diff --git "a/checkpoint-2317/trainer_state.json" "b/checkpoint-2317/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2317/trainer_state.json"
@@ -0,0 +1,13955 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 580,
+  "global_step": 2317,
+  "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
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 9268,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 2317,
+  "total_flos": 7.80134137284526e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}