
+      "loss": 1.2812,
+      "step": 8908
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1335819653324198e-08,
+      "loss": 0.7805,
+      "step": 8909
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1272834520029751e-08,
+      "loss": 0.9212,
+      "step": 8910
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.121002419648265e-08,
+      "loss": 0.9167,
+      "step": 8911
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1147388690058158e-08,
+      "loss": 0.9104,
+      "step": 8912
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1084928008111394e-08,
+      "loss": 0.9514,
+      "step": 8913
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1022642157975993e-08,
+      "loss": 1.0078,
+      "step": 8914
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0960531146966269e-08,
+      "loss": 0.7713,
+      "step": 8915
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0898594982375387e-08,
+      "loss": 0.8529,
+      "step": 8916
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0836833671475699e-08,
+      "loss": 0.6398,
+      "step": 8917
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0775247221519736e-08,
+      "loss": 0.72,
+      "step": 8918
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0713835639738878e-08,
+      "loss": 0.9086,
+      "step": 8919
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0652598933344194e-08,
+      "loss": 0.7805,
+      "step": 8920
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0591537109526429e-08,
+      "loss": 0.9761,
+      "step": 8921
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0530650175455514e-08,
+      "loss": 0.8119,
+      "step": 8922
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0469938138280899e-08,
+      "loss": 1.0591,
+      "step": 8923
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0409401005131547e-08,
+      "loss": 1.1173,
+      "step": 8924
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0349038783115771e-08,
+      "loss": 0.8418,
+      "step": 8925
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0288851479321736e-08,
+      "loss": 0.8035,
+      "step": 8926
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0228839100816456e-08,
+      "loss": 0.7651,
+      "step": 8927
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0169001654646958e-08,
+      "loss": 0.9067,
+      "step": 8928
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0109339147839126e-08,
+      "loss": 0.9783,
+      "step": 8929
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0049851587399016e-08,
+      "loss": 0.8019,
+      "step": 8930
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.99053898031188e-09,
+      "loss": 1.0432,
+      "step": 8931
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.931401333541979e-09,
+      "loss": 1.2199,
+      "step": 8932
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.872438654033588e-09,
+      "loss": 1.0222,
+      "step": 8933
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.81365094871034e-09,
+      "loss": 1.0594,
+      "step": 8934
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.755038224475043e-09,
+      "loss": 1.04,
+      "step": 8935
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.696600488210194e-09,
+      "loss": 0.9029,
+      "step": 8936
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.63833774677747e-09,
+      "loss": 0.9224,
+      "step": 8937
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.58025000701873e-09,
+      "loss": 0.7303,
+      "step": 8938
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.522337275754356e-09,
+      "loss": 1.0081,
+      "step": 8939
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.464599559784737e-09,
+      "loss": 0.8542,
+      "step": 8940
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.407036865889452e-09,
+      "loss": 0.5377,
+      "step": 8941
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.34964920082776e-09,
+      "loss": 0.7767,
+      "step": 8942
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.29243657133827e-09,
+      "loss": 1.131,
+      "step": 8943
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.235398984139109e-09,
+      "loss": 0.7489,
+      "step": 8944
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.178536445927422e-09,
+      "loss": 0.9749,
+      "step": 8945
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.121848963380364e-09,
+      "loss": 0.7677,
+      "step": 8946
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.065336543154279e-09,
+      "loss": 0.6981,
+      "step": 8947
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.008999191885025e-09,
+      "loss": 0.6463,
+      "step": 8948
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.952836916187812e-09,
+      "loss": 0.8977,
+      "step": 8949
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.89684972265753e-09,
+      "loss": 1.0727,
+      "step": 8950
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.84103761786792e-09,
+      "loss": 0.787,
+      "step": 8951
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.785400608372907e-09,
+      "loss": 0.7898,
+      "step": 8952
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.729938700705431e-09,
+      "loss": 0.6599,
+      "step": 8953
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.674651901377783e-09,
+      "loss": 0.9233,
+      "step": 8954
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.619540216882271e-09,
+      "loss": 1.1328,
+      "step": 8955
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.564603653689718e-09,
+      "loss": 0.7514,
+      "step": 8956
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.509842218251296e-09,
+      "loss": 0.7226,
+      "step": 8957
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.455255916997029e-09,
+      "loss": 0.934,
+      "step": 8958
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.40084475633679e-09,
+      "loss": 0.7436,
+      "step": 8959
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.3466087426593e-09,
+      "loss": 0.8596,
+      "step": 8960
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.292547882333301e-09,
+      "loss": 0.9502,
+      "step": 8961
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.238662181706714e-09,
+      "loss": 1.2499,
+      "step": 8962
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.184951647106976e-09,
+      "loss": 0.856,
+      "step": 8963
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.131416284840875e-09,
+      "loss": 1.0553,
+      "step": 8964
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.078056101194552e-09,
+      "loss": 0.7766,
+      "step": 8965
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.024871102433828e-09,
+      "loss": 1.0493,
+      "step": 8966
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.97186129480354e-09,
+      "loss": 0.6894,
+      "step": 8967
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.919026684528541e-09,
+      "loss": 0.9198,
+      "step": 8968
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.866367277812702e-09,
+      "loss": 0.8025,
+      "step": 8969
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.813883080839412e-09,
+      "loss": 1.0904,
+      "step": 8970
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.761574099771407e-09,
+      "loss": 0.7996,
+      "step": 8971
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.709440340750773e-09,
+      "loss": 0.8459,
+      "step": 8972
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.657481809899613e-09,
+      "loss": 0.9784,
+      "step": 8973
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.605698513318715e-09,
+      "loss": 0.8075,
+      "step": 8974
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.554090457088547e-09,
+      "loss": 0.999,
+      "step": 8975
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.502657647269262e-09,
+      "loss": 0.6953,
+      "step": 8976
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.451400089900029e-09,
+      "loss": 0.9022,
+      "step": 8977
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.4003177909997e-09,
+      "loss": 0.7323,
+      "step": 8978
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.349410756566643e-09,
+      "loss": 0.8715,
+      "step": 8979
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.298678992578078e-09,
+      "loss": 0.8624,
+      "step": 8980
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.2481225049914055e-09,
+      "loss": 0.9311,
+      "step": 8981
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.197741299742877e-09,
+      "loss": 1.1174,
+      "step": 8982
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.147535382748594e-09,
+      "loss": 0.9439,
+      "step": 8983
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.097504759903672e-09,
+      "loss": 0.9738,
+      "step": 8984
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.047649437082748e-09,
+      "loss": 0.6659,
+      "step": 8985
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.9979694201401375e-09,
+      "loss": 0.8816,
+      "step": 8986
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.948464714909175e-09,
+      "loss": 0.9139,
+      "step": 8987
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.89913532720321e-09,
+      "loss": 0.9546,
+      "step": 8988
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.8499812628139425e-09,
+      "loss": 1.104,
+      "step": 8989
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.801002527513922e-09,
+      "loss": 1.0249,
+      "step": 8990
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.752199127053715e-09,
+      "loss": 1.0575,
+      "step": 8991
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.7035710671644044e-09,
+      "loss": 0.9925,
+      "step": 8992
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.655118353555589e-09,
+      "loss": 0.9451,
+      "step": 8993
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.6068409919170514e-09,
+      "loss": 0.9925,
+      "step": 8994
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.558738987917423e-09,
+      "loss": 0.7219,
+      "step": 8995
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.510812347205186e-09,
+      "loss": 0.9237,
+      "step": 8996
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.463061075407506e-09,
+      "loss": 1.1237,
+      "step": 8997
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.415485178131897e-09,
+      "loss": 1.0656,
+      "step": 8998
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.368084660964723e-09,
+      "loss": 0.9521,
+      "step": 8999
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.320859529471868e-09,
+      "loss": 0.4908,
+      "step": 9000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.273809789198393e-09,
+      "loss": 0.6727,
+      "step": 9001
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.22693544566938e-09,
+      "loss": 1.111,
+      "step": 9002
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.1802365043885925e-09,
+      "loss": 1.2492,
+      "step": 9003
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.133712970839644e-09,
+      "loss": 0.817,
+      "step": 9004
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.0873648504854994e-09,
+      "loss": 1.3359,
+      "step": 9005
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.041192148768304e-09,
+      "loss": 1.1009,
+      "step": 9006
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.9951948711098884e-09,
+      "loss": 0.8066,
+      "step": 9007
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.94937302291143e-09,
+      "loss": 0.9947,
+      "step": 9008
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.903726609553295e-09,
+      "loss": 1.2051,
+      "step": 9009
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.8582556363953595e-09,
+      "loss": 1.3526,
+      "step": 9010
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.812960108777021e-09,
+      "loss": 1.1096,
+      "step": 9011
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.767840032016858e-09,
+      "loss": 0.993,
+      "step": 9012
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.722895411413298e-09,
+      "loss": 1.1181,
+      "step": 9013
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.678126252243454e-09,
+      "loss": 0.8259,
+      "step": 9014
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.633532559764454e-09,
+      "loss": 1.2145,
+      "step": 9015
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.589114339212442e-09,
+      "loss": 0.8869,
+      "step": 9016
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.5448715958030806e-09,
+      "loss": 1.02,
+      "step": 9017
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.500804334731713e-09,
+      "loss": 1.0225,
+      "step": 9018
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.4569125611727e-09,
+      "loss": 0.7262,
+      "step": 9019
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.413196280279753e-09,
+      "loss": 1.064,
+      "step": 9020
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.369655497186432e-09,
+      "loss": 0.9264,
+      "step": 9021
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.3262902170051475e-09,
+      "loss": 0.942,
+      "step": 9022
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.283100444827993e-09,
+      "loss": 0.8756,
+      "step": 9023
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.2400861857264115e-09,
+      "loss": 0.7576,
+      "step": 9024
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.197247444751363e-09,
+      "loss": 0.884,
+      "step": 9025
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.15458422693299e-09,
+      "loss": 0.9766,
+      "step": 9026
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.112096537280786e-09,
+      "loss": 0.6326,
+      "step": 9027
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.069784380783926e-09,
+      "loss": 1.5636,
+      "step": 9028
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.0276477624107696e-09,
+      "loss": 0.809,
+      "step": 9029
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.985686687109026e-09,
+      "loss": 0.9376,
+      "step": 9030
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.9439011598059195e-09,
+      "loss": 1.1086,
+      "step": 9031
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.902291185407859e-09,
+      "loss": 1.0838,
+      "step": 9032
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.860856768800936e-09,
+      "loss": 1.1406,
+      "step": 9033
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.819597914850427e-09,
+      "loss": 0.8469,
+      "step": 9034
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.778514628401121e-09,
+      "loss": 1.0223,
+      "step": 9035
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.737606914276993e-09,
+      "loss": 0.9793,
+      "step": 9036
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.696874777281701e-09,
+      "loss": 1.0476,
+      "step": 9037
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.65631822219792e-09,
+      "loss": 0.6893,
+      "step": 9038
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.615937253787838e-09,
+      "loss": 1.0328,
+      "step": 9039
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.575731876793332e-09,
+      "loss": 0.8901,
+      "step": 9040
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.535702095935124e-09,
+      "loss": 0.6393,
+      "step": 9041
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.495847915913787e-09,
+      "loss": 0.9004,
+      "step": 9042
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.456169341409077e-09,
+      "loss": 0.8908,
+      "step": 9043
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.416666377080103e-09,
+      "loss": 1.0887,
+      "step": 9044
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.377339027565486e-09,
+      "loss": 0.9454,
+      "step": 9045
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.338187297482865e-09,
+      "loss": 1.0016,
+      "step": 9046
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.299211191429897e-09,
+      "loss": 0.95,
+      "step": 9047
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.260410713983087e-09,
+      "loss": 1.0951,
+      "step": 9048
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.2217858696982914e-09,
+      "loss": 1.0118,
+      "step": 9049
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.183336663111215e-09,
+      "loss": 1.0446,
+      "step": 9050
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.1450630987364125e-09,
+      "loss": 1.0811,
+      "step": 9051
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.10696518106829e-09,
+      "loss": 1.1886,
+      "step": 9052
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.069042914580101e-09,
+      "loss": 1.0971,
+      "step": 9053
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.0312963037251185e-09,
+      "loss": 0.9008,
+      "step": 9054
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.993725352935296e-09,
+      "loss": 0.7735,
+      "step": 9055
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.956330066622438e-09,
+      "loss": 0.875,
+      "step": 9056
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.919110449177699e-09,
+      "loss": 1.1745,
+      "step": 9057
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.882066504971415e-09,
+      "loss": 0.816,
+      "step": 9058
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.845198238353276e-09,
+      "loss": 0.7205,
+      "step": 9059
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.808505653652483e-09,
+      "loss": 0.8078,
+      "step": 9060
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.771988755177591e-09,
+      "loss": 0.7334,
+      "step": 9061
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.735647547216503e-09,
+      "loss": 1.3086,
+      "step": 9062
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.6994820340366388e-09,
+      "loss": 0.6766,
+      "step": 9063
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.663492219884268e-09,
+      "loss": 0.7096,
+      "step": 9064
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.627678108985677e-09,
+      "loss": 0.9007,
+      "step": 9065
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.5920397055461685e-09,
+      "loss": 0.8283,
+      "step": 9066
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.5565770137505617e-09,
+      "loss": 0.9244,
+      "step": 9067
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.5212900377628587e-09,
+      "loss": 1.0741,
+      "step": 9068
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.4861787817264124e-09,
+      "loss": 6.1286,
+      "step": 9069
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.4512432497644243e-09,
+      "loss": 0.9546,
+      "step": 9070
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.4164834459787798e-09,
+      "loss": 0.8129,
+      "step": 9071
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.381899374451214e-09,
+      "loss": 0.7918,
+      "step": 9072
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.347491039242645e-09,
+      "loss": 0.8945,
+      "step": 9073
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.313258444393341e-09,
+      "loss": 0.8557,
+      "step": 9074
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.2792015939230867e-09,
+      "loss": 0.8453,
+      "step": 9075
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.2453204918306835e-09,
+      "loss": 1.3617,
+      "step": 9076
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.2116151420947815e-09,
+      "loss": 1.1189,
+      "step": 9077
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.178085548673049e-09,
+      "loss": 0.8606,
+      "step": 9078
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.1447317155025023e-09,
+      "loss": 0.903,
+      "step": 9079
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.1115536464996765e-09,
+      "loss": 1.1208,
+      "step": 9080
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.078551345560454e-09,
+      "loss": 0.8611,
+      "step": 9081
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0457248165600694e-09,
+      "loss": 1.0771,
+      "step": 9082
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.013074063353105e-09,
+      "loss": 1.055,
+      "step": 9083
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9805990897733282e-09,
+      "loss": 0.801,
+      "step": 9084
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9482998996341883e-09,
+      "loss": 0.9297,
+      "step": 9085
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9161764967283177e-09,
+      "loss": 0.9027,
+      "step": 9086
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8842288848277e-09,
+      "loss": 1.0562,
+      "step": 9087
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8524570676835004e-09,
+      "loss": 1.1456,
+      "step": 9088
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.820861049026735e-09,
+      "loss": 0.9301,
+      "step": 9089
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.789440832567436e-09,
+      "loss": 0.8406,
+      "step": 9090
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.7581964219948185e-09,
+      "loss": 0.6557,
+      "step": 9091
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.7271278209777817e-09,
+      "loss": 0.9996,
+      "step": 9092
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6962350331645736e-09,
+      "loss": 0.7729,
+      "step": 9093
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.665518062182626e-09,
+      "loss": 1.0555,
+      "step": 9094
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6349769116388868e-09,
+      "loss": 0.9134,
+      "step": 9095
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.604611585119321e-09,
+      "loss": 1.3159,
+      "step": 9096
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5744220861897425e-09,
+      "loss": 0.9785,
+      "step": 9097
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5444084183949833e-09,
+      "loss": 1.1961,
+      "step": 9098
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.514570585259224e-09,
+      "loss": 0.8305,
+      "step": 9099
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.484908590286161e-09,
+      "loss": 0.7338,
+      "step": 9100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.455422436958843e-09,
+      "loss": 0.9865,
+      "step": 9101
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.4261121287394994e-09,
+      "loss": 0.9321,
+      "step": 9102
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.396977669070044e-09,
+      "loss": 1.1184,
+      "step": 9103
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.3680190613712403e-09,
+      "loss": 1.1531,
+      "step": 9104
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.3392363090435354e-09,
+      "loss": 1.0413,
+      "step": 9105
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.3106294154665587e-09,
+      "loss": 0.9931,
+      "step": 9106
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.2821983839997897e-09,
+      "loss": 0.9666,
+      "step": 9107
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.253943217981225e-09,
+      "loss": 1.0733,
+      "step": 9108
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.2258639207288765e-09,
+      "loss": 0.8277,
+      "step": 9109
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.1979604955399413e-09,
+      "loss": 1.1861,
+      "step": 9110
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.170232945690631e-09,
+      "loss": 0.7767,
+      "step": 9111
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.1426812744370083e-09,
+      "loss": 1.0764,
+      "step": 9112
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.1153054850143183e-09,
+      "loss": 0.8454,
+      "step": 9113
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.088105580636823e-09,
+      "loss": 0.9391,
+      "step": 9114
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.0610815644984683e-09,
+      "loss": 0.8179,
+      "step": 9115
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.0342334397727147e-09,
+      "loss": 0.9743,
+      "step": 9116
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.0075612096118743e-09,
+      "loss": 0.9116,
+      "step": 9117
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.9810648771479425e-09,
+      "loss": 0.7762,
+      "step": 9118
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.9547444454922627e-09,
+      "loss": 1.1673,
+      "step": 9119
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.928599917735363e-09,
+      "loss": 0.8397,
+      "step": 9120
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.9026312969471217e-09,
+      "loss": 1.0684,
+      "step": 9121
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.876838586176932e-09,
+      "loss": 0.7326,
+      "step": 9122
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8512217884532056e-09,
+      "loss": 0.9759,
+      "step": 9123
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8257809067843689e-09,
+      "loss": 0.7732,
+      "step": 9124
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8005159441575326e-09,
+      "loss": 0.7851,
+      "step": 9125
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7754269035391568e-09,
+      "loss": 0.9031,
+      "step": 9126
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7505137878755517e-09,
+      "loss": 0.8264,
+      "step": 9127
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7257766000918774e-09,
+      "loss": 1.268,
+      "step": 9128
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7012153430929768e-09,
+      "loss": 0.8576,
+      "step": 9129
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.676830019762876e-09,
+      "loss": 0.7343,
+      "step": 9130
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6526206329647852e-09,
+      "loss": 0.8701,
+      "step": 9131
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6285871855415967e-09,
+      "loss": 0.9375,
+      "step": 9132
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.60472968031522e-09,
+      "loss": 0.83,
+      "step": 9133
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5810481200872472e-09,
+      "loss": 0.7977,
+      "step": 9134
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5575425076382877e-09,
+      "loss": 1.0081,
+      "step": 9135
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.534212845728633e-09,
+      "loss": 1.0876,
+      "step": 9136
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5110591370974258e-09,
+      "loss": 1.0214,
+      "step": 9137
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.488081384463491e-09,
+      "loss": 1.1699,
+      "step": 9138
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4652795905250038e-09,
+      "loss": 0.7738,
+      "step": 9139
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4426537579593224e-09,
+      "loss": 0.7698,
+      "step": 9140
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4202038894233216e-09,
+      "loss": 0.7563,
+      "step": 9141
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3979299875530592e-09,
+      "loss": 0.8597,
+      "step": 9142
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3758320549641102e-09,
+      "loss": 1.1325,
+      "step": 9143
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3539100942510651e-09,
+      "loss": 0.9449,
+      "step": 9144
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3321641079881985e-09,
+      "loss": 0.8755,
+      "step": 9145
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3105940987288012e-09,
+      "loss": 0.8144,
+      "step": 9146
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.289200069005847e-09,
+      "loss": 1.0284,
+      "step": 9147
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.267982021331493e-09,
+      "loss": 0.8795,
+      "step": 9148
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2469399581970797e-09,
+      "loss": 1.1605,
+      "step": 9149
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.226073882073464e-09,
+      "loss": 0.7756,
+      "step": 9150
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2053837954106862e-09,
+      "loss": 0.684,
+      "step": 9151
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.184869700638469e-09,
+      "loss": 0.895,
+      "step": 9152
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.164531600165386e-09,
+      "loss": 0.9234,
+      "step": 9153
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1443694963796936e-09,
+      "loss": 1.0467,
+      "step": 9154
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1243833916489976e-09,
+      "loss": 0.8252,
+      "step": 9155
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1045732883197547e-09,
+      "loss": 0.7516,
+      "step": 9156
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0849391887184368e-09,
+      "loss": 0.944,
+      "step": 9157
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0654810951505334e-09,
+      "loss": 0.8562,
+      "step": 9158
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0461990099005502e-09,
+      "loss": 0.8207,
+      "step": 9159
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0270929352328428e-09,
+      "loss": 1.0063,
+      "step": 9160
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0081628733909497e-09,
+      "loss": 1.0404,
+      "step": 9161
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.89408826597593e-10,
+      "loss": 0.9755,
+      "step": 9162
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.708307970550113e-10,
+      "loss": 1.1729,
+      "step": 9163
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.524287869444593e-10,
+      "loss": 0.8213,
+      "step": 9164
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.342027984270418e-10,
+      "loss": 1.136,
+      "step": 9165
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.161528336425473e-10,
+      "loss": 0.9168,
+      "step": 9166
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.982788947107801e-10,
+      "loss": 1.0418,
+      "step": 9167
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.805809837303946e-10,
+      "loss": 0.8413,
+      "step": 9168
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.630591027793955e-10,
+      "loss": 0.7068,
+      "step": 9169
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.457132539154699e-10,
+      "loss": 1.1693,
+      "step": 9170
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.285434391751556e-10,
+      "loss": 0.8465,
+      "step": 9171
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.115496605748395e-10,
+      "loss": 0.5939,
+      "step": 9172
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.947319201099257e-10,
+      "loss": 1.2555,
+      "step": 9173
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.780902197550011e-10,
+      "loss": 1.0588,
+      "step": 9174
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.616245614641693e-10,
+      "loss": 0.9921,
+      "step": 9175
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.453349471712167e-10,
+      "loss": 1.42,
+      "step": 9176
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.292213787884472e-10,
+      "loss": 1.0034,
+      "step": 9177
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.132838582083467e-10,
+      "loss": 0.8677,
+      "step": 9178
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.975223873020853e-10,
+      "loss": 0.8263,
+      "step": 9179
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.819369679205157e-10,
+      "loss": 0.8512,
+      "step": 9180
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.665276018936739e-10,
+      "loss": 1.0465,
+      "step": 9181
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.51294291030946e-10,
+      "loss": 0.7765,
+      "step": 9182
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.362370371212345e-10,
+      "loss": 0.8056,
+      "step": 9183
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.213558419322917e-10,
+      "loss": 0.998,
+      "step": 9184
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.0665070721172e-10,
+      "loss": 0.6157,
+      "step": 9185
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.921216346863045e-10,
+      "loss": 0.7699,
+      "step": 9186
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.77768626061681e-10,
+      "loss": 0.6766,
+      "step": 9187
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.635916830236676e-10,
+      "loss": 1.1989,
+      "step": 9188
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.495908072367661e-10,
+      "loss": 0.622,
+      "step": 9189
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.357660003449949e-10,
+      "loss": 0.8056,
+      "step": 9190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.221172639715555e-10,
+      "loss": 0.7291,
+      "step": 9191
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.086445997194988e-10,
+      "loss": 1.0384,
+      "step": 9192
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.953480091703932e-10,
+      "loss": 1.1424,
+      "step": 9193
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.822274938858229e-10,
+      "loss": 0.8935,
+      "step": 9194
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.692830554062222e-10,
+      "loss": 0.9053,
+      "step": 9195
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.565146952517085e-10,
+      "loss": 0.8371,
+      "step": 9196
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.4392241492158256e-10,
+      "loss": 1.0369,
+      "step": 9197
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.315062158944949e-10,
+      "loss": 0.7209,
+      "step": 9198
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.192660996281128e-10,
+      "loss": 0.667,
+      "step": 9199
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.07202067559953e-10,
+      "loss": 1.0014,
+      "step": 9200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.953141211065492e-10,
+      "loss": 0.8355,
+      "step": 9201
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.836022616639512e-10,
+      "loss": 1.1766,
+      "step": 9202
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.7206649060705923e-10,
+      "loss": 1.0663,
+      "step": 9203
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.6070680929062293e-10,
+      "loss": 0.7395,
+      "step": 9204
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4952321904857533e-10,
+      "loss": 0.6171,
+      "step": 9205
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.385157211938661e-10,
+      "loss": 0.9571,
+      "step": 9206
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.27684317019461e-10,
+      "loss": 1.0511,
+      "step": 9207
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.17029007796843e-10,
+      "loss": 1.1745,
+      "step": 9208
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.065497947771778e-10,
+      "loss": 0.898,
+      "step": 9209
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.962466791913143e-10,
+      "loss": 0.8422,
+      "step": 9210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.861196622486184e-10,
+      "loss": 0.8937,
+      "step": 9211
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7616874513847203e-10,
+      "loss": 0.8663,
+      "step": 9212
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.663939290294404e-10,
+      "loss": 0.8079,
+      "step": 9213
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.5679521506910553e-10,
+      "loss": 0.61,
+      "step": 9214
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.4737260438456586e-10,
+      "loss": 0.908,
+      "step": 9215
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.381260980822697e-10,
+      "loss": 1.0021,
+      "step": 9216
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.2905569724818166e-10,
+      "loss": 0.7124,
+      "step": 9217
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.2016140294711662e-10,
+      "loss": 0.915,
+      "step": 9218
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.114432162234059e-10,
+      "loss": 0.7569,
+      "step": 9219
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.029011381010637e-10,
+      "loss": 1.1805,
+      "step": 9220
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.9453516958278793e-10,
+      "loss": 1.1938,
+      "step": 9221
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.86345311651126e-10,
+      "loss": 1.1043,
+      "step": 9222
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7833156526780858e-10,
+      "loss": 0.7435,
+      "step": 9223
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.704939313737497e-10,
+      "loss": 0.9414,
+      "step": 9224
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.6283241088904667e-10,
+      "loss": 1.3407,
+      "step": 9225
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.5534700471364626e-10,
+      "loss": 0.926,
+      "step": 9226
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.480377137265121e-10,
+      "loss": 0.8795,
+      "step": 9227
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4090453878562448e-10,
+      "loss": 0.9758,
+      "step": 9228
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3394748072881323e-10,
+      "loss": 0.8707,
+      "step": 9229
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2716654037292497e-10,
+      "loss": 1.0753,
+      "step": 9230
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.205617185141561e-10,
+      "loss": 0.9091,
+      "step": 9231
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1413301592805292e-10,
+      "loss": 0.6132,
+      "step": 9232
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0788043336951159e-10,
+      "loss": 1.0086,
+      "step": 9233
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0180397157294463e-10,
+      "loss": 1.0535,
+      "step": 9234
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.59036312514483e-11,
+      "loss": 1.026,
+      "step": 9235
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.017941309813482e-11,
+      "loss": 0.9926,
+      "step": 9236
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.463131778513323e-11,
+      "loss": 0.9138,
+      "step": 9237
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.925934596392237e-11,
+      "loss": 0.6608,
+      "step": 9238
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.406349826499792e-11,
+      "loss": 1.1649,
+      "step": 9239
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.904377529887151e-11,
+      "loss": 0.7007,
+      "step": 9240
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.420017765490505e-11,
+      "loss": 0.5466,
+      "step": 9241
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.953270590164373e-11,
+      "loss": 0.8929,
+      "step": 9242
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.504136058714915e-11,
+      "loss": 0.903,
+      "step": 9243
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.072614223916583e-11,
+      "loss": 0.9701,
+      "step": 9244
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.658705136395547e-11,
+      "loss": 0.9035,
+      "step": 9245
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.262408844779575e-11,
+      "loss": 0.947,
+      "step": 9246
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.883725395581461e-11,
+      "loss": 1.1015,
+      "step": 9247
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.522654833298944e-11,
+      "loss": 0.8617,
+      "step": 9248
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.179197200298134e-11,
+      "loss": 1.0385,
+      "step": 9249
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.8533525369300872e-11,
+      "loss": 0.8315,
+      "step": 9250
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.5451208814641914e-11,
+      "loss": 1.0983,
+      "step": 9251
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.254502270054859e-11,
+      "loss": 0.8485,
+      "step": 9252
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.9814967368581017e-11,
+      "loss": 1.1176,
+      "step": 9253
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.726104313931609e-11,
+      "loss": 0.7517,
+      "step": 9254
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.488325031251403e-11,
+      "loss": 1.3063,
+      "step": 9255
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2681589167284902e-11,
+      "loss": 0.7197,
+      "step": 9256
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0656059962421694e-11,
+      "loss": 0.9057,
+      "step": 9257
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.806662935734177e-12,
+      "loss": 0.8768,
+      "step": 9258
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.13339830404891e-12,
+      "loss": 0.6736,
+      "step": 9259
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.636266264374968e-12,
+      "loss": 0.5764,
+      "step": 9260
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3152669920720756e-12,
+      "loss": 0.7504,
+      "step": 9261
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.1704006423494137e-12,
+      "loss": 0.9519,
+      "step": 9262
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2016673497660124e-12,
+      "loss": 0.6099,
+      "step": 9263
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4090672278976868e-12,
+      "loss": 0.7569,
+      "step": 9264
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.926003700031714e-13,
+      "loss": 1.2467,
+      "step": 9265
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.5226684835798496e-13,
+      "loss": 1.0045,
+      "step": 9266
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.806671458749805e-14,
+      "loss": 0.8938,
+      "step": 9267
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0,
+      "loss": 1.0135,
+      "step": 9268
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 9268,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 2317,
+  "total_flos": 3.120536549138104e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}